1
0

Compare commits

..

3 Commits

Author SHA1 Message Date
Erik Johnston
6c11bb956c Fix pypy repr for DomainSpecificString 2016-04-11 10:58:24 +01:00
Erik Johnston
58de897c77 Merge branch 'develop' into HEAD 2016-04-11 10:40:49 +01:00
TimePath
06602461e7 Support PyPy 2016-04-10 18:30:56 +10:00
354 changed files with 8866 additions and 28402 deletions

8
.gitignore vendored
View File

@@ -24,10 +24,10 @@ homeserver*.yaml
.coverage
htmlcov
demo/*/*.db
demo/*/*.log
demo/*/*.log.*
demo/*/*.pid
demo/*.db
demo/*.log
demo/*.log.*
demo/*.pid
demo/media_store.*
demo/etc

View File

@@ -1,17 +0,0 @@
sudo: false
language: python
python: 2.7
# tell travis to cache ~/.cache/pip
cache: pip
env:
- TOX_ENV=packaging
- TOX_ENV=pep8
- TOX_ENV=py27
install:
- pip install tox
script:
- tox -e $TOX_ENV

View File

@@ -1,670 +1,3 @@
Changes in synapse v0.18.6-rc2 (2016-12-30)
===========================================
Bug fixes:
* Fix memory leak in twisted by initialising logging correctly (PR #1731)
* Fix bug where fetching missing events took an unacceptable amount of time in
large rooms (PR #1734)
Changes in synapse v0.18.6-rc1 (2016-12-29)
===========================================
Bug fixes:
* Make sure that outbound connections are closed (PR #1725)
Changes in synapse v0.18.5 (2016-12-16)
=======================================
Bug fixes:
* Fix federation /backfill returning events it shouldn't (PR #1700)
* Fix crash in url preview (PR #1701)
Changes in synapse v0.18.5-rc3 (2016-12-13)
===========================================
Features:
* Add support for E2E for guests (PR #1653)
* Add new API appservice specific public room list (PR #1676)
* Add new room membership APIs (PR #1680)
Changes:
* Enable guest access for private rooms by default (PR #653)
* Limit the number of events that can be created on a given room concurrently
(PR #1620)
* Log the args that we have on UI auth completion (PR #1649)
* Stop generating refresh_tokens (PR #1654)
* Stop putting a time caveat on access tokens (PR #1656)
* Remove unspecced GET endpoints for e2e keys (PR #1694)
Bug fixes:
* Fix handling of 500 and 429's over federation (PR #1650)
* Fix Content-Type header parsing (PR #1660)
* Fix error when previewing sites that include unicode, thanks to kyrias (PR
#1664)
* Fix some cases where we drop read receipts (PR #1678)
* Fix bug where calls to ``/sync`` didn't correctly timeout (PR #1683)
* Fix bug where E2E key query would fail if a single remote host failed (PR
#1686)
Changes in synapse v0.18.5-rc2 (2016-11-24)
===========================================
Bug fixes:
* Don't send old events over federation, fixes bug in -rc1.
Changes in synapse v0.18.5-rc1 (2016-11-24)
===========================================
Features:
* Implement "event_fields" in filters (PR #1638)
Changes:
* Use external ldap auth pacakge (PR #1628)
* Split out federation transaction sending to a worker (PR #1635)
* Fail with a coherent error message if `/sync?filter=` is invalid (PR #1636)
* More efficient notif count queries (PR #1644)
Changes in synapse v0.18.4 (2016-11-22)
=======================================
Bug fixes:
* Add workaround for buggy clients that the fail to register (PR #1632)
Changes in synapse v0.18.4-rc1 (2016-11-14)
===========================================
Changes:
* Various database efficiency improvements (PR #1188, #1192)
* Update default config to blacklist more internal IPs, thanks to Euan Kemp (PR
#1198)
* Allow specifying duration in minutes in config, thanks to Daniel Dent (PR
#1625)
Bug fixes:
* Fix media repo to set CORs headers on responses (PR #1190)
* Fix registration to not error on non-ascii passwords (PR #1191)
* Fix create event code to limit the number of prev_events (PR #1615)
* Fix bug in transaction ID deduplication (PR #1624)
Changes in synapse v0.18.3 (2016-11-08)
=======================================
SECURITY UPDATE
Explicitly require authentication when using LDAP3. This is the default on
versions of ``ldap3`` above 1.0, but some distributions will package an older
version.
If you are using LDAP3 login and have a version of ``ldap3`` older than 1.0 it
is **CRITICAL to updgrade**.
Changes in synapse v0.18.2 (2016-11-01)
=======================================
No changes since v0.18.2-rc5
Changes in synapse v0.18.2-rc5 (2016-10-28)
===========================================
Bug fixes:
* Fix prometheus process metrics in worker processes (PR #1184)
Changes in synapse v0.18.2-rc4 (2016-10-27)
===========================================
Bug fixes:
* Fix ``user_threepids`` schema delta, which in some instances prevented
startup after upgrade (PR #1183)
Changes in synapse v0.18.2-rc3 (2016-10-27)
===========================================
Changes:
* Allow clients to supply access tokens as headers (PR #1098)
* Clarify error codes for GET /filter/, thanks to Alexander Maznev (PR #1164)
* Make password reset email field case insensitive (PR #1170)
* Reduce redundant database work in email pusher (PR #1174)
* Allow configurable rate limiting per AS (PR #1175)
* Check whether to ratelimit sooner to avoid work (PR #1176)
* Standardise prometheus metrics (PR #1177)
Bug fixes:
* Fix incredibly slow back pagination query (PR #1178)
* Fix infinite typing bug (PR #1179)
Changes in synapse v0.18.2-rc2 (2016-10-25)
===========================================
(This release did not include the changes advertised and was identical to RC1)
Changes in synapse v0.18.2-rc1 (2016-10-17)
===========================================
Changes:
* Remove redundant event_auth index (PR #1113)
* Reduce DB hits for replication (PR #1141)
* Implement pluggable password auth (PR #1155)
* Remove rate limiting from app service senders and fix get_or_create_user
requester, thanks to Patrik Oldsberg (PR #1157)
* window.postmessage for Interactive Auth fallback (PR #1159)
* Use sys.executable instead of hardcoded python, thanks to Pedro Larroy
(PR #1162)
* Add config option for adding additional TLS fingerprints (PR #1167)
* User-interactive auth on delete device (PR #1168)
Bug fixes:
* Fix not being allowed to set your own state_key, thanks to Patrik Oldsberg
(PR #1150)
* Fix interactive auth to return 401 from for incorrect password (PR #1160,
#1166)
* Fix email push notifs being dropped (PR #1169)
Changes in synapse v0.18.1 (2016-10-05)
======================================
No changes since v0.18.1-rc1
Changes in synapse v0.18.1-rc1 (2016-09-30)
===========================================
Features:
* Add total_room_count_estimate to ``/publicRooms`` (PR #1133)
Changes:
* Time out typing over federation (PR #1140)
* Restructure LDAP authentication (PR #1153)
Bug fixes:
* Fix 3pid invites when server is already in the room (PR #1136)
* Fix upgrading with SQLite taking lots of CPU for a few days
after upgrade (PR #1144)
* Fix upgrading from very old database versions (PR #1145)
* Fix port script to work with recently added tables (PR #1146)
Changes in synapse v0.18.0 (2016-09-19)
=======================================
The release includes major changes to the state storage database schemas, which
significantly reduce database size. Synapse will attempt to upgrade the current
data in the background. Servers with large SQLite database may experience
degradation of performance while this upgrade is in progress, therefore you may
want to consider migrating to using Postgres before upgrading very large SQLite
databases
Changes:
* Make public room search case insensitive (PR #1127)
Bug fixes:
* Fix and clean up publicRooms pagination (PR #1129)
Changes in synapse v0.18.0-rc1 (2016-09-16)
===========================================
Features:
* Add ``only=highlight`` on ``/notifications`` (PR #1081)
* Add server param to /publicRooms (PR #1082)
* Allow clients to ask for the whole of a single state event (PR #1094)
* Add is_direct param to /createRoom (PR #1108)
* Add pagination support to publicRooms (PR #1121)
* Add very basic filter API to /publicRooms (PR #1126)
* Add basic direct to device messaging support for E2E (PR #1074, #1084, #1104,
#1111)
Changes:
* Move to storing state_groups_state as deltas, greatly reducing DB size (PR
#1065)
* Reduce amount of state pulled out of the DB during common requests (PR #1069)
* Allow PDF to be rendered from media repo (PR #1071)
* Reindex state_groups_state after pruning (PR #1085)
* Clobber EDUs in send queue (PR #1095)
* Conform better to the CAS protocol specification (PR #1100)
* Limit how often we ask for keys from dead servers (PR #1114)
Bug fixes:
* Fix /notifications API when used with ``from`` param (PR #1080)
* Fix backfill when cannot find an event. (PR #1107)
Changes in synapse v0.17.3 (2016-09-09)
=======================================
This release fixes a major bug that stopped servers from handling rooms with
over 1000 members.
Changes in synapse v0.17.2 (2016-09-08)
=======================================
This release contains security bug fixes. Please upgrade.
No changes since v0.17.2-rc1
Changes in synapse v0.17.2-rc1 (2016-09-05)
===========================================
Features:
* Start adding store-and-forward direct-to-device messaging (PR #1046, #1050,
#1062, #1066)
Changes:
* Avoid pulling the full state of a room out so often (PR #1047, #1049, #1063,
#1068)
* Don't notify for online to online presence transitions. (PR #1054)
* Occasionally persist unpersisted presence updates (PR #1055)
* Allow application services to have an optional 'url' (PR #1056)
* Clean up old sent transactions from DB (PR #1059)
Bug fixes:
* Fix None check in backfill (PR #1043)
* Fix membership changes to be idempotent (PR #1067)
* Fix bug in get_pdu where it would sometimes return events with incorrect
signature
Changes in synapse v0.17.1 (2016-08-24)
=======================================
Changes:
* Delete old received_transactions rows (PR #1038)
* Pass through user-supplied content in /join/$room_id (PR #1039)
Bug fixes:
* Fix bug with backfill (PR #1040)
Changes in synapse v0.17.1-rc1 (2016-08-22)
===========================================
Features:
* Add notification API (PR #1028)
Changes:
* Don't print stack traces when failing to get remote keys (PR #996)
* Various federation /event/ perf improvements (PR #998)
* Only process one local membership event per room at a time (PR #1005)
* Move default display name push rule (PR #1011, #1023)
* Fix up preview URL API. Add tests. (PR #1015)
* Set ``Content-Security-Policy`` on media repo (PR #1021)
* Make notify_interested_services faster (PR #1022)
* Add usage stats to prometheus monitoring (PR #1037)
Bug fixes:
* Fix token login (PR #993)
* Fix CAS login (PR #994, #995)
* Fix /sync to not clobber status_msg (PR #997)
* Fix redacted state events to include prev_content (PR #1003)
* Fix some bugs in the auth/ldap handler (PR #1007)
* Fix backfill request to limit URI length, so that remotes don't reject the
requests due to path length limits (PR #1012)
* Fix AS push code to not send duplicate events (PR #1025)
Changes in synapse v0.17.0 (2016-08-08)
=======================================
This release contains significant security bug fixes regarding authenticating
events received over federation. PLEASE UPGRADE.
This release changes the LDAP configuration format in a backwards incompatible
way, see PR #843 for details.
Changes:
* Add federation /version API (PR #990)
* Make psutil dependency optional (PR #992)
Bug fixes:
* Fix URL preview API to exclude HTML comments in description (PR #988)
* Fix error handling of remote joins (PR #991)
Changes in synapse v0.17.0-rc4 (2016-08-05)
===========================================
Changes:
* Change the way we summarize URLs when previewing (PR #973)
* Add new ``/state_ids/`` federation API (PR #979)
* Speed up processing of ``/state/`` response (PR #986)
Bug fixes:
* Fix event persistence when event has already been partially persisted
(PR #975, #983, #985)
* Fix port script to also copy across backfilled events (PR #982)
Changes in synapse v0.17.0-rc3 (2016-08-02)
===========================================
Changes:
* Forbid non-ASes from registering users whose names begin with '_' (PR #958)
* Add some basic admin API docs (PR #963)
Bug fixes:
* Send the correct host header when fetching keys (PR #941)
* Fix joining a room that has missing auth events (PR #964)
* Fix various push bugs (PR #966, #970)
* Fix adding emails on registration (PR #968)
Changes in synapse v0.17.0-rc2 (2016-08-02)
===========================================
(This release did not include the changes advertised and was identical to RC1)
Changes in synapse v0.17.0-rc1 (2016-07-28)
===========================================
This release changes the LDAP configuration format in a backwards incompatible
way, see PR #843 for details.
Features:
* Add purge_media_cache admin API (PR #902)
* Add deactivate account admin API (PR #903)
* Add optional pepper to password hashing (PR #907, #910 by KentShikama)
* Add an admin option to shared secret registration (breaks backwards compat)
(PR #909)
* Add purge local room history API (PR #911, #923, #924)
* Add requestToken endpoints (PR #915)
* Add an /account/deactivate endpoint (PR #921)
* Add filter param to /messages. Add 'contains_url' to filter. (PR #922)
* Add device_id support to /login (PR #929)
* Add device_id support to /v2/register flow. (PR #937, #942)
* Add GET /devices endpoint (PR #939, #944)
* Add GET /device/{deviceId} (PR #943)
* Add update and delete APIs for devices (PR #949)
Changes:
* Rewrite LDAP Authentication against ldap3 (PR #843 by mweinelt)
* Linearize some federation endpoints based on (origin, room_id) (PR #879)
* Remove the legacy v0 content upload API. (PR #888)
* Use similar naming we use in email notifs for push (PR #894)
* Optionally include password hash in createUser endpoint (PR #905 by
KentShikama)
* Use a query that postgresql optimises better for get_events_around (PR #906)
* Fall back to 'username' if 'user' is not given for appservice registration.
(PR #927 by Half-Shot)
* Add metrics for psutil derived memory usage (PR #936)
* Record device_id in client_ips (PR #938)
* Send the correct host header when fetching keys (PR #941)
* Log the hostname the reCAPTCHA was completed on (PR #946)
* Make the device id on e2e key upload optional (PR #956)
* Add r0.2.0 to the "supported versions" list (PR #960)
* Don't include name of room for invites in push (PR #961)
Bug fixes:
* Fix substitution failure in mail template (PR #887)
* Put most recent 20 messages in email notif (PR #892)
* Ensure that the guest user is in the database when upgrading accounts
(PR #914)
* Fix various edge cases in auth handling (PR #919)
* Fix 500 ISE when sending alias event without a state_key (PR #925)
* Fix bug where we stored rejections in the state_group, persist all
rejections (PR #948)
* Fix lack of check of if the user is banned when handling 3pid invites
(PR #952)
* Fix a couple of bugs in the transaction and keyring code (PR #954, #955)
Changes in synapse v0.16.1-r1 (2016-07-08)
==========================================
THIS IS A CRITICAL SECURITY UPDATE.
This fixes a bug which allowed users' accounts to be accessed by unauthorised
users.
Changes in synapse v0.16.1 (2016-06-20)
=======================================
Bug fixes:
* Fix assorted bugs in ``/preview_url`` (PR #872)
* Fix TypeError when setting unicode passwords (PR #873)
Performance improvements:
* Turn ``use_frozen_events`` off by default (PR #877)
* Disable responding with canonical json for federation (PR #878)
Changes in synapse v0.16.1-rc1 (2016-06-15)
===========================================
Features: None
Changes:
* Log requester for ``/publicRoom`` endpoints when possible (PR #856)
* 502 on ``/thumbnail`` when can't connect to remote server (PR #862)
* Linearize fetching of gaps on incoming events (PR #871)
Bugs fixes:
* Fix bug where rooms where marked as published by default (PR #857)
* Fix bug where joining room with an event with invalid sender (PR #868)
* Fix bug where backfilled events were sent down sync streams (PR #869)
* Fix bug where outgoing connections could wedge indefinitely, causing push
notifications to be unreliable (PR #870)
Performance improvements:
* Improve ``/publicRooms`` performance(PR #859)
Changes in synapse v0.16.0 (2016-06-09)
=======================================
NB: As of v0.14 all AS config files must have an ID field.
Bug fixes:
* Don't make rooms published by default (PR #857)
Changes in synapse v0.16.0-rc2 (2016-06-08)
===========================================
Features:
* Add configuration option for tuning GC via ``gc.set_threshold`` (PR #849)
Changes:
* Record metrics about GC (PR #771, #847, #852)
* Add metric counter for number of persisted events (PR #841)
Bug fixes:
* Fix 'From' header in email notifications (PR #843)
* Fix presence where timeouts were not being fired for the first 8h after
restarts (PR #842)
* Fix bug where synapse sent malformed transactions to AS's when retrying
transactions (Commits 310197b, 8437906)
Performance improvements:
* Remove event fetching from DB threads (PR #835)
* Change the way we cache events (PR #836)
* Add events to cache when we persist them (PR #840)
Changes in synapse v0.16.0-rc1 (2016-06-03)
===========================================
Version 0.15 was not released. See v0.15.0-rc1 below for additional changes.
Features:
* Add email notifications for missed messages (PR #759, #786, #799, #810, #815,
#821)
* Add a ``url_preview_ip_range_whitelist`` config param (PR #760)
* Add /report endpoint (PR #762)
* Add basic ignore user API (PR #763)
* Add an openidish mechanism for proving that you own a given user_id (PR #765)
* Allow clients to specify a server_name to avoid 'No known servers' (PR #794)
* Add secondary_directory_servers option to fetch room list from other servers
(PR #808, #813)
Changes:
* Report per request metrics for all of the things using request_handler (PR
#756)
* Correctly handle ``NULL`` password hashes from the database (PR #775)
* Allow receipts for events we haven't seen in the db (PR #784)
* Make synctl read a cache factor from config file (PR #785)
* Increment badge count per missed convo, not per msg (PR #793)
* Special case m.room.third_party_invite event auth to match invites (PR #814)
Bug fixes:
* Fix typo in event_auth servlet path (PR #757)
* Fix password reset (PR #758)
Performance improvements:
* Reduce database inserts when sending transactions (PR #767)
* Queue events by room for persistence (PR #768)
* Add cache to ``get_user_by_id`` (PR #772)
* Add and use ``get_domain_from_id`` (PR #773)
* Use tree cache for ``get_linearized_receipts_for_room`` (PR #779)
* Remove unused indices (PR #782)
* Add caches to ``bulk_get_push_rules*`` (PR #804)
* Cache ``get_event_reference_hashes`` (PR #806)
* Add ``get_users_with_read_receipts_in_room`` cache (PR #809)
* Use state to calculate ``get_users_in_room`` (PR #811)
* Load push rules in storage layer so that they get cached (PR #825)
* Make ``get_joined_hosts_for_room`` use get_users_in_room (PR #828)
* Poke notifier on next reactor tick (PR #829)
* Change CacheMetrics to be quicker (PR #830)
Changes in synapse v0.15.0-rc1 (2016-04-26)
===========================================
Features:
* Add login support for Javascript Web Tokens, thanks to Niklas Riekenbrauck
(PR #671,#687)
* Add URL previewing support (PR #688)
* Add login support for LDAP, thanks to Christoph Witzany (PR #701)
* Add GET endpoint for pushers (PR #716)
Changes:
* Never notify for member events (PR #667)
* Deduplicate identical ``/sync`` requests (PR #668)
* Require user to have left room to forget room (PR #673)
* Use DNS cache if within TTL (PR #677)
* Let users see their own leave events (PR #699)
* Deduplicate membership changes (PR #700)
* Increase performance of pusher code (PR #705)
* Respond with error status 504 if failed to talk to remote server (PR #731)
* Increase search performance on postgres (PR #745)
Bug fixes:
* Fix bug where disabling all notifications still resulted in push (PR #678)
* Fix bug where users couldn't reject remote invites if remote refused (PR #691)
* Fix bug where synapse attempted to backfill from itself (PR #693)
* Fix bug where profile information was not correctly added when joining remote
rooms (PR #703)
* Fix bug where register API required incorrect key name for AS registration
(PR #727)
Changes in synapse v0.14.0 (2016-03-30)
=======================================
@@ -1178,7 +511,7 @@ Configuration:
* Add support for changing the bind host of the metrics listener via the
``metrics_bind_host`` option.
Changes in synapse v0.9.0-r5 (2015-05-21)
=========================================
@@ -1520,7 +853,7 @@ See UPGRADE for information about changes to the client server API, including
breaking backwards compatibility with VoIP calls and registration API.
Homeserver:
* When a user changes their displayname or avatar the server will now update
* When a user changes their displayname or avatar the server will now update
all their join states to reflect this.
* The server now adds "age" key to events to indicate how old they are. This
is clock independent, so at no point does any server or webclient have to
@@ -1578,7 +911,7 @@ Changes in synapse 0.2.2 (2014-09-06)
=====================================
Homeserver:
* When the server returns state events it now also includes the previous
* When the server returns state events it now also includes the previous
content.
* Add support for inviting people when creating a new room.
* Make the homeserver inform the room via `m.room.aliases` when a new alias
@@ -1590,7 +923,7 @@ Webclient:
* Handle `m.room.aliases` events.
* Asynchronously send messages and show a local echo.
* Inform the UI when a message failed to send.
* Only autoscroll on receiving a new message if the user was already at the
* Only autoscroll on receiving a new message if the user was already at the
bottom of the screen.
* Add support for ban/kick reasons.

View File

@@ -11,10 +11,8 @@ recursive-include synapse/storage/schema *.sql
recursive-include synapse/storage/schema *.py
recursive-include docs *
recursive-include res *
recursive-include scripts *
recursive-include scripts-dev *
recursive-include synapse *.pyi
recursive-include tests *.py
recursive-include synapse/static *.css
@@ -24,7 +22,5 @@ recursive-include synapse/static *.js
exclude jenkins.sh
exclude jenkins*.sh
exclude jenkins*
recursive-exclude jenkins *.sh
prune demo/etc

View File

@@ -11,8 +11,8 @@ VoIP. The basics you need to know to get up and running are:
like ``#matrix:matrix.org`` or ``#test:localhost:8448``.
- Matrix user IDs look like ``@matthew:matrix.org`` (although in the future
you will normally refer to yourself and others using a third party identifier
(3PID): email address, phone number, etc rather than manipulating Matrix user IDs)
you will normally refer to yourself and others using a 3PID: email
address, phone number, etc rather than manipulating Matrix user IDs)
The overall architecture is::
@@ -20,13 +20,12 @@ The overall architecture is::
https://somewhere.org/_matrix https://elsewhere.net/_matrix
``#matrix:matrix.org`` is the official support room for Matrix, and can be
accessed by any client from https://matrix.org/docs/projects/try-matrix-now or
via IRC bridge at irc://irc.freenode.net/matrix.
accessed by any client from https://matrix.org/blog/try-matrix-now or via IRC
bridge at irc://irc.freenode.net/matrix.
Synapse is currently in rapid development, but as of version 0.5 we believe it
is sufficiently stable to be run as an internet-facing service for real usage!
About Matrix
============
@@ -53,30 +52,39 @@ generation of fully open and interoperable messaging and VoIP apps for the
internet.
Synapse is a reference "homeserver" implementation of Matrix from the core
development team at matrix.org, written in Python/Twisted. It is intended to
showcase the concept of Matrix and let folks see the spec in the context of a
codebase and let you run your own homeserver and generally help bootstrap the
ecosystem.
development team at matrix.org, written in Python/Twisted for clarity and
simplicity. It is intended to showcase the concept of Matrix and let folks see
the spec in the context of a codebase and let you run your own homeserver and
generally help bootstrap the ecosystem.
In Matrix, every user runs one or more Matrix clients, which connect through to
a Matrix homeserver. The homeserver stores all their personal chat history and
user account information - much as a mail client connects through to an
IMAP/SMTP server. Just like email, you can either run your own Matrix
homeserver and control and own your own communications and history or use one
hosted by someone else (e.g. matrix.org) - there is no single point of control
or mandatory service provider in Matrix, unlike WhatsApp, Facebook, Hangouts,
etc.
a Matrix homeserver which stores all their personal chat history and user
account information - much as a mail client connects through to an IMAP/SMTP
server. Just like email, you can either run your own Matrix homeserver and
control and own your own communications and history or use one hosted by
someone else (e.g. matrix.org) - there is no single point of control or
mandatory service provider in Matrix, unlike WhatsApp, Facebook, Hangouts, etc.
Synapse ships with two basic demo Matrix clients: webclient (a basic group chat
web client demo implemented in AngularJS) and cmdclient (a basic Python
command line utility which lets you easily see what the JSON APIs are up to).
Meanwhile, iOS and Android SDKs and clients are available from:
- https://github.com/matrix-org/matrix-ios-sdk
- https://github.com/matrix-org/matrix-ios-kit
- https://github.com/matrix-org/matrix-ios-console
- https://github.com/matrix-org/matrix-android-sdk
We'd like to invite you to join #matrix:matrix.org (via
https://matrix.org/docs/projects/try-matrix-now), run a homeserver, take a look
at the `Matrix spec <https://matrix.org/docs/spec>`_, and experiment with the
`APIs <https://matrix.org/docs/api>`_ and `Client SDKs
<http://matrix.org/docs/projects/try-matrix-now.html#client-sdks>`_.
https://matrix.org/blog/try-matrix-now), run a homeserver, take a look at the
Matrix spec at https://matrix.org/docs/spec and API docs at
https://matrix.org/docs/api, experiment with the APIs and the demo clients, and
report any bugs via https://matrix.org/jira.
Thanks for using Matrix!
[1] End-to-end encryption is currently in beta: `blog post <https://matrix.org/blog/2016/11/21/matrixs-olm-end-to-end-encryption-security-assessment-released-and-implemented-cross-platform-on-riot-at-last>`_.
[1] End-to-end encryption is currently in development - see https://matrix.org/git/olm
Synapse Installation
====================
@@ -86,14 +94,9 @@ Synapse is the reference python/twisted Matrix homeserver implementation.
System requirements:
- POSIX-compliant system (tested on Linux & OS X)
- Python 2.7
- At least 1GB of free RAM if you want to join large public rooms like #matrix:matrix.org
- At least 512 MB RAM.
Installing from source
----------------------
(Prebuilt packages are available for some platforms - see `Platform-Specific
Instructions`_.)
Synapse is written in python but some of the libraries it uses are written in
Synapse is written in python but some of the libraries is uses are written in
C. So before we can install synapse itself we need a working C compiler and the
header files for python C extensions.
@@ -101,7 +104,7 @@ Installing prerequisites on Ubuntu or Debian::
sudo apt-get install build-essential python2.7-dev libffi-dev \
python-pip python-setuptools sqlite3 \
libssl-dev python-virtualenv libjpeg-dev libxslt1-dev
libssl-dev python-virtualenv libjpeg-dev
Installing prerequisites on ArchLinux::
@@ -120,7 +123,6 @@ Installing prerequisites on Mac OS X::
xcode-select --install
sudo easy_install pip
sudo pip install virtualenv
brew install pkg-config libffi
Installing prerequisites on Raspbian::
@@ -131,16 +133,6 @@ Installing prerequisites on Raspbian::
sudo pip install --upgrade ndg-httpsclient
sudo pip install --upgrade virtualenv
Installing prerequisites on openSUSE::
sudo zypper in -t pattern devel_basis
sudo zypper in python-pip python-setuptools sqlite3 python-virtualenv \
python-devel libffi-devel libopenssl-devel libjpeg62-devel
Installing prerequisites on OpenBSD::
doas pkg_add python libffi py-pip py-setuptools sqlite3 py-virtualenv \
libxslt
To install the synapse homeserver run::
virtualenv -p python2.7 ~/.synapse
@@ -152,74 +144,38 @@ This installs synapse, along with the libraries it uses, into a virtual
environment under ``~/.synapse``. Feel free to pick a different directory
if you prefer.
In case of problems, please see the _`Troubleshooting` section below.
In case of problems, please see the _Troubleshooting section below.
Alternatively, Silvio Fricke has contributed a Dockerfile to automate the
above in Docker at https://registry.hub.docker.com/u/silviof/docker-matrix/.
Also, Martin Giess has created an auto-deployment process with vagrant/ansible,
tested with VirtualBox/AWS/DigitalOcean - see https://github.com/EMnify/matrix-synapse-auto-deploy
Also, Martin Giess has created an auto-deployment process with vagrant/ansible,
tested with VirtualBox/AWS/DigitalOcean - see https://github.com/EMnify/matrix-synapse-auto-deploy
for details.
Configuring synapse
-------------------
Before you can start Synapse, you will need to generate a configuration
file. To do this, run (in your virtualenv, as before)::
To set up your homeserver, run (in your virtualenv, as before)::
cd ~/.synapse
python -m synapse.app.homeserver \
--server-name my.domain.name \
--server-name machine.my.domain.name \
--config-path homeserver.yaml \
--generate-config \
--report-stats=[yes|no]
... substituting an appropriate value for ``--server-name``. The server name
determines the "domain" part of user-ids for users on your server: these will
all be of the format ``@user:my.domain.name``. It also determines how other
matrix servers will reach yours for `Federation`_. For a test configuration,
set this to the hostname of your server. For a more production-ready setup, you
will probably want to specify your domain (``example.com``) rather than a
matrix-specific hostname here (in the same way that your email address is
probably ``user@example.com`` rather than ``user@email.example.com``) - but
doing so may require more advanced setup - see `Setting up
Federation`_. Beware that the server name cannot be changed later.
...substituting your host and domain name as appropriate.
This command will generate you a config file that you can then customise, but it will
This will generate you a config file that you can then customise, but it will
also generate a set of keys for you. These keys will allow your Home Server to
identify itself to other Home Servers, so don't lose or delete them. It would be
wise to back them up somewhere safe. (If, for whatever reason, you do need to
wise to back them up somewhere safe. If, for whatever reason, you do need to
change your Home Server's keys, you may find that other Home Servers have the
old key cached. If you update the signing key, you should change the name of the
key in the ``<server name>.signing.key`` file (the second word) to something
different. See `the spec`__ for more information on key management.)
key in the <server name>.signing.key file (the second word) to something different.
.. __: `key_management`_
The default configuration exposes two HTTP ports: 8008 and 8448. Port 8008 is
configured without TLS; it is not recommended this be exposed outside your
local network. Port 8448 is configured to use TLS with a self-signed
certificate. This is fine for testing with but, to avoid your clients
complaining about the certificate, you will almost certainly want to use
another certificate for production purposes. (Note that a self-signed
certificate is fine for `Federation`_). You can do so by changing
``tls_certificate_path``, ``tls_private_key_path`` and ``tls_dh_params_path``
in ``homeserver.yaml``; alternatively, you can use a reverse-proxy, but be sure
to read `Using a reverse proxy with Synapse`_ when doing so.
Apart from port 8448 using TLS, both ports are the same in the default
configuration.
Registering a user
------------------
You will need at least one user on your server in order to use a Matrix
client. Users can be registered either `via a Matrix client`__, or via a
commandline script.
.. __: `client-user-reg`_
To get started, it is easiest to use the command line to register new users::
By default, registration of new users is disabled. You can either enable
registration in the config by specifying ``enable_registration: true``
(it is then recommended to also set up CAPTCHA - see docs/CAPTCHA_SETUP), or
you can use the command line to register new users::
$ source ~/.synapse/bin/activate
$ synctl start # if not already running
@@ -229,19 +185,8 @@ To get started, it is easiest to use the command line to register new users::
Confirm password:
Success!
This process uses a setting ``registration_shared_secret`` in
``homeserver.yaml``, which is shared between Synapse itself and the
``register_new_matrix_user`` script. It doesn't matter what it is (a random
value is generated by ``--generate-config``), but it should be kept secret, as
anyone with knowledge of it can register users on your server even if
``enable_registration`` is ``false``.
Setting up a TURN server
------------------------
For reliable VoIP calls to be routed via this homeserver, you MUST configure
a TURN server. See `<docs/turn-howto.rst>`_ for details.
a TURN server. See docs/turn-howto.rst for details.
Running Synapse
===============
@@ -253,66 +198,29 @@ run (e.g. ``~/.synapse``), and::
source ./bin/activate
synctl start
Using PostgreSQL
================
Connecting to Synapse from a client
===================================
As of Synapse 0.9, `PostgreSQL <http://www.postgresql.org>`_ is supported as an
alternative to the `SQLite <http://sqlite.org/>`_ database that Synapse has
traditionally used for convenience and simplicity.
The easiest way to try out your new Synapse installation is by connecting to it
from a web client. The easiest option is probably the one at
http://riot.im/app. You will need to specify a "Custom server" when you log on
or register: set this to ``https://localhost:8448`` - remember to specify the
port (``:8448``) unless you changed the configuration. (Leave the identity
server as the default - see `Identity servers`_.)
The advantages of Postgres include:
If all goes well you should at least be able to log in, create a room, and
start sending messages.
* significant performance improvements due to the superior threading and
caching model, smarter query optimiser
* allowing the DB to be run on separate hardware
* allowing basic active/backup high-availability with a "hot spare" synapse
pointing at the same DB master, as well as enabling DB replication in
synapse itself.
(The homeserver runs a web client by default at https://localhost:8448/, though
as of the time of writing it is somewhat outdated and not really recommended -
https://github.com/matrix-org/synapse/issues/1527).
The only disadvantage is that the code is relatively new as of April 2015 and
may have a few regressions relative to SQLite.
.. _`client-user-reg`:
For information on how to install and use PostgreSQL, please see
`docs/postgres.rst <docs/postgres.rst>`_.
Registering a new user from a client
------------------------------------
By default, registration of new users via Matrix clients is disabled. To enable
it, specify ``enable_registration: true`` in ``homeserver.yaml``. (It is then
recommended to also set up CAPTCHA - see `<docs/CAPTCHA_SETUP.rst>`_.)
Once ``enable_registration`` is set to ``true``, it is possible to register a
user via `riot.im <https://riot.im/app/#/register>`_ or other Matrix clients.
Your new user name will be formed partly from the ``server_name`` (see
`Configuring synapse`_), and partly from a localpart you specify when you
create the account. Your name will take the form of::
@localpart:my.domain.name
(pronounced "at localpart on my dot domain dot name").
As when logging in, you will need to specify a "Custom server". Specify your
desired ``localpart`` in the 'User name' box.
Security Note
=============
Matrix serves raw user generated data in some APIs - specifically the `content
repository endpoints <http://matrix.org/docs/spec/client_server/latest.html#get-matrix-media-r0-download-servername-mediaid>`_.
Whilst we have tried to mitigate against possible XSS attacks (e.g.
https://github.com/matrix-org/synapse/pull/1021) we recommend running
matrix homeservers on a dedicated domain name, to limit any malicious user generated
content served to web browsers a matrix API from being able to attack webapps hosted
on the same domain. This is particularly true of sharing a matrix webclient and
server on the same domain.
See https://github.com/vector-im/vector-web/issues/1977 and
https://developer.github.com/changes/2014-04-25-user-content-security for more details.
Platform-Specific Instructions
Platform Specific Instructions
==============================
Debian
@@ -320,7 +228,7 @@ Debian
Matrix provides official Debian packages via apt from http://matrix.org/packages/debian/.
Note that these packages do not include a client - choose one from
https://matrix.org/docs/projects/try-matrix-now/ (or build your own with one of our SDKs :)
https://matrix.org/blog/try-matrix-now/ (or build your own with one of our SDKs :)
Fedora
------
@@ -374,32 +282,6 @@ Synapse can be installed via FreeBSD Ports or Packages contributed by Brendan Mo
- Ports: ``cd /usr/ports/net/py-matrix-synapse && make install clean``
- Packages: ``pkg install py27-matrix-synapse``
OpenBSD
-------
There is currently no port for OpenBSD. Additionally, OpenBSD's security
settings require a slightly more difficult installation process.
1) Create a new directory in ``/usr/local`` called ``_synapse``. Also, create a
new user called ``_synapse`` and set that directory as the new user's home.
This is required because, by default, OpenBSD only allows binaries which need
write and execute permissions on the same memory space to be run from
``/usr/local``.
2) ``su`` to the new ``_synapse`` user and change to their home directory.
3) Create a new virtualenv: ``virtualenv -p python2.7 ~/.synapse``
4) Source the virtualenv configuration located at
``/usr/local/_synapse/.synapse/bin/activate``. This is done in ``ksh`` by
using the ``.`` command, rather than ``bash``'s ``source``.
5) Optionally, use ``pip`` to install ``lxml``, which Synapse needs to parse
webpages for their titles.
6) Use ``pip`` to install this repository: ``pip install
https://github.com/matrix-org/synapse/tarball/master``
7) Optionally, change ``_synapse``'s shell to ``/bin/false`` to reduce the
chance of a compromised Synapse server being used to take over your box.
After this, you may proceed with the rest of the install directions.
NixOS
-----
@@ -439,7 +321,6 @@ Troubleshooting:
you do, you may need to create a symlink to ``libsodium.a`` so ``ld`` can find
it: ``ln -s /usr/local/lib/libsodium.a /usr/lib/libsodium.a``
Troubleshooting
===============
@@ -513,292 +394,9 @@ you will need to explicitly call Python2.7 - either running as::
...or by editing synctl with the correct python executable.
Upgrading an existing Synapse
=============================
The instructions for upgrading synapse are in `UPGRADE.rst`_.
Please check these instructions as upgrading may require extra steps for some
versions of synapse.
.. _UPGRADE.rst: UPGRADE.rst
.. _federation:
Setting up Federation
=====================
Federation is the process by which users on different servers can participate
in the same room. For this to work, those other servers must be able to contact
yours to send messages.
As explained in `Configuring synapse`_, the ``server_name`` in your
``homeserver.yaml`` file determines the way that other servers will reach
yours. By default, they will treat it as a hostname and try to connect to
port 8448. This is easy to set up and will work with the default configuration,
provided you set the ``server_name`` to match your machine's public DNS
hostname.
For a more flexible configuration, you can set up a DNS SRV record. This allows
you to run your server on a machine that might not have the same name as your
domain name. For example, you might want to run your server at
``synapse.example.com``, but have your Matrix user-ids look like
``@user:example.com``. (A SRV record also allows you to change the port from
the default 8448. However, if you are thinking of using a reverse-proxy, be
sure to read `Reverse-proxying the federation port`_ first.)
To use a SRV record, first create your SRV record and publish it in DNS. This
should have the format ``_matrix._tcp.<yourdomain.com> <ttl> IN SRV 10 0 <port>
<synapse.server.name>``. The DNS record should then look something like::
$ dig -t srv _matrix._tcp.example.com
_matrix._tcp.example.com. 3600 IN SRV 10 0 8448 synapse.example.com.
You can then configure your homeserver to use ``<yourdomain.com>`` as the domain in
its user-ids, by setting ``server_name``::
python -m synapse.app.homeserver \
--server-name <yourdomain.com> \
--config-path homeserver.yaml \
--generate-config
python -m synapse.app.homeserver --config-path homeserver.yaml
If you've already generated the config file, you need to edit the ``server_name``
in your ``homeserver.yaml`` file. If you've already started Synapse and a
database has been created, you will have to recreate the database.
If all goes well, you should be able to `connect to your server with a client`__,
and then join a room via federation. (Try ``#matrix-dev:matrix.org`` as a first
step. "Matrix HQ"'s sheer size and activity level tends to make even the
largest boxes pause for thought.)
.. __: `Connecting to Synapse from a client`_
Troubleshooting
---------------
The typical failure mode with federation is that when you try to join a room,
it is rejected with "401: Unauthorized". Generally this means that other
servers in the room couldn't access yours. (Joining a room over federation is a
complicated dance which requires connections in both directions).
So, things to check are:
* If you are trying to use a reverse-proxy, read `Reverse-proxying the
federation port`_.
* If you are not using a SRV record, check that your ``server_name`` (the part
of your user-id after the ``:``) matches your hostname, and that port 8448 on
that hostname is reachable from outside your network.
* If you *are* using a SRV record, check that it matches your ``server_name``
(it should be ``_matrix._tcp.<server_name>``), and that the port and hostname
it specifies are reachable from outside your network.
Running a Demo Federation of Synapses
-------------------------------------
If you want to get up and running quickly with a trio of homeservers in a
private federation, there is a script in the ``demo`` directory. This is mainly
useful just for development purposes. See `<demo/README>`_.
Using PostgreSQL
================
As of Synapse 0.9, `PostgreSQL <http://www.postgresql.org>`_ is supported as an
alternative to the `SQLite <http://sqlite.org/>`_ database that Synapse has
traditionally used for convenience and simplicity.
The advantages of Postgres include:
* significant performance improvements due to the superior threading and
caching model, smarter query optimiser
* allowing the DB to be run on separate hardware
* allowing basic active/backup high-availability with a "hot spare" synapse
pointing at the same DB master, as well as enabling DB replication in
synapse itself.
For information on how to install and use PostgreSQL, please see
`docs/postgres.rst <docs/postgres.rst>`_.
.. _reverse-proxy:
Using a reverse proxy with Synapse
==================================
It is possible to put a reverse proxy such as
`nginx <https://nginx.org/en/docs/http/ngx_http_proxy_module.html>`_,
`Apache <https://httpd.apache.org/docs/current/mod/mod_proxy_http.html>`_ or
`HAProxy <http://www.haproxy.org/>`_ in front of Synapse. One advantage of
doing so is that it means that you can expose the default https port (443) to
Matrix clients without needing to run Synapse with root privileges.
The most important thing to know here is that Matrix clients and other Matrix
servers do not necessarily need to connect to your server via the same
port. Indeed, clients will use port 443 by default, whereas servers default to
port 8448. Where these are different, we refer to the 'client port' and the
'federation port'.
The next most important thing to know is that using a reverse-proxy on the
federation port has a number of pitfalls. It is possible, but be sure to read
`Reverse-proxying the federation port`_.
The recommended setup is therefore to configure your reverse-proxy on port 443
for client connections, but to also expose port 8448 for server-server
connections. All the Matrix endpoints begin ``/_matrix``, so an example nginx
configuration might look like::
server {
listen 443 ssl;
listen [::]:443 ssl;
server_name matrix.example.com;
location /_matrix {
proxy_pass http://localhost:8008;
proxy_set_header X-Forwarded-For $remote_addr;
}
}
You will also want to set ``bind_address: 127.0.0.1`` and ``x_forwarded: true``
for port 8008 in ``homeserver.yaml`` to ensure that client IP addresses are
recorded correctly.
Having done so, you can then use ``https://matrix.example.com`` (instead of
``https://matrix.example.com:8448``) as the "Custom server" when `Connecting to
Synapse from a client`_.
Reverse-proxying the federation port
------------------------------------
There are two issues to consider before using a reverse-proxy on the federation
port:
* Due to the way SSL certificates are managed in the Matrix federation protocol
(see `spec`__), Synapse needs to be configured with the path to the SSL
certificate, *even if you do not terminate SSL at Synapse*.
.. __: `key_management`_
* Synapse does not currently support SNI on the federation protocol
(`bug #1491 <https://github.com/matrix-org/synapse/issues/1491>`_), which
means that using name-based virtual hosting is unreliable.
Furthermore, a number of the normal reasons for using a reverse-proxy do not
apply:
* Other servers will connect on port 8448 by default, so there is no need to
listen on port 443 (for federation, at least), which avoids the need for root
privileges and virtual hosting.
* A self-signed SSL certificate is fine for federation, so there is no need to
automate renewals. (The certificate generated by ``--generate-config`` is
valid for 10 years.)
If you want to set up a reverse-proxy on the federation port despite these
caveats, you will need to do the following:
* In ``homeserver.yaml``, set ``tls_certificate_path`` to the path to the SSL
certificate file used by your reverse-proxy, and set ``no_tls`` to ``True``.
(``tls_private_key_path`` will be ignored if ``no_tls`` is ``True``.)
* In your reverse-proxy configuration:
* If there are other virtual hosts on the same port, make sure that the
*default* one uses the certificate configured above.
* Forward ``/_matrix`` to Synapse.
* If your reverse-proxy is not listening on port 8448, publish a SRV record to
tell other servers how to find you. See `Setting up Federation`_.
When updating the SSL certificate, just update the file pointed to by
``tls_certificate_path``: there is no need to restart synapse. (You may like to
use a symbolic link to help make this process atomic.)
The most common mistake when setting up federation is not to tell Synapse about
your SSL certificate. To check it, you can visit
``https://matrix.org/federationtester/api/report?server_name=<your_server_name>``.
Unfortunately, there is no UI for this yet, but, you should see
``"MatchingTLSFingerprint": true``. If not, check that
``Certificates[0].SHA256Fingerprint`` (the fingerprint of the certificate
presented by your reverse-proxy) matches ``Keys.tls_fingerprints[0].sha256``
(the fingerprint of the certificate Synapse is using).
Identity Servers
================
Identity servers have the job of mapping email addresses and other 3rd Party
IDs (3PIDs) to Matrix user IDs, as well as verifying the ownership of 3PIDs
before creating that mapping.
**They are not where accounts or credentials are stored - these live on home
servers. Identity Servers are just for mapping 3rd party IDs to matrix IDs.**
This process is very security-sensitive, as there is obvious risk of spam if it
is too easy to sign up for Matrix accounts or harvest 3PID data. In the longer
term, we hope to create a decentralised system to manage it (`matrix-doc #712
<https://github.com/matrix-org/matrix-doc/issues/712>`_), but in the meantime,
the role of managing trusted identity in the Matrix ecosystem is farmed out to
a cluster of known trusted ecosystem partners, who run 'Matrix Identity
Servers' such as `Sydent <https://github.com/matrix-org/sydent>`_, whose role
is purely to authenticate and track 3PID logins and publish end-user public
keys.
You can host your own copy of Sydent, but this will prevent you reaching other
users in the Matrix ecosystem via their email address, and prevent them finding
you. We therefore recommend that you use one of the centralised identity servers
at ``https://matrix.org`` or ``https://vector.im`` for now.
To reiterate: the Identity server will only be used if you choose to associate
an email address with your account, or send an invite to another user via their
email address.
URL Previews
============
Synapse 0.15.0 introduces a new API for previewing URLs at
``/_matrix/media/r0/preview_url``. This is disabled by default. To turn it on
you must enable the ``url_preview_enabled: True`` config parameter and
explicitly specify the IP ranges that Synapse is not allowed to spider for
previewing in the ``url_preview_ip_range_blacklist`` configuration parameter.
This is critical from a security perspective to stop arbitrary Matrix users
spidering 'internal' URLs on your network. At the very least we recommend that
your loopback and RFC1918 IP addresses are blacklisted.
This also requires the optional lxml and netaddr python dependencies to be
installed.
Password reset
==============
If a user has registered an email address to their account using an identity
server, they can request a password-reset token via clients such as Vector.
A manual password reset can be done via direct database access as follows.
First calculate the hash of the new password::
$ source ~/.synapse/bin/activate
$ ./scripts/hash_password
Password:
Confirm password:
$2a$12$xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Then update the `users` table in the database::
UPDATE users SET password_hash='$2a$12$xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx'
WHERE name='@test:test.com';
Synapse Development
===================
Before setting up a development environment for synapse, make sure you have the
system dependencies (such as the python header files) installed - see
`Installing from source`_.
To check out a synapse for development, clone the git repo into a working
directory of your choice::
@@ -810,8 +408,8 @@ to install using pip and a virtualenv::
virtualenv env
source env/bin/activate
python synapse/python_dependencies.py | xargs pip install
pip install lxml mock
python synapse/python_dependencies.py | xargs -n1 pip install
pip install setuptools_trial mock
This will run a process of downloading and installing all the needed
dependencies into a virtual env.
@@ -819,7 +417,7 @@ dependencies into a virtual env.
Once this is done, you may wish to run Synapse's unit tests, to
check that everything is installed as it should be::
PYTHONPATH="." trial tests
python setup.py test
This should end with a 'PASSED' result::
@@ -828,6 +426,165 @@ This should end with a 'PASSED' result::
PASSED (successes=143)
Upgrading an existing Synapse
=============================
The instructions for upgrading synapse are in `UPGRADE.rst`_.
Please check these instructions as upgrading may require extra steps for some
versions of synapse.
.. _UPGRADE.rst: UPGRADE.rst
Setting up Federation
=====================
In order for other homeservers to send messages to your server, it will need to
be publicly visible on the internet, and they will need to know its host name.
You have two choices here, which will influence the form of your Matrix user
IDs:
1) Use the machine's own hostname as available on public DNS in the form of
its A or AAAA records. This is easier to set up initially, perhaps for
testing, but lacks the flexibility of SRV.
2) Set up a SRV record for your domain name. This requires you create a SRV
record in DNS, but gives the flexibility to run the server on your own
choice of TCP port, on a machine that might not be the same name as the
domain name.
For the first form, simply pass the required hostname (of the machine) as the
--server-name parameter::
python -m synapse.app.homeserver \
--server-name machine.my.domain.name \
--config-path homeserver.yaml \
--generate-config
python -m synapse.app.homeserver --config-path homeserver.yaml
Alternatively, you can run ``synctl start`` to guide you through the process.
For the second form, first create your SRV record and publish it in DNS. This
needs to be named _matrix._tcp.YOURDOMAIN, and point at at least one hostname
and port where the server is running. (At the current time synapse does not
support clustering multiple servers into a single logical homeserver). The DNS
record would then look something like::
$ dig -t srv _matrix._tcp.machine.my.domain.name
_matrix._tcp IN SRV 10 0 8448 machine.my.domain.name.
At this point, you should then run the homeserver with the hostname of this
SRV record, as that is the name other machines will expect it to have::
python -m synapse.app.homeserver \
--server-name YOURDOMAIN \
--config-path homeserver.yaml \
--generate-config
python -m synapse.app.homeserver --config-path homeserver.yaml
If you've already generated the config file, you need to edit the "server_name"
in you ```homeserver.yaml``` file. If you've already started Synapse and a
database has been created, you will have to recreate the database.
You may additionally want to pass one or more "-v" options, in order to
increase the verbosity of logging output; at least for initial testing.
Running a Demo Federation of Synapses
-------------------------------------
If you want to get up and running quickly with a trio of homeservers in a
private federation (``localhost:8080``, ``localhost:8081`` and
``localhost:8082``) which you can then access through the webclient running at
http://localhost:8080. Simply run::
demo/start.sh
This is mainly useful just for development purposes.
Running The Demo Web Client
===========================
The homeserver runs a web client by default at https://localhost:8448/.
If this is the first time you have used the client from that browser (it uses
HTML5 local storage to remember its config), you will need to log in to your
account. If you don't yet have an account, because you've just started the
homeserver for the first time, then you'll need to register one.
Registering A New Account
-------------------------
Your new user name will be formed partly from the hostname your server is
running as, and partly from a localpart you specify when you create the
account. Your name will take the form of::
@localpart:my.domain.here
(pronounced "at localpart on my dot domain dot here")
Specify your desired localpart in the topmost box of the "Register for an
account" form, and click the "Register" button. Hostnames can contain ports if
required due to lack of SRV records (e.g. @matthew:localhost:8448 on an
internal synapse sandbox running on localhost).
If registration fails, you may need to enable it in the homeserver (see
`Synapse Installation`_ above)
Logging In To An Existing Account
---------------------------------
Just enter the ``@localpart:my.domain.here`` Matrix user ID and password into
the form and click the Login button.
Identity Servers
================
The job of authenticating 3PIDs and tracking which 3PIDs are associated with a
given Matrix user is very security-sensitive, as there is obvious risk of spam
if it is too easy to sign up for Matrix accounts or harvest 3PID data.
Meanwhile the job of publishing the end-to-end encryption public keys for
Matrix users is also very security-sensitive for similar reasons.
Therefore the role of managing trusted identity in the Matrix ecosystem is
farmed out to a cluster of known trusted ecosystem partners, who run 'Matrix
Identity Servers' such as ``sydent``, whose role is purely to authenticate and
track 3PID logins and publish end-user public keys.
It's currently early days for identity servers as Matrix is not yet using 3PIDs
as the primary means of identity and E2E encryption is not complete. As such,
we are running a single identity server (https://matrix.org) at the current
time.
Password reset
==============
If a user has registered an email address to their account using an identity
server, they can request a password-reset token via clients such as Vector.
A manual password reset can be done via direct database access as follows.
First calculate the hash of the new password:
$ source ~/.synapse/bin/activate
$ ./scripts/hash_password
Password:
Confirm password:
$2a$12$xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Then update the `users` table in the database:
UPDATE users SET password_hash='$2a$12$xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx'
WHERE name='@test:test.com';
Where's the spec?!
==================
The source of the matrix spec lives at https://github.com/matrix-org/matrix-doc.
A recent HTML snapshot of this lives at http://matrix.org/docs/spec
Building Internal API Documentation
===================================
@@ -842,7 +599,8 @@ Building internal API documentation::
python setup.py build_sphinx
Help!! Synapse eats all my RAM!
Halp!! Synapse eats all my RAM!
===============================
Synapse's architecture is quite RAM hungry currently - we deliberately
@@ -857,5 +615,3 @@ around a ~700MB footprint. You can dial it down further to 0.02 if
desired, which targets roughly ~512MB. Conversely you can dial it up if
you need performance for lots of users and have a box with a lot of RAM.
.. _`key_management`: https://matrix.org/docs/spec/server_server/unstable.html#retrieving-server-keys

View File

@@ -27,15 +27,7 @@ running:
# Pull the latest version of the master branch.
git pull
# Update the versions of synapse's python dependencies.
python synapse/python_dependencies.py | xargs -n1 pip install --upgrade
Upgrading to v0.15.0
====================
If you want to use the new URL previewing API (/_matrix/media/r0/preview_url)
then you have to explicitly enable it in the config and update your dependencies
dependencies. See README.rst for details.
python synapse/python_dependencies.py | xargs -n1 pip install
Upgrading to v0.11.0

View File

@@ -9,7 +9,6 @@ Description=Synapse Matrix homeserver
Type=simple
User=synapse
Group=synapse
EnvironmentFile=-/etc/sysconfig/synapse
WorkingDirectory=/var/lib/synapse
ExecStart=/usr/bin/python2.7 -m synapse.app.homeserver --config-path=/etc/synapse/homeserver.yaml --log-config=/etc/synapse/log_config.yaml

View File

@@ -10,13 +10,13 @@ https://developers.google.com/recaptcha/
Setting ReCaptcha Keys
----------------------
The keys are a config option on the home server config. If they are not
visible, you can generate them via --generate-config. Set the following value::
The keys are a config option on the home server config. If they are not
visible, you can generate them via --generate-config. Set the following value:
recaptcha_public_key: YOUR_PUBLIC_KEY
recaptcha_private_key: YOUR_PRIVATE_KEY
In addition, you MUST enable captchas via::
In addition, you MUST enable captchas via:
enable_registration_captcha: true
@@ -25,6 +25,7 @@ Configuring IP used for auth
The ReCaptcha API requires that the IP address of the user who solved the
captcha is sent. If the client is connecting through a proxy or load balancer,
it may be required to use the X-Forwarded-For (XFF) header instead of the origin
IP address. This can be configured as an option on the home server like so::
IP address. This can be configured as an option on the home server like so:
captcha_ip_origin_is_x_forwarded: true

View File

@@ -1,12 +0,0 @@
Admin APIs
==========
This directory includes documentation for the various synapse specific admin
APIs available.
Only users that are server admins can use these APIs. A user can be marked as a
server admin by updating the database directly, e.g.:
``UPDATE users SET admin = 1 WHERE name = '@foo:bar.com'``
Restarting may be required for the changes to register.

View File

@@ -1,15 +0,0 @@
Purge History API
=================
The purge history API allows server admins to purge historic events from their
database, reclaiming disk space.
Depending on the amount of history being purged a call to the API may take
several minutes or longer. During this period users will not be able to
paginate further back in the room from the point being purged from.
The API is simply:
``POST /_matrix/client/r0/admin/purge_history/<room_id>/<event_id>``
including an ``access_token`` of a server admin.

View File

@@ -1,19 +0,0 @@
Purge Remote Media API
======================
The purge remote media API allows server admins to purge old cached remote
media.
The API is::
POST /_matrix/client/r0/admin/purge_media_cache
{
"before_ts": <unix_timestamp_in_ms>
}
Which will remove all cached media that was last accessed before
``<unix_timestamp_in_ms>``.
If the user re-requests purged remote media, synapse will re-request the media
from the originating server.

View File

@@ -32,4 +32,5 @@ The format of the AS configuration file is as follows:
See the spec_ for further details on how application services work.
.. _spec: https://matrix.org/docs/spec/application_service/unstable.html
.. _spec: https://github.com/matrix-org/matrix-doc/blob/master/specification/25_application_service_api.rst#application-service-api

View File

@@ -43,10 +43,7 @@ Basically, PEP8
together, or want to deliberately extend or preserve vertical/horizontal
space)
Comments should follow the `google code style <http://google.github.io/styleguide/pyguide.html?showone=Comments#Comments>`_.
This is so that we can generate documentation with
`sphinx <http://sphinxcontrib-napoleon.readthedocs.org/en/latest/>`_. See the
`examples <http://sphinxcontrib-napoleon.readthedocs.io/en/latest/example_google.html>`_
in the sphinx documentation.
Comments should follow the google code style. This is so that we can generate
documentation with sphinx (http://sphinxcontrib-napoleon.readthedocs.org/en/latest/)
Code should pass pep8 --max-line-length=100 without any warnings.

View File

@@ -1,10 +0,0 @@
What do I do about "Unexpected logging context" debug log-lines everywhere?
<Mjark> The logging context lives in thread local storage
<Mjark> Sometimes it gets out of sync with what it should actually be, usually because something scheduled something to run on the reactor without preserving the logging context.
<Matthew> what is the impact of it getting out of sync? and how and when should we preserve log context?
<Mjark> The impact is that some of the CPU and database metrics will be under-reported, and some log lines will be mis-attributed.
<Mjark> It should happen auto-magically in all the APIs that do IO or otherwise defer to the reactor.
<Erik> Mjark: the other place is if we branch, e.g. using defer.gatherResults
Unanswered: how and when should we preserve log context?

View File

@@ -15,45 +15,36 @@ How to monitor Synapse metrics using Prometheus
Restart synapse
3: Add a prometheus target for synapse. It needs to set the ``metrics_path``
to a non-default value::
3: Check out synapse-prometheus-config
https://github.com/matrix-org/synapse-prometheus-config
- job_name: "synapse"
metrics_path: "/_synapse/metrics"
static_configs:
- targets:
"my.server.here:9092"
4: Add ``synapse.html`` and ``synapse.rules``
The ``.html`` file needs to appear in prometheus's ``consoles`` directory,
and the ``.rules`` file needs to be invoked somewhere in the main config
file. A symlink to each from the git checkout into the prometheus directory
might be easiest to ensure ``git pull`` keeps it updated.
Standard Metric Names
---------------------
5: Add a prometheus target for synapse
This is easiest if prometheus runs on the same machine as synapse, as it can
then just use localhost::
As of synapse version 0.18.2, the format of the process-wide metrics has been
changed to fit prometheus standard naming conventions. Additionally the units
have been changed to seconds, from miliseconds.
global: {
rule_file: "synapse.rules"
}
================================== =============================
New name Old name
---------------------------------- -----------------------------
process_cpu_user_seconds_total process_resource_utime / 1000
process_cpu_system_seconds_total process_resource_stime / 1000
process_open_fds (no 'type' label) process_fds
================================== =============================
job: {
name: "synapse"
The python-specific counts of garbage collector performance have been renamed.
target_group: {
target: "http://localhost:9092/"
}
}
=========================== ======================
New name Old name
--------------------------- ----------------------
python_gc_time reactor_gc_time
python_gc_unreachable_total reactor_gc_unreachable
python_gc_counts reactor_gc_counts
=========================== ======================
6: Start prometheus::
The twisted-specific reactor metrics have been renamed.
./prometheus -config.file=prometheus.conf
==================================== =====================
New name Old name
------------------------------------ ---------------------
python_twisted_reactor_pending_calls reactor_pending_calls
python_twisted_reactor_tick_time reactor_tick_time
==================================== =====================
7: Wait a few seconds for it to start and perform the first scrape,
then visit the console:
http://server-where-prometheus-runs:9090/consoles/synapse.html

View File

@@ -1,58 +0,0 @@
Replication Architecture
========================
Motivation
----------
We'd like to be able to split some of the work that synapse does into multiple
python processes. In theory multiple synapse processes could share a single
postgresql database and we'd scale up by running more synapse processes.
However much of synapse assumes that only one process is interacting with the
database, both for assigning unique identifiers when inserting into tables,
notifying components about new updates, and for invalidating its caches.
So running multiple copies of the current code isn't an option. One way to
run multiple processes would be to have a single writer process and multiple
reader processes connected to the same database. In order to do this we'd need
a way for the reader process to invalidate its in-memory caches when an update
happens on the writer. One way to do this is for the writer to present an
append-only log of updates which the readers can consume to invalidate their
caches and to push updates to listening clients or pushers.
Synapse already stores much of its data as an append-only log so that it can
correctly respond to /sync requests so the amount of code changes needed to
expose the append-only log to the readers should be fairly minimal.
Architecture
------------
The Replication API
~~~~~~~~~~~~~~~~~~~
Synapse will optionally expose a long poll HTTP API for extracting updates. The
API will have a similar shape to /sync in that clients provide tokens
indicating where in the log they have reached and a timeout. The synapse server
then either responds with updates immediately if it already has updates or it
waits until the timeout for more updates. If the timeout expires and nothing
happened then the server returns an empty response.
However unlike the /sync API this replication API is returning synapse specific
data rather than trying to implement a matrix specification. The replication
results are returned as arrays of rows where the rows are mostly lifted
directly from the database. This avoids unnecessary JSON parsing on the server
and hopefully avoids an impedance mismatch between the data returned and the
required updates to the datastore.
This does not replicate all the database tables as many of the database tables
are indexes that can be recovered from the contents of other tables.
The format and parameters for the api are documented in
``synapse/replication/resource.py``.
The Slaved DataStore
~~~~~~~~~~~~~~~~~~~~
There are read-only version of the synapse storage layer in
``synapse/replication/slave/storage`` that use the response of the replication
API to invalidate their caches.

View File

@@ -9,35 +9,31 @@ the Home Server to generate credentials that are valid for use on the TURN
server through the use of a secret shared between the Home Server and the
TURN server.
This document describes how to install coturn
(https://github.com/coturn/coturn) which also supports the TURN REST API,
This document described how to install coturn
(https://code.google.com/p/coturn/) which also supports the TURN REST API,
and integrate it with synapse.
coturn Setup
============
You may be able to setup coturn via your package manager, or set it up manually using the usual ``configure, make, make install`` process.
1. Check out coturn::
git clone https://github.com/coturn/coturn.git coturn
svn checkout http://coturn.googlecode.com/svn/trunk/ coturn
cd coturn
2. Configure it::
./configure
You may need to install ``libevent2``: if so, you should do so
You may need to install libevent2: if so, you should do so
in the way recommended by your operating system.
You can ignore warnings about lack of database support: a
database is unnecessary for this purpose.
3. Build and install it::
make
make install
4. Create or edit the config file in ``/etc/turnserver.conf``. The relevant
4. Make a config file in /etc/turnserver.conf. You can customise
a config file from turnserver.conf.default. The relevant
lines, with example values, are::
lt-cred-mech
@@ -45,7 +41,7 @@ You may be able to setup coturn via your package manager, or set it up manually
static-auth-secret=[your secret key here]
realm=turn.myserver.org
See turnserver.conf for explanations of the options.
See turnserver.conf.default for explanations of the options.
One way to generate the static-auth-secret is with pwgen::
pwgen -s 64 1
@@ -58,7 +54,6 @@ You may be able to setup coturn via your package manager, or set it up manually
import your private key and certificate.
7. Start the turn server::
bin/turnserver -o

View File

@@ -1,74 +0,0 @@
URL Previews
============
Design notes on a URL previewing service for Matrix:
Options are:
1. Have an AS which listens for URLs, downloads them, and inserts an event that describes their metadata.
* Pros:
* Decouples the implementation entirely from Synapse.
* Uses existing Matrix events & content repo to store the metadata.
* Cons:
* Which AS should provide this service for a room, and why should you trust it?
* Doesn't work well with E2E; you'd have to cut the AS into every room
* the AS would end up subscribing to every room anyway.
2. Have a generic preview API (nothing to do with Matrix) that provides a previewing service:
* Pros:
* Simple and flexible; can be used by any clients at any point
* Cons:
* If each HS provides one of these independently, all the HSes in a room may needlessly DoS the target URI
* We need somewhere to store the URL metadata rather than just using Matrix itself
* We can't piggyback on matrix to distribute the metadata between HSes.
3. Make the synapse of the sending user responsible for spidering the URL and inserting an event asynchronously which describes the metadata.
* Pros:
* Works transparently for all clients
* Piggy-backs nicely on using Matrix for distributing the metadata.
* No confusion as to which AS
* Cons:
* Doesn't work with E2E
* We might want to decouple the implementation of the spider from the HS, given spider behaviour can be quite complicated and evolve much more rapidly than the HS. It's more like a bot than a core part of the server.
4. Make the sending client use the preview API and insert the event itself when successful.
* Pros:
* Works well with E2E
* No custom server functionality
* Lets the client customise the preview that they send (like on FB)
* Cons:
* Entirely specific to the sending client, whereas it'd be nice if /any/ URL was correctly previewed if clients support it.
5. Have the option of specifying a shared (centralised) previewing service used by a room, to avoid all the different HSes in the room DoSing the target.
Best solution is probably a combination of both 2 and 4.
* Sending clients do their best to create and send a preview at the point of sending the message, perhaps delaying the message until the preview is computed? (This also lets the user validate the preview before sending)
* Receiving clients have the option of going and creating their own preview if one doesn't arrive soon enough (or if the original sender didn't create one)
This is a bit magical though in that the preview could come from two entirely different sources - the sending HS or your local one. However, this can always be exposed to users: "Generate your own URL previews if none are available?"
This is tantamount also to senders calculating their own thumbnails for sending in advance of the main content - we are trusting the sender not to lie about the content in the thumbnail. Whereas currently thumbnails are calculated by the receiving homeserver to avoid this attack.
However, this kind of phishing attack does exist whether we let senders pick their thumbnails or not, in that a malicious sender can send normal text messages around the attachment claiming it to be legitimate. We could rely on (future) reputation/abuse management to punish users who phish (be it with bogus metadata or bogus descriptions). Bogus metadata is particularly bad though, especially if it's avoidable.
As a first cut, let's do #2 and have the receiver hit the API to calculate its own previews (as it does currently for image thumbnails). We can then extend/optimise this to option 4 as a special extra if needed.
API
---
GET /_matrix/media/r0/preview_url?url=http://wherever.com
200 OK
{
"og:type" : "article"
"og:url" : "https://twitter.com/matrixdotorg/status/684074366691356672"
"og:title" : "Matrix on Twitter"
"og:image" : "https://pbs.twimg.com/profile_images/500400952029888512/yI0qtFi7_400x400.png"
"og:description" : "“Synapse 0.12 is out! Lots of polishing, performance &amp;amp; bugfixes: /sync API, /r0 prefix, fulltext search, 3PID invites https://t.co/5alhXLLEGP”"
"og:site_name" : "Twitter"
}
* Downloads the URL
* If HTML, just stores it in RAM and parses it for OG meta tags
* Download any media OG meta tags to the media repo, and refer to them in the OG via mxc:// URIs.
* If a media filetype we know we can thumbnail: store it on disk, and hand it to the thumbnailer. Generate OG meta tags from the thumbnailer contents.
* Otherwise, don't bother downloading further.

View File

@@ -1,98 +0,0 @@
Scaling synapse via workers
---------------------------
Synapse has experimental support for splitting out functionality into
multiple separate python processes, helping greatly with scalability. These
processes are called 'workers', and are (eventually) intended to scale
horizontally independently.
All processes continue to share the same database instance, and as such, workers
only work with postgres based synapse deployments (sharing a single sqlite
across multiple processes is a recipe for disaster, plus you should be using
postgres anyway if you care about scalability).
The workers communicate with the master synapse process via a synapse-specific
HTTP protocol called 'replication' - analogous to MySQL or Postgres style
database replication; feeding a stream of relevant data to the workers so they
can be kept in sync with the main synapse process and database state.
To enable workers, you need to add a replication listener to the master synapse, e.g.::
listeners:
- port: 9092
bind_address: '127.0.0.1'
type: http
tls: false
x_forwarded: false
resources:
- names: [replication]
compress: false
Under **no circumstances** should this replication API listener be exposed to the
public internet; it currently implements no authentication whatsoever and is
unencrypted HTTP.
You then create a set of configs for the various worker processes. These should be
worker configuration files should be stored in a dedicated subdirectory, to allow
synctl to manipulate them.
The current available worker applications are:
* synapse.app.pusher - handles sending push notifications to sygnal and email
* synapse.app.synchrotron - handles /sync endpoints. can scales horizontally through multiple instances.
* synapse.app.appservice - handles output traffic to Application Services
* synapse.app.federation_reader - handles receiving federation traffic (including public_rooms API)
* synapse.app.media_repository - handles the media repository.
* synapse.app.client_reader - handles client API endpoints like /publicRooms
Each worker configuration file inherits the configuration of the main homeserver
configuration file. You can then override configuration specific to that worker,
e.g. the HTTP listener that it provides (if any); logging configuration; etc.
You should minimise the number of overrides though to maintain a usable config.
You must specify the type of worker application (worker_app) and the replication
endpoint that it's talking to on the main synapse process (worker_replication_url).
For instance::
worker_app: synapse.app.synchrotron
# The replication listener on the synapse to talk to.
worker_replication_url: http://127.0.0.1:9092/_synapse/replication
worker_listeners:
- type: http
port: 8083
resources:
- names:
- client
worker_daemonize: True
worker_pid_file: /home/matrix/synapse/synchrotron.pid
worker_log_config: /home/matrix/synapse/config/synchrotron_log_config.yaml
...is a full configuration for a synchrotron worker instance, which will expose a
plain HTTP /sync endpoint on port 8083 separately from the /sync endpoint provided
by the main synapse.
Obviously you should configure your loadbalancer to route the /sync endpoint to
the synchrotron instance(s) in this instance.
Finally, to actually run your worker-based synapse, you must pass synctl the -a
commandline option to tell it to operate on all the worker configurations found
in the given directory, e.g.::
synctl -a $CONFIG/workers start
Currently one should always restart all workers when restarting or upgrading
synapse, unless you explicitly know it's safe not to. For instance, restarting
synapse without restarting all the synchrotrons may result in broken typing
notifications.
To manipulate a specific worker, you pass the -w option to synctl::
synctl -w $CONFIG/workers/synchrotron.yaml restart
All of the above is highly experimental and subject to change as Synapse evolves,
but documenting it here to help folks needing highly scalable Synapses similar
to the one running matrix.org!

View File

@@ -1,25 +0,0 @@
#!/bin/bash
set -eux
: ${WORKSPACE:="$(pwd)"}
export WORKSPACE
export PYTHONDONTWRITEBYTECODE=yep
export SYNAPSE_CACHE_FACTOR=1
./jenkins/prepare_synapse.sh
./jenkins/clone.sh sytest https://github.com/matrix-org/sytest.git
./jenkins/clone.sh dendron https://github.com/matrix-org/dendron.git
./dendron/jenkins/build_dendron.sh
./sytest/jenkins/prep_sytest_for_postgres.sh
./sytest/jenkins/install_and_run.sh \
--synapse-directory $WORKSPACE \
--dendron $WORKSPACE/dendron/bin/dendron \
--pusher \
--synchrotron \
--federation-reader \
--client-reader \
--appservice \
--federation-sender \

View File

@@ -4,14 +4,58 @@ set -eux
: ${WORKSPACE:="$(pwd)"}
export WORKSPACE
export PYTHONDONTWRITEBYTECODE=yep
export SYNAPSE_CACHE_FACTOR=1
./jenkins/prepare_synapse.sh
./jenkins/clone.sh sytest https://github.com/matrix-org/sytest.git
# Output test results as junit xml
export TRIAL_FLAGS="--reporter=subunit"
export TOXSUFFIX="| subunit-1to2 | subunit2junitxml --no-passthrough --output-to=results.xml"
# Write coverage reports to a separate file for each process
export COVERAGE_OPTS="-p"
export DUMP_COVERAGE_COMMAND="coverage help"
./sytest/jenkins/prep_sytest_for_postgres.sh
# Output flake8 violations to violations.flake8.log
# Don't exit with non-0 status code on Jenkins,
# so that the build steps continue and a later step can decided whether to
# UNSTABLE or FAILURE this build.
export PEP8SUFFIX="--output-file=violations.flake8.log || echo flake8 finished with status code \$?"
./sytest/jenkins/install_and_run.sh \
--synapse-directory $WORKSPACE \
rm .coverage* || echo "No coverage files to remove"
tox --notest -e py27
TOX_BIN=$WORKSPACE/.tox/py27/bin
$TOX_BIN/pip install psycopg2
: ${GIT_BRANCH:="origin/$(git rev-parse --abbrev-ref HEAD)"}
if [[ ! -e .sytest-base ]]; then
git clone https://github.com/matrix-org/sytest.git .sytest-base --mirror
else
(cd .sytest-base; git fetch -p)
fi
rm -rf sytest
git clone .sytest-base sytest --shared
cd sytest
git checkout "${GIT_BRANCH}" || (echo >&2 "No ref ${GIT_BRANCH} found, falling back to develop" ; git checkout develop)
: ${PORT_BASE:=8000}
./jenkins/prep_sytest_for_postgres.sh
echo >&2 "Running sytest with PostgreSQL";
./jenkins/install_and_run.sh --coverage \
--python $TOX_BIN/python \
--synapse-directory $WORKSPACE \
--port-base $PORT_BASE
cd ..
cp sytest/.coverage.* .
# Combine the coverage reports
echo "Combining:" .coverage.*
$TOX_BIN/python -m coverage combine
# Output coverage to coverage.xml
$TOX_BIN/coverage xml -o coverage.xml

View File

@@ -4,12 +4,52 @@ set -eux
: ${WORKSPACE:="$(pwd)"}
export WORKSPACE
export PYTHONDONTWRITEBYTECODE=yep
export SYNAPSE_CACHE_FACTOR=1
./jenkins/prepare_synapse.sh
./jenkins/clone.sh sytest https://github.com/matrix-org/sytest.git
# Output test results as junit xml
export TRIAL_FLAGS="--reporter=subunit"
export TOXSUFFIX="| subunit-1to2 | subunit2junitxml --no-passthrough --output-to=results.xml"
# Write coverage reports to a separate file for each process
export COVERAGE_OPTS="-p"
export DUMP_COVERAGE_COMMAND="coverage help"
./sytest/jenkins/install_and_run.sh \
--synapse-directory $WORKSPACE \
# Output flake8 violations to violations.flake8.log
# Don't exit with non-0 status code on Jenkins,
# so that the build steps continue and a later step can decided whether to
# UNSTABLE or FAILURE this build.
export PEP8SUFFIX="--output-file=violations.flake8.log || echo flake8 finished with status code \$?"
rm .coverage* || echo "No coverage files to remove"
tox --notest -e py27
TOX_BIN=$WORKSPACE/.tox/py27/bin
: ${GIT_BRANCH:="origin/$(git rev-parse --abbrev-ref HEAD)"}
if [[ ! -e .sytest-base ]]; then
git clone https://github.com/matrix-org/sytest.git .sytest-base --mirror
else
(cd .sytest-base; git fetch -p)
fi
rm -rf sytest
git clone .sytest-base sytest --shared
cd sytest
git checkout "${GIT_BRANCH}" || (echo >&2 "No ref ${GIT_BRANCH} found, falling back to develop" ; git checkout develop)
: ${PORT_BASE:=8500}
./jenkins/install_and_run.sh --coverage \
--python $TOX_BIN/python \
--synapse-directory $WORKSPACE \
--port-base $PORT_BASE
cd ..
cp sytest/.coverage.* .
# Combine the coverage reports
echo "Combining:" .coverage.*
$TOX_BIN/python -m coverage combine
# Output coverage to coverage.xml
$TOX_BIN/coverage xml -o coverage.xml

View File

@@ -22,9 +22,4 @@ export PEP8SUFFIX="--output-file=violations.flake8.log || echo flake8 finished w
rm .coverage* || echo "No coverage files to remove"
tox --notest -e py27
TOX_BIN=$WORKSPACE/.tox/py27/bin
python synapse/python_dependencies.py | xargs -n1 $TOX_BIN/pip install
$TOX_BIN/pip install lxml
tox -e py27

86
jenkins.sh Executable file
View File

@@ -0,0 +1,86 @@
#!/bin/bash
set -eux
: ${WORKSPACE:="$(pwd)"}
export PYTHONDONTWRITEBYTECODE=yep
export SYNAPSE_CACHE_FACTOR=1
# Output test results as junit xml
export TRIAL_FLAGS="--reporter=subunit"
export TOXSUFFIX="| subunit-1to2 | subunit2junitxml --no-passthrough --output-to=results.xml"
# Write coverage reports to a separate file for each process
export COVERAGE_OPTS="-p"
export DUMP_COVERAGE_COMMAND="coverage help"
# Output flake8 violations to violations.flake8.log
# Don't exit with non-0 status code on Jenkins,
# so that the build steps continue and a later step can decided whether to
# UNSTABLE or FAILURE this build.
export PEP8SUFFIX="--output-file=violations.flake8.log || echo flake8 finished with status code \$?"
rm .coverage* || echo "No coverage files to remove"
tox
: ${GIT_BRANCH:="origin/$(git rev-parse --abbrev-ref HEAD)"}
TOX_BIN=$WORKSPACE/.tox/py27/bin
if [[ ! -e .sytest-base ]]; then
git clone https://github.com/matrix-org/sytest.git .sytest-base --mirror
else
(cd .sytest-base; git fetch -p)
fi
rm -rf sytest
git clone .sytest-base sytest --shared
cd sytest
git checkout "${GIT_BRANCH}" || (echo >&2 "No ref ${GIT_BRANCH} found, falling back to develop" ; git checkout develop)
: ${PERL5LIB:=$WORKSPACE/perl5/lib/perl5}
: ${PERL_MB_OPT:=--install_base=$WORKSPACE/perl5}
: ${PERL_MM_OPT:=INSTALL_BASE=$WORKSPACE/perl5}
export PERL5LIB PERL_MB_OPT PERL_MM_OPT
./install-deps.pl
: ${PORT_BASE:=8000}
echo >&2 "Running sytest with SQLite3";
./run-tests.pl --coverage -O tap --synapse-directory $WORKSPACE \
--python $TOX_BIN/python --all --port-base $PORT_BASE > results-sqlite3.tap
RUN_POSTGRES=""
for port in $(($PORT_BASE + 1)) $(($PORT_BASE + 2)); do
if psql synapse_jenkins_$port <<< ""; then
RUN_POSTGRES="$RUN_POSTGRES:$port"
cat > localhost-$port/database.yaml << EOF
name: psycopg2
args:
database: synapse_jenkins_$port
EOF
fi
done
# Run if both postgresql databases exist
if test "$RUN_POSTGRES" = ":$(($PORT_BASE + 1)):$(($PORT_BASE + 2))"; then
echo >&2 "Running sytest with PostgreSQL";
$TOX_BIN/pip install psycopg2
./run-tests.pl --coverage -O tap --synapse-directory $WORKSPACE \
--python $TOX_BIN/python --all --port-base $PORT_BASE > results-postgresql.tap
else
echo >&2 "Skipping running sytest with PostgreSQL, $RUN_POSTGRES"
fi
cd ..
cp sytest/.coverage.* .
# Combine the coverage reports
echo "Combining:" .coverage.*
$TOX_BIN/python -m coverage combine
# Output coverage to coverage.xml
$TOX_BIN/coverage xml -o coverage.xml

View File

@@ -1,44 +0,0 @@
#! /bin/bash
# This clones a project from github into a named subdirectory
# If the project has a branch with the same name as this branch
# then it will checkout that branch after cloning.
# Otherwise it will checkout "origin/develop."
# The first argument is the name of the directory to checkout
# the branch into.
# The second argument is the URL of the remote repository to checkout.
# Usually something like https://github.com/matrix-org/sytest.git
set -eux
NAME=$1
PROJECT=$2
BASE=".$NAME-base"
# Update our mirror.
if [ ! -d ".$NAME-base" ]; then
# Create a local mirror of the source repository.
# This saves us from having to download the entire repository
# when this script is next run.
git clone "$PROJECT" "$BASE" --mirror
else
# Fetch any updates from the source repository.
(cd "$BASE"; git fetch -p)
fi
# Remove the existing repository so that we have a clean copy
rm -rf "$NAME"
# Cloning with --shared means that we will share portions of the
# .git directory with our local mirror.
git clone "$BASE" "$NAME" --shared
# Jenkins may have supplied us with the name of the branch in the
# environment. Otherwise we will have to guess based on the current
# commit.
: ${GIT_BRANCH:="origin/$(git rev-parse --abbrev-ref HEAD)"}
cd "$NAME"
# check out the relevant branch
git checkout "${GIT_BRANCH}" || (
echo >&2 "No ref ${GIT_BRANCH} found, falling back to develop"
git checkout "origin/develop"
)

View File

@@ -1,20 +0,0 @@
#! /bin/bash
cd "`dirname $0`/.."
TOX_DIR=$WORKSPACE/.tox
mkdir -p $TOX_DIR
if ! [ $TOX_DIR -ef .tox ]; then
ln -s "$TOX_DIR" .tox
fi
# set up the virtualenv
tox -e py27 --notest -v
TOX_BIN=$TOX_DIR/py27/bin
$TOX_BIN/pip install setuptools
{ python synapse/python_dependencies.py
echo lxml psycopg2
} | xargs $TOX_BIN/pip install

View File

@@ -1,7 +0,0 @@
.header {
border-bottom: 4px solid #e4f7ed ! important;
}
.notif_link a, .footer a {
color: #76CFA6 ! important;
}

View File

@@ -1,156 +0,0 @@
body {
margin: 0px;
}
pre, code {
word-break: break-word;
white-space: pre-wrap;
}
#page {
font-family: 'Open Sans', Helvetica, Arial, Sans-Serif;
font-color: #454545;
font-size: 12pt;
width: 100%;
padding: 20px;
}
#inner {
width: 640px;
}
.header {
width: 100%;
height: 87px;
color: #454545;
border-bottom: 4px solid #e5e5e5;
}
.logo {
text-align: right;
margin-left: 20px;
}
.salutation {
padding-top: 10px;
font-weight: bold;
}
.summarytext {
}
.room {
width: 100%;
color: #454545;
border-bottom: 1px solid #e5e5e5;
}
.room_header td {
padding-top: 38px;
padding-bottom: 10px;
border-bottom: 1px solid #e5e5e5;
}
.room_name {
vertical-align: middle;
font-size: 18px;
font-weight: bold;
}
.room_header h2 {
margin-top: 0px;
margin-left: 75px;
font-size: 20px;
}
.room_avatar {
width: 56px;
line-height: 0px;
text-align: center;
vertical-align: middle;
}
.room_avatar img {
width: 48px;
height: 48px;
object-fit: cover;
border-radius: 24px;
}
.notif {
border-bottom: 1px solid #e5e5e5;
margin-top: 16px;
padding-bottom: 16px;
}
.historical_message .sender_avatar {
opacity: 0.3;
}
/* spell out opacity and historical_message class names for Outlook aka Word */
.historical_message .sender_name {
color: #e3e3e3;
}
.historical_message .message_time {
color: #e3e3e3;
}
.historical_message .message_body {
color: #c7c7c7;
}
.historical_message td,
.message td {
padding-top: 10px;
}
.sender_avatar {
width: 56px;
text-align: center;
vertical-align: top;
}
.sender_avatar img {
margin-top: -2px;
width: 32px;
height: 32px;
border-radius: 16px;
}
.sender_name {
display: inline;
font-size: 13px;
color: #a2a2a2;
}
.message_time {
text-align: right;
width: 100px;
font-size: 11px;
color: #a2a2a2;
}
.message_body {
}
.notif_link td {
padding-top: 10px;
padding-bottom: 10px;
font-weight: bold;
}
.notif_link a, .footer a {
color: #454545;
text-decoration: none;
}
.debug {
font-size: 10px;
color: #888;
}
.footer {
margin-top: 20px;
text-align: center;
}

View File

@@ -1,45 +0,0 @@
{% for message in notif.messages %}
<tr class="{{ "historical_message" if message.is_historical else "message" }}">
<td class="sender_avatar">
{% if loop.index0 == 0 or notif.messages[loop.index0 - 1].sender_name != notif.messages[loop.index0].sender_name %}
{% if message.sender_avatar_url %}
<img alt="" class="sender_avatar" src="{{ message.sender_avatar_url|mxc_to_http(32,32) }}" />
{% else %}
{% if message.sender_hash % 3 == 0 %}
<img class="sender_avatar" src="https://vector.im/beta/img/76cfa6.png" />
{% elif message.sender_hash % 3 == 1 %}
<img class="sender_avatar" src="https://vector.im/beta/img/50e2c2.png" />
{% else %}
<img class="sender_avatar" src="https://vector.im/beta/img/f4c371.png" />
{% endif %}
{% endif %}
{% endif %}
</td>
<td class="message_contents">
{% if loop.index0 == 0 or notif.messages[loop.index0 - 1].sender_name != notif.messages[loop.index0].sender_name %}
<div class="sender_name">{% if message.msgtype == "m.emote" %}*{% endif %} {{ message.sender_name }}</div>
{% endif %}
<div class="message_body">
{% if message.msgtype == "m.text" %}
{{ message.body_text_html }}
{% elif message.msgtype == "m.emote" %}
{{ message.body_text_html }}
{% elif message.msgtype == "m.notice" %}
{{ message.body_text_html }}
{% elif message.msgtype == "m.image" %}
<img src="{{ message.image_url|mxc_to_http(640, 480, scale) }}" />
{% elif message.msgtype == "m.file" %}
<span class="filename">{{ message.body_text_plain }}</span>
{% endif %}
</div>
</td>
<td class="message_time">{{ message.ts|format_ts("%H:%M") }}</td>
</tr>
{% endfor %}
<tr class="notif_link">
<td></td>
<td>
<a href="{{ notif.link }}">View {{ room.title }}</a>
</td>
<td></td>
</tr>

View File

@@ -1,16 +0,0 @@
{% for message in notif.messages %}
{% if message.msgtype == "m.emote" %}* {% endif %}{{ message.sender_name }} ({{ message.ts|format_ts("%H:%M") }})
{% if message.msgtype == "m.text" %}
{{ message.body_text_plain }}
{% elif message.msgtype == "m.emote" %}
{{ message.body_text_plain }}
{% elif message.msgtype == "m.notice" %}
{{ message.body_text_plain }}
{% elif message.msgtype == "m.image" %}
{{ message.body_text_plain }}
{% elif message.msgtype == "m.file" %}
{{ message.body_text_plain }}
{% endif %}
{% endfor %}
View {{ room.title }} at {{ notif.link }}

View File

@@ -1,55 +0,0 @@
<!doctype html>
<html lang="en">
<head>
<style type="text/css">
{% include 'mail.css' without context %}
{% include "mail-%s.css" % app_name ignore missing without context %}
</style>
</head>
<body>
<table id="page">
<tr>
<td> </td>
<td id="inner">
<table class="header">
<tr>
<td>
<div class="salutation">Hi {{ user_display_name }},</div>
<div class="summarytext">{{ summary_text }}</div>
</td>
<td class="logo">
{% if app_name == "Riot" %}
<img src="http://matrix.org/img/riot-logo-email.png" width="83" height="83" alt="[Riot]"/>
{% elif app_name == "Vector" %}
<img src="http://matrix.org/img/vector-logo-email.png" width="64" height="83" alt="[Vector]"/>
{% else %}
<img src="http://matrix.org/img/matrix-120x51.png" width="120" height="51" alt="[matrix]"/>
{% endif %}
</td>
</tr>
</table>
{% for room in rooms %}
{% include 'room.html' with context %}
{% endfor %}
<div class="footer">
<a href="{{ unsubscribe_link }}">Unsubscribe</a>
<br/>
<br/>
<div class="debug">
Sending email at {{ reason.now|format_ts("%c") }} due to activity in room {{ reason.room_name }} because
an event was received at {{ reason.received_at|format_ts("%c") }}
which is more than {{ "%.1f"|format(reason.delay_before_mail_ms / (60*1000)) }} ({{ reason.delay_before_mail_ms }}) mins ago,
{% if reason.last_sent_ts %}
and the last time we sent a mail for this room was {{ reason.last_sent_ts|format_ts("%c") }},
which is more than {{ "%.1f"|format(reason.throttle_ms / (60*1000)) }} (current throttle_ms) mins ago.
{% else %}
and we don't have a last time we sent a mail for this room.
{% endif %}
</div>
</div>
</td>
<td> </td>
</tr>
</table>
</body>
</html>

View File

@@ -1,10 +0,0 @@
Hi {{ user_display_name }},
{{ summary_text }}
{% for room in rooms %}
{% include 'room.txt' with context %}
{% endfor %}
You can disable these notifications at {{ unsubscribe_link }}

View File

@@ -1,33 +0,0 @@
<table class="room">
<tr class="room_header">
<td class="room_avatar">
{% if room.avatar_url %}
<img alt="" src="{{ room.avatar_url|mxc_to_http(48,48) }}" />
{% else %}
{% if room.hash % 3 == 0 %}
<img alt="" src="https://vector.im/beta/img/76cfa6.png" />
{% elif room.hash % 3 == 1 %}
<img alt="" src="https://vector.im/beta/img/50e2c2.png" />
{% else %}
<img alt="" src="https://vector.im/beta/img/f4c371.png" />
{% endif %}
{% endif %}
</td>
<td class="room_name" colspan="2">
{{ room.title }}
</td>
</tr>
{% if room.invite %}
<tr>
<td></td>
<td>
<a href="{{ room.link }}">Join the conversation.</a>
</td>
<td></td>
</tr>
{% else %}
{% for notif in room.notifs %}
{% include 'notif.html' with context %}
{% endfor %}
{% endif %}
</table>

View File

@@ -1,9 +0,0 @@
{{ room.title }}
{% if room.invite %}
You've been invited, join at {{ room.link }}
{% else %}
{% for notif in room.notifs %}
{% include 'notif.txt' with context %}
{% endfor %}
{% endif %}

View File

@@ -116,19 +116,17 @@ def get_json(origin_name, origin_key, destination, path):
authorization_headers = []
for key, sig in signed_json["signatures"][origin_name].items():
header = "X-Matrix origin=%s,key=\"%s\",sig=\"%s\"" % (
origin_name, key, sig,
)
authorization_headers.append(bytes(header))
sys.stderr.write(header)
sys.stderr.write("\n")
authorization_headers.append(bytes(
"X-Matrix origin=%s,key=\"%s\",sig=\"%s\"" % (
origin_name, key, sig,
)
))
result = requests.get(
lookup(destination, path),
headers={"Authorization": authorization_headers[0]},
verify=False,
)
sys.stderr.write("Status Code: %d\n" % (result.status_code,))
return result.json()
@@ -143,7 +141,6 @@ def main():
)
json.dump(result, sys.stdout)
print ""
if __name__ == "__main__":
main()

View File

@@ -1,16 +1,10 @@
#!/usr/bin/env python
import argparse
import sys
import bcrypt
import getpass
import yaml
bcrypt_rounds=12
password_pepper = ""
def prompt_for_pass():
password = getpass.getpass("Password: ")
@@ -34,22 +28,12 @@ if __name__ == "__main__":
default=None,
help="New password for user. Will prompt if omitted.",
)
parser.add_argument(
"-c", "--config",
type=argparse.FileType('r'),
help="Path to server config file. Used to read in bcrypt_rounds and password_pepper.",
)
args = parser.parse_args()
if "config" in args and args.config:
config = yaml.safe_load(args.config)
bcrypt_rounds = config.get("bcrypt_rounds", bcrypt_rounds)
password_config = config.get("password_config", {})
password_pepper = password_config.get("pepper", password_pepper)
password = args.password
if not password:
password = prompt_for_pass()
print bcrypt.hashpw(password + password_pepper, bcrypt.gensalt(bcrypt_rounds))
print bcrypt.hashpw(password, bcrypt.gensalt(bcrypt_rounds))

View File

@@ -25,26 +25,18 @@ import urllib2
import yaml
def request_registration(user, password, server_location, shared_secret, admin=False):
def request_registration(user, password, server_location, shared_secret):
mac = hmac.new(
key=shared_secret,
msg=user,
digestmod=hashlib.sha1,
)
mac.update(user)
mac.update("\x00")
mac.update(password)
mac.update("\x00")
mac.update("admin" if admin else "notadmin")
mac = mac.hexdigest()
).hexdigest()
data = {
"user": user,
"password": password,
"mac": mac,
"type": "org.matrix.login.shared_secret",
"admin": admin,
}
server_location = server_location.rstrip("/")
@@ -76,7 +68,7 @@ def request_registration(user, password, server_location, shared_secret, admin=F
sys.exit(1)
def register_new_user(user, password, server_location, shared_secret, admin):
def register_new_user(user, password, server_location, shared_secret):
if not user:
try:
default_user = getpass.getuser()
@@ -107,14 +99,7 @@ def register_new_user(user, password, server_location, shared_secret, admin):
print "Passwords do not match"
sys.exit(1)
if not admin:
admin = raw_input("Make admin [no]: ")
if admin in ("y", "yes", "true"):
admin = True
else:
admin = False
request_registration(user, password, server_location, shared_secret, bool(admin))
request_registration(user, password, server_location, shared_secret)
if __name__ == "__main__":
@@ -134,11 +119,6 @@ if __name__ == "__main__":
default=None,
help="New password for user. Will prompt if omitted.",
)
parser.add_argument(
"-a", "--admin",
action="store_true",
help="Register new user as an admin. Will prompt if omitted.",
)
group = parser.add_mutually_exclusive_group(required=True)
group.add_argument(
@@ -171,4 +151,4 @@ if __name__ == "__main__":
else:
secret = args.shared_secret
register_new_user(args.user, args.password, args.server_url, secret, args.admin)
register_new_user(args.user, args.password, args.server_url, secret)

View File

@@ -34,12 +34,11 @@ logger = logging.getLogger("synapse_port_db")
BOOLEAN_COLUMNS = {
"events": ["processed", "outlier", "contains_url"],
"events": ["processed", "outlier"],
"rooms": ["is_public"],
"event_edges": ["is_state"],
"presence_list": ["accepted"],
"presence_stream": ["currently_active"],
"public_room_list_stream": ["visibility"],
}
@@ -72,14 +71,6 @@ APPEND_ONLY_TABLES = [
"event_to_state_groups",
"rejections",
"event_search",
"presence_stream",
"push_rules_stream",
"current_state_resets",
"ex_outlier_stream",
"cache_invalidation_stream",
"public_room_list_stream",
"state_group_edges",
"stream_ordering_to_exterm",
]
@@ -101,12 +92,8 @@ class Store(object):
_simple_select_onecol_txn = SQLBaseStore.__dict__["_simple_select_onecol_txn"]
_simple_select_onecol = SQLBaseStore.__dict__["_simple_select_onecol"]
_simple_select_one = SQLBaseStore.__dict__["_simple_select_one"]
_simple_select_one_txn = SQLBaseStore.__dict__["_simple_select_one_txn"]
_simple_select_one_onecol = SQLBaseStore.__dict__["_simple_select_one_onecol"]
_simple_select_one_onecol_txn = SQLBaseStore.__dict__[
"_simple_select_one_onecol_txn"
]
_simple_select_one_onecol_txn = SQLBaseStore.__dict__["_simple_select_one_onecol_txn"]
_simple_update_one = SQLBaseStore.__dict__["_simple_update_one"]
_simple_update_one_txn = SQLBaseStore.__dict__["_simple_update_one_txn"]
@@ -171,40 +158,31 @@ class Porter(object):
def setup_table(self, table):
if table in APPEND_ONLY_TABLES:
# It's safe to just carry on inserting.
row = yield self.postgres_store._simple_select_one(
next_chunk = yield self.postgres_store._simple_select_one_onecol(
table="port_from_sqlite3",
keyvalues={"table_name": table},
retcols=("forward_rowid", "backward_rowid"),
retcol="rowid",
allow_none=True,
)
total_to_port = None
if row is None:
if next_chunk is None:
if table == "sent_transactions":
forward_chunk, already_ported, total_to_port = (
next_chunk, already_ported, total_to_port = (
yield self._setup_sent_transactions()
)
backward_chunk = 0
else:
yield self.postgres_store._simple_insert(
table="port_from_sqlite3",
values={
"table_name": table,
"forward_rowid": 1,
"backward_rowid": 0,
}
values={"table_name": table, "rowid": 1}
)
forward_chunk = 1
backward_chunk = 0
next_chunk = 1
already_ported = 0
else:
forward_chunk = row["forward_rowid"]
backward_chunk = row["backward_rowid"]
if total_to_port is None:
already_ported, total_to_port = yield self._get_total_count_to_port(
table, forward_chunk, backward_chunk
table, next_chunk
)
else:
def delete_all(txn):
@@ -218,124 +196,32 @@ class Porter(object):
yield self.postgres_store._simple_insert(
table="port_from_sqlite3",
values={
"table_name": table,
"forward_rowid": 1,
"backward_rowid": 0,
}
values={"table_name": table, "rowid": 0}
)
forward_chunk = 1
backward_chunk = 0
next_chunk = 1
already_ported, total_to_port = yield self._get_total_count_to_port(
table, forward_chunk, backward_chunk
table, next_chunk
)
defer.returnValue(
(table, already_ported, total_to_port, forward_chunk, backward_chunk)
)
defer.returnValue((table, already_ported, total_to_port, next_chunk))
@defer.inlineCallbacks
def handle_table(self, table, postgres_size, table_size, forward_chunk,
backward_chunk):
def handle_table(self, table, postgres_size, table_size, next_chunk):
if not table_size:
return
self.progress.add_table(table, postgres_size, table_size)
if table == "event_search":
yield self.handle_search_table(
postgres_size, table_size, forward_chunk, backward_chunk
)
return
forward_select = (
select = (
"SELECT rowid, * FROM %s WHERE rowid >= ? ORDER BY rowid LIMIT ?"
% (table,)
)
backward_select = (
"SELECT rowid, * FROM %s WHERE rowid <= ? ORDER BY rowid LIMIT ?"
% (table,)
)
do_forward = [True]
do_backward = [True]
while True:
def r(txn):
forward_rows = []
backward_rows = []
if do_forward[0]:
txn.execute(forward_select, (forward_chunk, self.batch_size,))
forward_rows = txn.fetchall()
if not forward_rows:
do_forward[0] = False
if do_backward[0]:
txn.execute(backward_select, (backward_chunk, self.batch_size,))
backward_rows = txn.fetchall()
if not backward_rows:
do_backward[0] = False
if forward_rows or backward_rows:
headers = [column[0] for column in txn.description]
else:
headers = None
return headers, forward_rows, backward_rows
headers, frows, brows = yield self.sqlite_store.runInteraction(
"select", r
)
if frows or brows:
if frows:
forward_chunk = max(row[0] for row in frows) + 1
if brows:
backward_chunk = min(row[0] for row in brows) - 1
rows = frows + brows
self._convert_rows(table, headers, rows)
def insert(txn):
self.postgres_store.insert_many_txn(
txn, table, headers[1:], rows
)
self.postgres_store._simple_update_one_txn(
txn,
table="port_from_sqlite3",
keyvalues={"table_name": table},
updatevalues={
"forward_rowid": forward_chunk,
"backward_rowid": backward_chunk,
},
)
yield self.postgres_store.execute(insert)
postgres_size += len(rows)
self.progress.update(table, postgres_size)
else:
return
@defer.inlineCallbacks
def handle_search_table(self, postgres_size, table_size, forward_chunk,
backward_chunk):
select = (
"SELECT es.rowid, es.*, e.origin_server_ts, e.stream_ordering"
" FROM event_search as es"
" INNER JOIN events AS e USING (event_id, room_id)"
" WHERE es.rowid >= ?"
" ORDER BY es.rowid LIMIT ?"
)
while True:
def r(txn):
txn.execute(select, (forward_chunk, self.batch_size,))
txn.execute(select, (next_chunk, self.batch_size,))
rows = txn.fetchall()
headers = [column[0] for column in txn.description]
@@ -344,51 +230,59 @@ class Porter(object):
headers, rows = yield self.sqlite_store.runInteraction("select", r)
if rows:
forward_chunk = rows[-1][0] + 1
next_chunk = rows[-1][0] + 1
# We have to treat event_search differently since it has a
# different structure in the two different databases.
def insert(txn):
sql = (
"INSERT INTO event_search (event_id, room_id, key,"
" sender, vector, origin_server_ts, stream_ordering)"
" VALUES (?,?,?,?,to_tsvector('english', ?),?,?)"
)
rows_dict = [
dict(zip(headers, row))
for row in rows
]
txn.executemany(sql, [
(
row["event_id"],
row["room_id"],
row["key"],
row["sender"],
row["value"],
row["origin_server_ts"],
row["stream_ordering"],
if table == "event_search":
# We have to treat event_search differently since it has a
# different structure in the two different databases.
def insert(txn):
sql = (
"INSERT INTO event_search (event_id, room_id, key, sender, vector)"
" VALUES (?,?,?,?,to_tsvector('english', ?))"
)
for row in rows_dict
])
self.postgres_store._simple_update_one_txn(
txn,
table="port_from_sqlite3",
keyvalues={"table_name": "event_search"},
updatevalues={
"forward_rowid": forward_chunk,
"backward_rowid": backward_chunk,
},
)
rows_dict = [
dict(zip(headers, row))
for row in rows
]
txn.executemany(sql, [
(
row["event_id"],
row["room_id"],
row["key"],
row["sender"],
row["value"],
)
for row in rows_dict
])
self.postgres_store._simple_update_one_txn(
txn,
table="port_from_sqlite3",
keyvalues={"table_name": table},
updatevalues={"rowid": next_chunk},
)
else:
self._convert_rows(table, headers, rows)
def insert(txn):
self.postgres_store.insert_many_txn(
txn, table, headers[1:], rows
)
self.postgres_store._simple_update_one_txn(
txn,
table="port_from_sqlite3",
keyvalues={"table_name": table},
updatevalues={"rowid": next_chunk},
)
yield self.postgres_store.execute(insert)
postgres_size += len(rows)
self.progress.update("event_search", postgres_size)
self.progress.update(table, postgres_size)
else:
return
@@ -462,32 +356,10 @@ class Porter(object):
txn.execute(
"CREATE TABLE port_from_sqlite3 ("
" table_name varchar(100) NOT NULL UNIQUE,"
" forward_rowid bigint NOT NULL,"
" backward_rowid bigint NOT NULL"
" rowid bigint NOT NULL"
")"
)
# The old port script created a table with just a "rowid" column.
# We want people to be able to rerun this script from an old port
# so that they can pick up any missing events that were not
# ported across.
def alter_table(txn):
txn.execute(
"ALTER TABLE IF EXISTS port_from_sqlite3"
" RENAME rowid TO forward_rowid"
)
txn.execute(
"ALTER TABLE IF EXISTS port_from_sqlite3"
" ADD backward_rowid bigint NOT NULL DEFAULT 0"
)
try:
yield self.postgres_store.runInteraction(
"alter_table", alter_table
)
except Exception as e:
logger.info("Failed to create port table: %s", e)
try:
yield self.postgres_store.runInteraction(
"create_port_table", create_port_table
@@ -547,7 +419,7 @@ class Porter(object):
@defer.inlineCallbacks
def _setup_sent_transactions(self):
# Only save things from the last day
yesterday = int(time.time() * 1000) - 86400000
yesterday = int(time.time()*1000) - 86400000
# And save the max transaction id from each destination
select = (
@@ -603,11 +475,7 @@ class Porter(object):
yield self.postgres_store._simple_insert(
table="port_from_sqlite3",
values={
"table_name": "sent_transactions",
"forward_rowid": next_chunk,
"backward_rowid": 0,
}
values={"table_name": "sent_transactions", "rowid": next_chunk}
)
def get_sent_table_size(txn):
@@ -628,18 +496,13 @@ class Porter(object):
defer.returnValue((next_chunk, inserted_rows, total_count))
@defer.inlineCallbacks
def _get_remaining_count_to_port(self, table, forward_chunk, backward_chunk):
frows = yield self.sqlite_store.execute_sql(
def _get_remaining_count_to_port(self, table, next_chunk):
rows = yield self.sqlite_store.execute_sql(
"SELECT count(*) FROM %s WHERE rowid >= ?" % (table,),
forward_chunk,
next_chunk,
)
brows = yield self.sqlite_store.execute_sql(
"SELECT count(*) FROM %s WHERE rowid <= ?" % (table,),
backward_chunk,
)
defer.returnValue(frows[0][0] + brows[0][0])
defer.returnValue(rows[0][0])
@defer.inlineCallbacks
def _get_already_ported_count(self, table):
@@ -650,10 +513,10 @@ class Porter(object):
defer.returnValue(rows[0][0])
@defer.inlineCallbacks
def _get_total_count_to_port(self, table, forward_chunk, backward_chunk):
def _get_total_count_to_port(self, table, next_chunk):
remaining, done = yield defer.gatherResults(
[
self._get_remaining_count_to_port(table, forward_chunk, backward_chunk),
self._get_remaining_count_to_port(table, next_chunk),
self._get_already_ported_count(table),
],
consumeErrors=True,
@@ -784,7 +647,7 @@ class CursesProgress(Progress):
color = curses.color_pair(2) if perc == 100 else curses.color_pair(1)
self.stdscr.addstr(
i + 2, left_margin + max_len - len(table),
i+2, left_margin + max_len - len(table),
table,
curses.A_BOLD | color,
)
@@ -792,18 +655,18 @@ class CursesProgress(Progress):
size = 20
progress = "[%s%s]" % (
"#" * int(perc * size / 100),
" " * (size - int(perc * size / 100)),
"#" * int(perc*size/100),
" " * (size - int(perc*size/100)),
)
self.stdscr.addstr(
i + 2, left_margin + max_len + middle_space,
i+2, left_margin + max_len + middle_space,
"%s %3d%% (%d/%d)" % (progress, perc, data["num_done"], data["total"]),
)
if self.finished:
self.stdscr.addstr(
rows - 1, 0,
rows-1, 0,
"Press any key to exit...",
)

View File

@@ -16,5 +16,7 @@ ignore =
[flake8]
max-line-length = 90
# W503 requires that binary operators be at the end, not start, of lines. Erik doesn't like it.
ignore = W503
ignore = W503 ; W503 requires that binary operators be at the end, not start, of lines. Erik doesn't like it.
[pep8]
max-line-length = 90

View File

@@ -23,45 +23,6 @@ import sys
here = os.path.abspath(os.path.dirname(__file__))
# Some notes on `setup.py test`:
#
# Once upon a time we used to try to make `setup.py test` run `tox` to run the
# tests. That's a bad idea for three reasons:
#
# 1: `setup.py test` is supposed to find out whether the tests work in the
# *current* environmentt, not whatever tox sets up.
# 2: Empirically, trying to install tox during the test run wasn't working ("No
# module named virtualenv").
# 3: The tox documentation advises against it[1].
#
# Even further back in time, we used to use setuptools_trial [2]. That has its
# own set of issues: for instance, it requires installation of Twisted to build
# an sdist (because the recommended mode of usage is to add it to
# `setup_requires`). That in turn means that in order to successfully run tox
# you have to have the python header files installed for whichever version of
# python tox uses (which is python3 on recent ubuntus, for example).
#
# So, for now at least, we stick with what appears to be the convention among
# Twisted projects, and don't attempt to do anything when someone runs
# `setup.py test`; instead we direct people to run `trial` directly if they
# care.
#
# [1]: http://tox.readthedocs.io/en/2.5.0/example/basic.html#integration-with-setup-py-test-command
# [2]: https://pypi.python.org/pypi/setuptools_trial
class TestCommand(Command):
user_options = []
def initialize_options(self):
pass
def finalize_options(self):
pass
def run(self):
print ("""Synapse's tests cannot be run via setup.py. To run them, try:
PYTHONPATH="." trial tests
""")
def read_file(path_segments):
"""Read a file from the package. Takes a list of strings to join to
make the path"""
@@ -78,6 +39,38 @@ def exec_file(path_segments):
return result
class Tox(Command):
user_options = [('tox-args=', 'a', "Arguments to pass to tox")]
def initialize_options(self):
self.tox_args = None
def finalize_options(self):
self.test_args = []
self.test_suite = True
def run(self):
#import here, cause outside the eggs aren't loaded
try:
import tox
except ImportError:
try:
self.distribution.fetch_build_eggs("tox")
import tox
except:
raise RuntimeError(
"The tests need 'tox' to run. Please install 'tox'."
)
import shlex
args = self.tox_args
if args:
args = shlex.split(self.tox_args)
else:
args = []
errno = tox.cmdline(args=args)
sys.exit(errno)
version = exec_file(("synapse", "__init__.py"))["__version__"]
dependencies = exec_file(("synapse", "python_dependencies.py"))
long_description = read_file(("README.rst",))
@@ -93,5 +86,5 @@ setup(
zip_safe=False,
long_description=long_description,
scripts=["synctl"] + glob.glob("scripts/*"),
cmdclass={'test': TestCommand},
cmdclass={'test': Tox},
)

View File

@@ -16,4 +16,4 @@
""" This is a reference implementation of a Matrix home server.
"""
__version__ = "0.18.6-rc2"
__version__ = "0.14.0"

View File

@@ -13,22 +13,22 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import logging
import pymacaroons
"""This module contains classes for authenticating the user."""
from canonicaljson import encode_canonical_json
from signedjson.key import decode_verify_key_bytes
from signedjson.sign import verify_signed_json, SignatureVerifyException
from twisted.internet import defer
from unpaddedbase64 import decode_base64
import synapse.types
from twisted.internet import defer
from synapse.api.constants import EventTypes, Membership, JoinRules
from synapse.api.errors import AuthError, Codes, SynapseError, EventSizeError
from synapse.types import UserID, get_domain_from_id
from synapse.util.logcontext import preserve_context_over_fn
from synapse.types import Requester, RoomID, UserID, EventID
from synapse.util.logutils import log_function
from synapse.util.metrics import Measure
from synapse.util.logcontext import preserve_context_over_fn
from unpaddedbase64 import decode_base64
import logging
import pymacaroons
logger = logging.getLogger(__name__)
@@ -39,34 +39,23 @@ AuthEventTypes = (
EventTypes.ThirdPartyInvite,
)
# guests always get this device id.
GUEST_DEVICE_ID = "guest_device"
class Auth(object):
"""
FIXME: This class contains a mix of functions for authenticating users
of our client-server API and authenticating events added to room graphs.
"""
def __init__(self, hs):
self.hs = hs
self.clock = hs.get_clock()
self.store = hs.get_datastore()
self.state = hs.get_state_handler()
self.TOKEN_NOT_FOUND_HTTP_STATUS = 401
self._KNOWN_CAVEAT_PREFIXES = set([
"gen = ",
"guest = ",
"type = ",
"time < ",
"user_id = ",
])
@defer.inlineCallbacks
def check_from_context(self, event, context, do_sig_check=True):
auth_events_ids = yield self.compute_auth_events(
event, context.prev_state_ids, for_verification=True,
)
auth_events = yield self.store.get_events(auth_events_ids)
auth_events = {
(e.type, e.state_key): e for e in auth_events.values()
}
self.check(event, auth_events=auth_events, do_sig_check=do_sig_check)
def check(self, event, auth_events, do_sig_check=True):
def check(self, event, auth_events):
""" Checks if this event is correctly authed.
Args:
@@ -77,35 +66,11 @@ class Auth(object):
Returns:
True if the auth checks pass.
"""
with Measure(self.clock, "auth.check"):
self.check_size_limits(event)
self.check_size_limits(event)
try:
if not hasattr(event, "room_id"):
raise AuthError(500, "Event has no room_id: %s" % event)
if do_sig_check:
sender_domain = get_domain_from_id(event.sender)
event_id_domain = get_domain_from_id(event.event_id)
is_invite_via_3pid = (
event.type == EventTypes.Member
and event.membership == Membership.INVITE
and "third_party_invite" in event.content
)
# Check the sender's domain has signed the event
if not event.signatures.get(sender_domain):
# We allow invites via 3pid to have a sender from a different
# HS, as the sender must match the sender of the original
# 3pid invite. This is checked further down with the
# other dedicated membership checks.
if not is_invite_via_3pid:
raise AuthError(403, "Event not signed by sender's server")
# Check the event_id's domain has signed the event
if not event.signatures.get(event_id_domain):
raise AuthError(403, "Event not signed by sending server")
if auth_events is None:
# Oh, we don't know what the state of the room was, so we
# are trusting that this is allowed (at least for now)
@@ -113,12 +78,6 @@ class Auth(object):
return True
if event.type == EventTypes.Create:
room_id_domain = get_domain_from_id(event.room_id)
if room_id_domain != sender_domain:
raise AuthError(
403,
"Creation event's room_id domain does not match sender's"
)
# FIXME
return True
@@ -130,8 +89,8 @@ class Auth(object):
"Room %r does not exist" % (event.room_id,)
)
creating_domain = get_domain_from_id(event.room_id)
originating_domain = get_domain_from_id(event.sender)
creating_domain = RoomID.from_string(event.room_id).domain
originating_domain = UserID.from_string(event.sender).domain
if creating_domain != originating_domain:
if not self.can_federate(event, auth_events):
raise AuthError(
@@ -141,22 +100,6 @@ class Auth(object):
# FIXME: Temp hack
if event.type == EventTypes.Aliases:
if not event.is_state():
raise AuthError(
403,
"Alias event must be a state event",
)
if not event.state_key:
raise AuthError(
403,
"Alias event must have non-empty state_key"
)
sender_domain = get_domain_from_id(event.sender)
if event.state_key != sender_domain:
raise AuthError(
403,
"Alias event's state_key does not match sender's domain"
)
return True
logger.debug(
@@ -175,24 +118,6 @@ class Auth(object):
return allowed
self.check_event_sender_in_room(event, auth_events)
# Special case to allow m.room.third_party_invite events wherever
# a user is allowed to issue invites. Fixes
# https://github.com/vector-im/vector-web/issues/1208 hopefully
if event.type == EventTypes.ThirdPartyInvite:
user_level = self._get_user_power_level(event.user_id, auth_events)
invite_level = self._get_named_level(auth_events, "invite", 0)
if user_level < invite_level:
raise AuthError(
403, (
"You cannot issue a third party invite for %s." %
(event.content.display_name,)
)
)
else:
return True
self._can_send_event(event, auth_events)
if event.type == EventTypes.PowerLevels:
@@ -202,6 +127,13 @@ class Auth(object):
self.check_redaction(event, auth_events)
logger.debug("Allowing! %s", event)
except AuthError as e:
logger.info(
"Event auth check failed on event %s with msg: %s",
event, e.msg
)
logger.info("Denying! %s", event)
raise
def check_size_limits(self, event):
def too_big(field):
@@ -287,17 +219,21 @@ class Auth(object):
@defer.inlineCallbacks
def check_host_in_room(self, room_id, host):
with Measure(self.clock, "check_host_in_room"):
latest_event_ids = yield self.store.get_latest_event_ids_in_room(room_id)
curr_state = yield self.state.get_current_state(room_id)
entry = yield self.state.resolve_state_groups(
room_id, latest_event_ids
)
for event in curr_state.values():
if event.type == EventTypes.Member:
try:
if UserID.from_string(event.state_key).domain != host:
continue
except:
logger.warn("state_key not user_id: %s", event.state_key)
continue
ret = yield self.store.is_host_joined(
room_id, host, entry.state_group, entry.state
)
defer.returnValue(ret)
if event.content["membership"] == Membership.JOIN:
defer.returnValue(True)
defer.returnValue(False)
def check_event_sender_in_room(self, event, auth_events):
key = (EventTypes.Member, event.user_id, )
@@ -335,8 +271,8 @@ class Auth(object):
target_user_id = event.state_key
creating_domain = get_domain_from_id(event.room_id)
target_domain = get_domain_from_id(target_user_id)
creating_domain = RoomID.from_string(event.room_id).domain
target_domain = UserID.from_string(target_user_id).domain
if creating_domain != target_domain:
if not self.can_federate(event, auth_events):
raise AuthError(
@@ -392,10 +328,6 @@ class Auth(object):
if Membership.INVITE == membership and "third_party_invite" in event.content:
if not self._verify_third_party_invite(event, auth_events):
raise AuthError(403, "You are not invited to this room.")
if target_banned:
raise AuthError(
403, "%s is banned from the room" % (target_user_id,)
)
return True
if Membership.JOIN != membership:
@@ -500,9 +432,6 @@ class Auth(object):
if not invite_event:
return False
if invite_event.sender != event.sender:
return False
if event.user_id != invite_event.user_id:
return False
@@ -583,38 +512,33 @@ class Auth(object):
return default
@defer.inlineCallbacks
def get_user_by_req(self, request, allow_guest=False, rights="access"):
def get_user_by_req(self, request, allow_guest=False):
""" Get a registered user's ID.
Args:
request - An HTTP request with an access_token query parameter.
Returns:
defer.Deferred: resolves to a ``synapse.types.Requester`` object
tuple of:
UserID (str)
Access token ID (str)
Raises:
AuthError if no user by that token exists or the token is invalid.
"""
# Can optionally look elsewhere in the request (e.g. headers)
try:
user_id, app_service = yield self._get_appservice_user_id(request)
user_id = yield self._get_appservice_user_id(request.args)
if user_id:
request.authenticated_entity = user_id
defer.returnValue(
synapse.types.create_requester(user_id, app_service=app_service)
Requester(UserID.from_string(user_id), "", False)
)
access_token = get_access_token_from_request(
request, self.TOKEN_NOT_FOUND_HTTP_STATUS
)
user_info = yield self.get_user_by_access_token(access_token, rights)
access_token = request.args["access_token"][0]
user_info = yield self.get_user_by_access_token(access_token)
user = user_info["user"]
token_id = user_info["token_id"]
is_guest = user_info["is_guest"]
# device_id may not be present if get_user_by_access_token has been
# stubbed out.
device_id = user_info.get("device_id")
ip_addr = self.hs.get_ip_from_request(request)
user_agent = request.requestHeaders.getRawHeaders(
"User-Agent",
@@ -626,8 +550,7 @@ class Auth(object):
user=user,
access_token=access_token,
ip=ip_addr,
user_agent=user_agent,
device_id=device_id,
user_agent=user_agent
)
if is_guest and not allow_guest:
@@ -637,9 +560,7 @@ class Auth(object):
request.authenticated_entity = user.to_string()
defer.returnValue(synapse.types.create_requester(
user, token_id, is_guest, device_id, app_service=app_service)
)
defer.returnValue(Requester(user, token_id, is_guest))
except KeyError:
raise AuthError(
self.TOKEN_NOT_FOUND_HTTP_STATUS, "Missing access token.",
@@ -647,21 +568,19 @@ class Auth(object):
)
@defer.inlineCallbacks
def _get_appservice_user_id(self, request):
app_service = self.store.get_app_service_by_token(
get_access_token_from_request(
request, self.TOKEN_NOT_FOUND_HTTP_STATUS
)
def _get_appservice_user_id(self, request_args):
app_service = yield self.store.get_app_service_by_token(
request_args["access_token"][0]
)
if app_service is None:
defer.returnValue((None, None))
defer.returnValue(None)
if "user_id" not in request.args:
defer.returnValue((app_service.sender, app_service))
if "user_id" not in request_args:
defer.returnValue(app_service.sender)
user_id = request.args["user_id"][0]
user_id = request_args["user_id"][0]
if app_service.sender == user_id:
defer.returnValue((app_service.sender, app_service))
defer.returnValue(app_service.sender)
if not app_service.is_interested_in_user(user_id):
raise AuthError(
@@ -673,98 +592,61 @@ class Auth(object):
403,
"Application service has not registered this user"
)
defer.returnValue((user_id, app_service))
defer.returnValue(user_id)
@defer.inlineCallbacks
def get_user_by_access_token(self, token, rights="access"):
""" Validate access token and get user_id from it
def get_user_by_access_token(self, token):
""" Get a registered user's ID.
Args:
token (str): The access token to get the user by.
rights (str): The operation being performed; the access token must
allow this.
Returns:
dict : dict that includes the user and the ID of their access token.
Raises:
AuthError if no user by that token exists or the token is invalid.
"""
try:
macaroon = pymacaroons.Macaroon.deserialize(token)
except Exception: # deserialize can throw more-or-less anything
# doesn't look like a macaroon: treat it as an opaque token which
# must be in the database.
# TODO: it would be nice to get rid of this, but apparently some
# people use access tokens which aren't macaroons
r = yield self._look_up_user_by_access_token(token)
defer.returnValue(r)
ret = yield self.get_user_from_macaroon(token)
except AuthError:
# TODO(daniel): Remove this fallback when all existing access tokens
# have been re-issued as macaroons.
ret = yield self._look_up_user_by_access_token(token)
defer.returnValue(ret)
@defer.inlineCallbacks
def get_user_from_macaroon(self, macaroon_str):
try:
user_id = self.get_user_id_from_macaroon(macaroon)
user = UserID.from_string(user_id)
self.validate_macaroon(
macaroon, rights, self.hs.config.expire_access_token,
user_id=user_id,
)
macaroon = pymacaroons.Macaroon.deserialize(macaroon_str)
self.validate_macaroon(macaroon, "access", False)
user_prefix = "user_id = "
user = None
guest = False
for caveat in macaroon.caveats:
if caveat.caveat_id == "guest = true":
if caveat.caveat_id.startswith(user_prefix):
user = UserID.from_string(caveat.caveat_id[len(user_prefix):])
elif caveat.caveat_id == "guest = true":
guest = True
if user is None:
raise AuthError(
self.TOKEN_NOT_FOUND_HTTP_STATUS, "No user caveat in macaroon",
errcode=Codes.UNKNOWN_TOKEN
)
if guest:
# Guest access tokens are not stored in the database (there can
# only be one access token per guest, anyway).
#
# In order to prevent guest access tokens being used as regular
# user access tokens (and hence getting around the invalidation
# process), we look up the user id and check that it is indeed
# a guest user.
#
# It would of course be much easier to store guest access
# tokens in the database as well, but that would break existing
# guest tokens.
stored_user = yield self.store.get_user_by_id(user_id)
if not stored_user:
raise AuthError(
self.TOKEN_NOT_FOUND_HTTP_STATUS,
"Unknown user_id %s" % user_id,
errcode=Codes.UNKNOWN_TOKEN
)
if not stored_user["is_guest"]:
raise AuthError(
self.TOKEN_NOT_FOUND_HTTP_STATUS,
"Guest access token used for regular user",
errcode=Codes.UNKNOWN_TOKEN
)
ret = {
"user": user,
"is_guest": True,
"token_id": None,
# all guests get the same device id
"device_id": GUEST_DEVICE_ID,
}
elif rights == "delete_pusher":
# We don't store these tokens in the database
ret = {
"user": user,
"is_guest": False,
"token_id": None,
"device_id": None,
}
else:
# This codepath exists for several reasons:
# * so that we can actually return a token ID, which is used
# in some parts of the schema (where we probably ought to
# use device IDs instead)
# * the only way we currently have to invalidate an
# access_token is by removing it from the database, so we
# have to check here that it is still in the db
# * some attributes (notably device_id) aren't stored in the
# macaroon. They probably should be.
# TODO: build the dictionary from the macaroon once the
# above are fixed
ret = yield self._look_up_user_by_access_token(token)
# This codepath exists so that we can actually return a
# token ID, because we use token IDs in place of device
# identifiers throughout the codebase.
# TODO(daniel): Remove this fallback when device IDs are
# properly implemented.
ret = yield self._look_up_user_by_access_token(macaroon_str)
if ret["user"] != user:
logger.error(
"Macaroon user (%s) != DB user (%s)",
@@ -783,67 +665,31 @@ class Auth(object):
errcode=Codes.UNKNOWN_TOKEN
)
def get_user_id_from_macaroon(self, macaroon):
"""Retrieve the user_id given by the caveats on the macaroon.
Does *not* validate the macaroon.
Args:
macaroon (pymacaroons.Macaroon): The macaroon to validate
Returns:
(str) user id
Raises:
AuthError if there is no user_id caveat in the macaroon
"""
user_prefix = "user_id = "
for caveat in macaroon.caveats:
if caveat.caveat_id.startswith(user_prefix):
return caveat.caveat_id[len(user_prefix):]
raise AuthError(
self.TOKEN_NOT_FOUND_HTTP_STATUS, "No user caveat in macaroon",
errcode=Codes.UNKNOWN_TOKEN
)
def validate_macaroon(self, macaroon, type_string, verify_expiry, user_id):
def validate_macaroon(self, macaroon, type_string, verify_expiry):
"""
validate that a Macaroon is understood by and was signed by this server.
Args:
macaroon(pymacaroons.Macaroon): The macaroon to validate
type_string(str): The kind of token required (e.g. "access",
"delete_pusher")
type_string(str): The kind of token this is (e.g. "access", "refresh")
verify_expiry(bool): Whether to verify whether the macaroon has expired.
user_id (str): The user_id required
This should really always be True, but no clients currently implement
token refresh, so we can't enforce expiry yet.
"""
v = pymacaroons.Verifier()
# the verifier runs a test for every caveat on the macaroon, to check
# that it is met for the current request. Each caveat must match at
# least one of the predicates specified by satisfy_exact or
# specify_general.
v.satisfy_exact("gen = 1")
v.satisfy_exact("type = " + type_string)
v.satisfy_exact("user_id = %s" % user_id)
v.satisfy_general(lambda c: c.startswith("user_id = "))
v.satisfy_exact("guest = true")
# verify_expiry should really always be True, but there exist access
# tokens in the wild which expire when they should not, so we can't
# enforce expiry yet (so we have to allow any caveat starting with
# 'time < ' in access tokens).
#
# On the other hand, short-term login tokens (as used by CAS login, for
# example) have an expiry time which we do want to enforce.
if verify_expiry:
v.satisfy_general(self._verify_expiry)
else:
v.satisfy_general(lambda c: c.startswith("time < "))
# access_tokens include a nonce for uniqueness: any value is acceptable
v.satisfy_general(lambda c: c.startswith("nonce = "))
v.verify(macaroon, self.hs.config.macaroon_secret_key)
v = pymacaroons.Verifier()
v.satisfy_general(self._verify_recognizes_caveats)
v.verify(macaroon, self.hs.config.macaroon_secret_key)
def _verify_expiry(self, caveat):
@@ -854,6 +700,15 @@ class Auth(object):
now = self.hs.get_clock().time_msec()
return now < expiry
def _verify_recognizes_caveats(self, caveat):
first_space = caveat.find(" ")
if first_space < 0:
return False
second_space = caveat.find(" ", first_space + 1)
if second_space < 0:
return False
return caveat[:second_space + 1] in self._KNOWN_CAVEAT_PREFIXES
@defer.inlineCallbacks
def _look_up_user_by_access_token(self, token):
ret = yield self.store.get_user_by_access_token(token)
@@ -863,23 +718,18 @@ class Auth(object):
self.TOKEN_NOT_FOUND_HTTP_STATUS, "Unrecognised access token.",
errcode=Codes.UNKNOWN_TOKEN
)
# we use ret.get() below because *lots* of unit tests stub out
# get_user_by_access_token in a way where it only returns a couple of
# the fields.
user_info = {
"user": UserID.from_string(ret.get("name")),
"token_id": ret.get("token_id", None),
"is_guest": False,
"device_id": ret.get("device_id"),
}
defer.returnValue(user_info)
@defer.inlineCallbacks
def get_appservice_by_req(self, request):
try:
token = get_access_token_from_request(
request, self.TOKEN_NOT_FOUND_HTTP_STATUS
)
service = self.store.get_app_service_by_token(token)
token = request.args["access_token"][0]
service = yield self.store.get_app_service_by_token(token)
if not service:
logger.warn("Unrecognised appservice access token: %s" % (token,))
raise AuthError(
@@ -888,7 +738,7 @@ class Auth(object):
errcode=Codes.UNKNOWN_TOKEN
)
request.authenticated_entity = service.sender
return defer.succeed(service)
defer.returnValue(service)
except KeyError:
raise AuthError(
self.TOKEN_NOT_FOUND_HTTP_STATUS, "Missing access token."
@@ -899,7 +749,7 @@ class Auth(object):
@defer.inlineCallbacks
def add_auth_events(self, builder, context):
auth_ids = yield self.compute_auth_events(builder, context.prev_state_ids)
auth_ids = self.compute_auth_events(builder, context.current_state)
auth_events_entries = yield self.store.add_event_hashes(
auth_ids
@@ -907,32 +757,30 @@ class Auth(object):
builder.auth_events = auth_events_entries
@defer.inlineCallbacks
def compute_auth_events(self, event, current_state_ids, for_verification=False):
def compute_auth_events(self, event, current_state):
if event.type == EventTypes.Create:
defer.returnValue([])
return []
auth_ids = []
key = (EventTypes.PowerLevels, "", )
power_level_event_id = current_state_ids.get(key)
power_level_event = current_state.get(key)
if power_level_event_id:
auth_ids.append(power_level_event_id)
if power_level_event:
auth_ids.append(power_level_event.event_id)
key = (EventTypes.JoinRules, "", )
join_rule_event_id = current_state_ids.get(key)
join_rule_event = current_state.get(key)
key = (EventTypes.Member, event.user_id, )
member_event_id = current_state_ids.get(key)
member_event = current_state.get(key)
key = (EventTypes.Create, "", )
create_event_id = current_state_ids.get(key)
if create_event_id:
auth_ids.append(create_event_id)
create_event = current_state.get(key)
if create_event:
auth_ids.append(create_event.event_id)
if join_rule_event_id:
join_rule_event = yield self.store.get_event(join_rule_event_id)
if join_rule_event:
join_rule = join_rule_event.content.get("join_rule")
is_public = join_rule == JoinRules.PUBLIC if join_rule else False
else:
@@ -941,21 +789,15 @@ class Auth(object):
if event.type == EventTypes.Member:
e_type = event.content["membership"]
if e_type in [Membership.JOIN, Membership.INVITE]:
if join_rule_event_id:
auth_ids.append(join_rule_event_id)
if join_rule_event:
auth_ids.append(join_rule_event.event_id)
if e_type == Membership.JOIN:
if member_event_id and not is_public:
auth_ids.append(member_event_id)
if member_event and not is_public:
auth_ids.append(member_event.event_id)
else:
if member_event_id:
auth_ids.append(member_event_id)
if for_verification:
key = (EventTypes.Member, event.state_key, )
existing_event_id = current_state_ids.get(key)
if existing_event_id:
auth_ids.append(existing_event_id)
if member_event:
auth_ids.append(member_event.event_id)
if e_type == Membership.INVITE:
if "third_party_invite" in event.content:
@@ -963,15 +805,14 @@ class Auth(object):
EventTypes.ThirdPartyInvite,
event.content["third_party_invite"]["signed"]["token"]
)
third_party_invite_id = current_state_ids.get(key)
if third_party_invite_id:
auth_ids.append(third_party_invite_id)
elif member_event_id:
member_event = yield self.store.get_event(member_event_id)
third_party_invite = current_state.get(key)
if third_party_invite:
auth_ids.append(third_party_invite.event_id)
elif member_event:
if member_event.content["membership"] == Membership.JOIN:
auth_ids.append(member_event.event_id)
defer.returnValue(auth_ids)
return auth_ids
def _get_send_level(self, etype, state_key, auth_events):
key = (EventTypes.PowerLevels, "", )
@@ -1020,6 +861,16 @@ class Auth(object):
403,
"You are not allowed to set others state"
)
else:
sender_domain = UserID.from_string(
event.user_id
).domain
if sender_domain != event.state_key:
raise AuthError(
403,
"You are not allowed to set others state"
)
return True
@@ -1043,8 +894,8 @@ class Auth(object):
if user_level >= redact_level:
return False
redacter_domain = get_domain_from_id(event.event_id)
redactee_domain = get_domain_from_id(event.redacts)
redacter_domain = EventID.from_string(event.event_id).domain
redactee_domain = EventID.from_string(event.redacts).domain
if redacter_domain == redactee_domain:
return True
@@ -1177,68 +1028,3 @@ class Auth(object):
"This server requires you to be a moderator in the room to"
" edit its room list entry"
)
def has_access_token(request):
"""Checks if the request has an access_token.
Returns:
bool: False if no access_token was given, True otherwise.
"""
query_params = request.args.get("access_token")
auth_headers = request.requestHeaders.getRawHeaders("Authorization")
return bool(query_params) or bool(auth_headers)
def get_access_token_from_request(request, token_not_found_http_status=401):
"""Extracts the access_token from the request.
Args:
request: The http request.
token_not_found_http_status(int): The HTTP status code to set in the
AuthError if the token isn't found. This is used in some of the
legacy APIs to change the status code to 403 from the default of
401 since some of the old clients depended on auth errors returning
403.
Returns:
str: The access_token
Raises:
AuthError: If there isn't an access_token in the request.
"""
auth_headers = request.requestHeaders.getRawHeaders("Authorization")
query_params = request.args.get("access_token")
if auth_headers:
# Try the get the access_token from a "Authorization: Bearer"
# header
if query_params is not None:
raise AuthError(
token_not_found_http_status,
"Mixing Authorization headers and access_token query parameters.",
errcode=Codes.MISSING_TOKEN,
)
if len(auth_headers) > 1:
raise AuthError(
token_not_found_http_status,
"Too many Authorization headers.",
errcode=Codes.MISSING_TOKEN,
)
parts = auth_headers[0].split(" ")
if parts[0] == "Bearer" and len(parts) == 2:
return parts[1]
else:
raise AuthError(
token_not_found_http_status,
"Invalid Authorization header.",
errcode=Codes.MISSING_TOKEN,
)
else:
# Try to get the access_token from the query params.
if not query_params:
raise AuthError(
token_not_found_http_status,
"Missing access token.",
errcode=Codes.MISSING_TOKEN
)
return query_params[0]

View File

@@ -85,8 +85,3 @@ class RoomCreationPreset(object):
PRIVATE_CHAT = "private_chat"
PUBLIC_CHAT = "public_chat"
TRUSTED_PRIVATE_CHAT = "trusted_private_chat"
class ThirdPartyEntityKind(object):
USER = "user"
LOCATION = "location"

View File

@@ -39,14 +39,11 @@ class Codes(object):
CAPTCHA_NEEDED = "M_CAPTCHA_NEEDED"
CAPTCHA_INVALID = "M_CAPTCHA_INVALID"
MISSING_PARAM = "M_MISSING_PARAM"
INVALID_PARAM = "M_INVALID_PARAM"
TOO_LARGE = "M_TOO_LARGE"
EXCLUSIVE = "M_EXCLUSIVE"
THREEPID_AUTH_FAILED = "M_THREEPID_AUTH_FAILED"
THREEPID_IN_USE = "M_THREEPID_IN_USE"
THREEPID_NOT_FOUND = "M_THREEPID_NOT_FOUND"
THREEPID_IN_USE = "THREEPID_IN_USE"
INVALID_USERNAME = "M_INVALID_USERNAME"
SERVER_NOT_TRUSTED = "M_SERVER_NOT_TRUSTED"
class CodeMessageException(RuntimeError):

View File

@@ -15,9 +15,7 @@
from synapse.api.errors import SynapseError
from synapse.types import UserID, RoomID
from twisted.internet import defer
import ujson as json
import json
class Filtering(object):
@@ -26,10 +24,10 @@ class Filtering(object):
super(Filtering, self).__init__()
self.store = hs.get_datastore()
@defer.inlineCallbacks
def get_user_filter(self, user_localpart, filter_id):
result = yield self.store.get_user_filter(user_localpart, filter_id)
defer.returnValue(FilterCollection(result))
result = self.store.get_user_filter(user_localpart, filter_id)
result.addCallback(FilterCollection)
return result
def add_user_filter(self, user_localpart, user_filter):
self.check_valid_filter(user_filter)
@@ -71,21 +69,6 @@ class Filtering(object):
if key in user_filter_json["room"]:
self._check_definition(user_filter_json["room"][key])
if "event_fields" in user_filter_json:
if type(user_filter_json["event_fields"]) != list:
raise SynapseError(400, "event_fields must be a list of strings")
for field in user_filter_json["event_fields"]:
if not isinstance(field, basestring):
raise SynapseError(400, "Event field must be a string")
# Don't allow '\\' in event field filters. This makes matching
# events a lot easier as we can then use a negative lookbehind
# assertion to split '\.' If we allowed \\ then it would
# incorrectly split '\\.' See synapse.events.utils.serialize_event
if r'\\' in field:
raise SynapseError(
400, r'The escape character \ cannot itself be escaped'
)
def _check_definition_room_lists(self, definition):
"""Check that "rooms" and "not_rooms" are lists of room ids if they
are present
@@ -167,7 +150,6 @@ class FilterCollection(object):
self.include_leave = filter_json.get("room", {}).get(
"include_leave", False
)
self.event_fields = filter_json.get("event_fields", [])
def __repr__(self):
return "<FilterCollection %s>" % (json.dumps(self._filter_json),)
@@ -202,51 +184,11 @@ class FilterCollection(object):
def filter_room_account_data(self, events):
return self._room_account_data.filter(self._room_filter.filter(events))
def blocks_all_presence(self):
return (
self._presence_filter.filters_all_types() or
self._presence_filter.filters_all_senders()
)
def blocks_all_room_ephemeral(self):
return (
self._room_ephemeral_filter.filters_all_types() or
self._room_ephemeral_filter.filters_all_senders() or
self._room_ephemeral_filter.filters_all_rooms()
)
def blocks_all_room_timeline(self):
return (
self._room_timeline_filter.filters_all_types() or
self._room_timeline_filter.filters_all_senders() or
self._room_timeline_filter.filters_all_rooms()
)
class Filter(object):
def __init__(self, filter_json):
self.filter_json = filter_json
self.types = self.filter_json.get("types", None)
self.not_types = self.filter_json.get("not_types", [])
self.rooms = self.filter_json.get("rooms", None)
self.not_rooms = self.filter_json.get("not_rooms", [])
self.senders = self.filter_json.get("senders", None)
self.not_senders = self.filter_json.get("not_senders", [])
self.contains_url = self.filter_json.get("contains_url", None)
def filters_all_types(self):
return "*" in self.not_types
def filters_all_senders(self):
return "*" in self.not_senders
def filters_all_rooms(self):
return "*" in self.not_rooms
def check(self, event):
"""Checks whether the filter matches the given event.
@@ -265,10 +207,9 @@ class Filter(object):
event.get("room_id", None),
sender,
event.get("type", None),
"url" in event.get("content", {})
)
def check_fields(self, room_id, sender, event_type, contains_url):
def check_fields(self, room_id, sender, event_type):
"""Checks whether the filter matches the given event fields.
Returns:
@@ -282,20 +223,15 @@ class Filter(object):
for name, match_func in literal_keys.items():
not_name = "not_%s" % (name,)
disallowed_values = getattr(self, not_name)
disallowed_values = self.filter_json.get(not_name, [])
if any(map(match_func, disallowed_values)):
return False
allowed_values = getattr(self, name)
allowed_values = self.filter_json.get(name, None)
if allowed_values is not None:
if not any(map(match_func, allowed_values)):
return False
contains_url_filter = self.filter_json.get("contains_url")
if contains_url_filter is not None:
if contains_url_filter != contains_url:
return False
return True
def filter_rooms(self, room_ids):

View File

@@ -23,7 +23,7 @@ class Ratelimiter(object):
def __init__(self):
self.message_counts = collections.OrderedDict()
def send_message(self, user_id, time_now_s, msg_rate_hz, burst_count, update=True):
def send_message(self, user_id, time_now_s, msg_rate_hz, burst_count):
"""Can the user send a message?
Args:
user_id: The user sending a message.
@@ -32,15 +32,12 @@ class Ratelimiter(object):
second.
burst_count: How many messages the user can send before being
limited.
update (bool): Whether to update the message rates or not. This is
useful to check if a message would be allowed to be sent before
its ready to be actually sent.
Returns:
A pair of a bool indicating if they can send a message now and a
time in seconds of when they can next send a message.
"""
self.prune_message_counts(time_now_s)
message_count, time_start, _ignored = self.message_counts.get(
message_count, time_start, _ignored = self.message_counts.pop(
user_id, (0., time_now_s, None),
)
time_delta = time_now_s - time_start
@@ -55,10 +52,9 @@ class Ratelimiter(object):
allowed = True
message_count += 1
if update:
self.message_counts[user_id] = (
message_count, time_start, msg_rate_hz
)
self.message_counts[user_id] = (
message_count, time_start, msg_rate_hz
)
if msg_rate_hz > 0:
time_allowed = (

View File

@@ -25,3 +25,4 @@ SERVER_KEY_PREFIX = "/_matrix/key/v1"
SERVER_KEY_V2_PREFIX = "/_matrix/key/v2"
MEDIA_PREFIX = "/_matrix/media/r0"
LEGACY_MEDIA_PREFIX = "/_matrix/media/v1"
APP_SERVICE_PREFIX = "/_matrix/appservice/v1"

View File

@@ -16,11 +16,13 @@
import sys
sys.dont_write_bytecode = True
from synapse import python_dependencies # noqa: E402
from synapse.python_dependencies import (
check_requirements, MissingRequirementError
) # NOQA
try:
python_dependencies.check_requirements()
except python_dependencies.MissingRequirementError as e:
check_requirements()
except MissingRequirementError as e:
message = "\n".join([
"Missing Requirement: %s" % (e.message,),
"To install run:",

View File

@@ -1,214 +0,0 @@
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Copyright 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import synapse
from synapse.server import HomeServer
from synapse.config._base import ConfigError
from synapse.config.logger import setup_logging
from synapse.config.homeserver import HomeServerConfig
from synapse.http.site import SynapseSite
from synapse.metrics.resource import MetricsResource, METRICS_PREFIX
from synapse.replication.slave.storage.directory import DirectoryStore
from synapse.replication.slave.storage.events import SlavedEventStore
from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
from synapse.replication.slave.storage.registration import SlavedRegistrationStore
from synapse.storage.engines import create_engine
from synapse.util.async import sleep
from synapse.util.httpresourcetree import create_resource_tree
from synapse.util.logcontext import LoggingContext
from synapse.util.manhole import manhole
from synapse.util.rlimit import change_resource_limit
from synapse.util.versionstring import get_version_string
from synapse import events
from twisted.internet import reactor, defer
from twisted.web.resource import Resource
from daemonize import Daemonize
import sys
import logging
import gc
logger = logging.getLogger("synapse.app.appservice")
class AppserviceSlaveStore(
DirectoryStore, SlavedEventStore, SlavedApplicationServiceStore,
SlavedRegistrationStore,
):
pass
class AppserviceServer(HomeServer):
def get_db_conn(self, run_new_connection=True):
# Any param beginning with cp_ is a parameter for adbapi, and should
# not be passed to the database engine.
db_params = {
k: v for k, v in self.db_config.get("args", {}).items()
if not k.startswith("cp_")
}
db_conn = self.database_engine.module.connect(**db_params)
if run_new_connection:
self.database_engine.on_new_connection(db_conn)
return db_conn
def setup(self):
logger.info("Setting up.")
self.datastore = AppserviceSlaveStore(self.get_db_conn(), self)
logger.info("Finished setting up.")
def _listen_http(self, listener_config):
port = listener_config["port"]
bind_address = listener_config.get("bind_address", "")
site_tag = listener_config.get("tag", port)
resources = {}
for res in listener_config["resources"]:
for name in res["names"]:
if name == "metrics":
resources[METRICS_PREFIX] = MetricsResource(self)
root_resource = create_resource_tree(resources, Resource())
reactor.listenTCP(
port,
SynapseSite(
"synapse.access.http.%s" % (site_tag,),
site_tag,
listener_config,
root_resource,
),
interface=bind_address
)
logger.info("Synapse appservice now listening on port %d", port)
def start_listening(self, listeners):
for listener in listeners:
if listener["type"] == "http":
self._listen_http(listener)
elif listener["type"] == "manhole":
reactor.listenTCP(
listener["port"],
manhole(
username="matrix",
password="rabbithole",
globals={"hs": self},
),
interface=listener.get("bind_address", '127.0.0.1')
)
else:
logger.warn("Unrecognized listener type: %s", listener["type"])
@defer.inlineCallbacks
def replicate(self):
http_client = self.get_simple_http_client()
store = self.get_datastore()
replication_url = self.config.worker_replication_url
appservice_handler = self.get_application_service_handler()
@defer.inlineCallbacks
def replicate(results):
stream = results.get("events")
if stream:
max_stream_id = stream["position"]
yield appservice_handler.notify_interested_services(max_stream_id)
while True:
try:
args = store.stream_positions()
args["timeout"] = 30000
result = yield http_client.get_json(replication_url, args=args)
yield store.process_replication(result)
replicate(result)
except:
logger.exception("Error replicating from %r", replication_url)
yield sleep(30)
def start(config_options):
try:
config = HomeServerConfig.load_config(
"Synapse appservice", config_options
)
except ConfigError as e:
sys.stderr.write("\n" + e.message + "\n")
sys.exit(1)
assert config.worker_app == "synapse.app.appservice"
setup_logging(config.worker_log_config, config.worker_log_file)
events.USE_FROZEN_DICTS = config.use_frozen_dicts
database_engine = create_engine(config.database_config)
if config.notify_appservices:
sys.stderr.write(
"\nThe appservices must be disabled in the main synapse process"
"\nbefore they can be run in a separate worker."
"\nPlease add ``notify_appservices: false`` to the main config"
"\n"
)
sys.exit(1)
# Force the pushers to start since they will be disabled in the main config
config.notify_appservices = True
ps = AppserviceServer(
config.server_name,
db_config=config.database_config,
config=config,
version_string="Synapse/" + get_version_string(synapse),
database_engine=database_engine,
)
ps.setup()
ps.start_listening(config.worker_listeners)
def run():
with LoggingContext("run"):
logger.info("Running")
change_resource_limit(config.soft_file_limit)
if config.gc_thresholds:
gc.set_threshold(*config.gc_thresholds)
reactor.run()
def start():
ps.replicate()
ps.get_datastore().start_profiling()
ps.get_state_handler().start_caching()
reactor.callWhenRunning(start)
if config.worker_daemonize:
daemon = Daemonize(
app="synapse-appservice",
pid=config.worker_pid_file,
action=run,
auto_close_fds=False,
verbose=True,
logger=logger,
)
daemon.start()
else:
run()
if __name__ == '__main__':
with LoggingContext("main"):
start(sys.argv[1:])

View File

@@ -1,220 +0,0 @@
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Copyright 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import synapse
from synapse.config._base import ConfigError
from synapse.config.homeserver import HomeServerConfig
from synapse.config.logger import setup_logging
from synapse.http.site import SynapseSite
from synapse.http.server import JsonResource
from synapse.metrics.resource import MetricsResource, METRICS_PREFIX
from synapse.replication.slave.storage._base import BaseSlavedStore
from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
from synapse.replication.slave.storage.events import SlavedEventStore
from synapse.replication.slave.storage.keys import SlavedKeyStore
from synapse.replication.slave.storage.room import RoomStore
from synapse.replication.slave.storage.directory import DirectoryStore
from synapse.replication.slave.storage.registration import SlavedRegistrationStore
from synapse.rest.client.v1.room import PublicRoomListRestServlet
from synapse.server import HomeServer
from synapse.storage.client_ips import ClientIpStore
from synapse.storage.engines import create_engine
from synapse.util.async import sleep
from synapse.util.httpresourcetree import create_resource_tree
from synapse.util.logcontext import LoggingContext
from synapse.util.manhole import manhole
from synapse.util.rlimit import change_resource_limit
from synapse.util.versionstring import get_version_string
from synapse.crypto import context_factory
from synapse import events
from twisted.internet import reactor, defer
from twisted.web.resource import Resource
from daemonize import Daemonize
import sys
import logging
import gc
logger = logging.getLogger("synapse.app.client_reader")
class ClientReaderSlavedStore(
SlavedEventStore,
SlavedKeyStore,
RoomStore,
DirectoryStore,
SlavedApplicationServiceStore,
SlavedRegistrationStore,
BaseSlavedStore,
ClientIpStore, # After BaseSlavedStore because the constructor is different
):
pass
class ClientReaderServer(HomeServer):
def get_db_conn(self, run_new_connection=True):
# Any param beginning with cp_ is a parameter for adbapi, and should
# not be passed to the database engine.
db_params = {
k: v for k, v in self.db_config.get("args", {}).items()
if not k.startswith("cp_")
}
db_conn = self.database_engine.module.connect(**db_params)
if run_new_connection:
self.database_engine.on_new_connection(db_conn)
return db_conn
def setup(self):
logger.info("Setting up.")
self.datastore = ClientReaderSlavedStore(self.get_db_conn(), self)
logger.info("Finished setting up.")
def _listen_http(self, listener_config):
port = listener_config["port"]
bind_address = listener_config.get("bind_address", "")
site_tag = listener_config.get("tag", port)
resources = {}
for res in listener_config["resources"]:
for name in res["names"]:
if name == "metrics":
resources[METRICS_PREFIX] = MetricsResource(self)
elif name == "client":
resource = JsonResource(self, canonical_json=False)
PublicRoomListRestServlet(self).register(resource)
resources.update({
"/_matrix/client/r0": resource,
"/_matrix/client/unstable": resource,
"/_matrix/client/v2_alpha": resource,
"/_matrix/client/api/v1": resource,
})
root_resource = create_resource_tree(resources, Resource())
reactor.listenTCP(
port,
SynapseSite(
"synapse.access.http.%s" % (site_tag,),
site_tag,
listener_config,
root_resource,
),
interface=bind_address
)
logger.info("Synapse client reader now listening on port %d", port)
def start_listening(self, listeners):
for listener in listeners:
if listener["type"] == "http":
self._listen_http(listener)
elif listener["type"] == "manhole":
reactor.listenTCP(
listener["port"],
manhole(
username="matrix",
password="rabbithole",
globals={"hs": self},
),
interface=listener.get("bind_address", '127.0.0.1')
)
else:
logger.warn("Unrecognized listener type: %s", listener["type"])
@defer.inlineCallbacks
def replicate(self):
http_client = self.get_simple_http_client()
store = self.get_datastore()
replication_url = self.config.worker_replication_url
while True:
try:
args = store.stream_positions()
args["timeout"] = 30000
result = yield http_client.get_json(replication_url, args=args)
yield store.process_replication(result)
except:
logger.exception("Error replicating from %r", replication_url)
yield sleep(5)
def start(config_options):
try:
config = HomeServerConfig.load_config(
"Synapse client reader", config_options
)
except ConfigError as e:
sys.stderr.write("\n" + e.message + "\n")
sys.exit(1)
assert config.worker_app == "synapse.app.client_reader"
setup_logging(config.worker_log_config, config.worker_log_file)
events.USE_FROZEN_DICTS = config.use_frozen_dicts
database_engine = create_engine(config.database_config)
tls_server_context_factory = context_factory.ServerContextFactory(config)
ss = ClientReaderServer(
config.server_name,
db_config=config.database_config,
tls_server_context_factory=tls_server_context_factory,
config=config,
version_string="Synapse/" + get_version_string(synapse),
database_engine=database_engine,
)
ss.setup()
ss.get_handlers()
ss.start_listening(config.worker_listeners)
def run():
with LoggingContext("run"):
logger.info("Running")
change_resource_limit(config.soft_file_limit)
if config.gc_thresholds:
gc.set_threshold(*config.gc_thresholds)
reactor.run()
def start():
ss.get_state_handler().start_caching()
ss.get_datastore().start_profiling()
ss.replicate()
reactor.callWhenRunning(start)
if config.worker_daemonize:
daemon = Daemonize(
app="synapse-client-reader",
pid=config.worker_pid_file,
action=run,
auto_close_fds=False,
verbose=True,
logger=logger,
)
daemon.start()
else:
run()
if __name__ == '__main__':
with LoggingContext("main"):
start(sys.argv[1:])

View File

@@ -1,211 +0,0 @@
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Copyright 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import synapse
from synapse.config._base import ConfigError
from synapse.config.homeserver import HomeServerConfig
from synapse.config.logger import setup_logging
from synapse.http.site import SynapseSite
from synapse.metrics.resource import MetricsResource, METRICS_PREFIX
from synapse.replication.slave.storage._base import BaseSlavedStore
from synapse.replication.slave.storage.events import SlavedEventStore
from synapse.replication.slave.storage.keys import SlavedKeyStore
from synapse.replication.slave.storage.room import RoomStore
from synapse.replication.slave.storage.transactions import TransactionStore
from synapse.replication.slave.storage.directory import DirectoryStore
from synapse.server import HomeServer
from synapse.storage.engines import create_engine
from synapse.util.async import sleep
from synapse.util.httpresourcetree import create_resource_tree
from synapse.util.logcontext import LoggingContext
from synapse.util.manhole import manhole
from synapse.util.rlimit import change_resource_limit
from synapse.util.versionstring import get_version_string
from synapse.api.urls import FEDERATION_PREFIX
from synapse.federation.transport.server import TransportLayerServer
from synapse.crypto import context_factory
from synapse import events
from twisted.internet import reactor, defer
from twisted.web.resource import Resource
from daemonize import Daemonize
import sys
import logging
import gc
logger = logging.getLogger("synapse.app.federation_reader")
class FederationReaderSlavedStore(
SlavedEventStore,
SlavedKeyStore,
RoomStore,
DirectoryStore,
TransactionStore,
BaseSlavedStore,
):
pass
class FederationReaderServer(HomeServer):
def get_db_conn(self, run_new_connection=True):
# Any param beginning with cp_ is a parameter for adbapi, and should
# not be passed to the database engine.
db_params = {
k: v for k, v in self.db_config.get("args", {}).items()
if not k.startswith("cp_")
}
db_conn = self.database_engine.module.connect(**db_params)
if run_new_connection:
self.database_engine.on_new_connection(db_conn)
return db_conn
def setup(self):
logger.info("Setting up.")
self.datastore = FederationReaderSlavedStore(self.get_db_conn(), self)
logger.info("Finished setting up.")
def _listen_http(self, listener_config):
port = listener_config["port"]
bind_address = listener_config.get("bind_address", "")
site_tag = listener_config.get("tag", port)
resources = {}
for res in listener_config["resources"]:
for name in res["names"]:
if name == "metrics":
resources[METRICS_PREFIX] = MetricsResource(self)
elif name == "federation":
resources.update({
FEDERATION_PREFIX: TransportLayerServer(self),
})
root_resource = create_resource_tree(resources, Resource())
reactor.listenTCP(
port,
SynapseSite(
"synapse.access.http.%s" % (site_tag,),
site_tag,
listener_config,
root_resource,
),
interface=bind_address
)
logger.info("Synapse federation reader now listening on port %d", port)
def start_listening(self, listeners):
for listener in listeners:
if listener["type"] == "http":
self._listen_http(listener)
elif listener["type"] == "manhole":
reactor.listenTCP(
listener["port"],
manhole(
username="matrix",
password="rabbithole",
globals={"hs": self},
),
interface=listener.get("bind_address", '127.0.0.1')
)
else:
logger.warn("Unrecognized listener type: %s", listener["type"])
@defer.inlineCallbacks
def replicate(self):
http_client = self.get_simple_http_client()
store = self.get_datastore()
replication_url = self.config.worker_replication_url
while True:
try:
args = store.stream_positions()
args["timeout"] = 30000
result = yield http_client.get_json(replication_url, args=args)
yield store.process_replication(result)
except:
logger.exception("Error replicating from %r", replication_url)
yield sleep(5)
def start(config_options):
try:
config = HomeServerConfig.load_config(
"Synapse federation reader", config_options
)
except ConfigError as e:
sys.stderr.write("\n" + e.message + "\n")
sys.exit(1)
assert config.worker_app == "synapse.app.federation_reader"
setup_logging(config.worker_log_config, config.worker_log_file)
events.USE_FROZEN_DICTS = config.use_frozen_dicts
database_engine = create_engine(config.database_config)
tls_server_context_factory = context_factory.ServerContextFactory(config)
ss = FederationReaderServer(
config.server_name,
db_config=config.database_config,
tls_server_context_factory=tls_server_context_factory,
config=config,
version_string="Synapse/" + get_version_string(synapse),
database_engine=database_engine,
)
ss.setup()
ss.get_handlers()
ss.start_listening(config.worker_listeners)
def run():
with LoggingContext("run"):
logger.info("Running")
change_resource_limit(config.soft_file_limit)
if config.gc_thresholds:
gc.set_threshold(*config.gc_thresholds)
reactor.run()
def start():
ss.get_state_handler().start_caching()
ss.get_datastore().start_profiling()
ss.replicate()
reactor.callWhenRunning(start)
if config.worker_daemonize:
daemon = Daemonize(
app="synapse-federation-reader",
pid=config.worker_pid_file,
action=run,
auto_close_fds=False,
verbose=True,
logger=logger,
)
daemon.start()
else:
run()
if __name__ == '__main__':
with LoggingContext("main"):
start(sys.argv[1:])

View File

@@ -1,331 +0,0 @@
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Copyright 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import synapse
from synapse.server import HomeServer
from synapse.config._base import ConfigError
from synapse.config.logger import setup_logging
from synapse.config.homeserver import HomeServerConfig
from synapse.crypto import context_factory
from synapse.http.site import SynapseSite
from synapse.federation import send_queue
from synapse.federation.units import Edu
from synapse.metrics.resource import MetricsResource, METRICS_PREFIX
from synapse.replication.slave.storage.deviceinbox import SlavedDeviceInboxStore
from synapse.replication.slave.storage.events import SlavedEventStore
from synapse.replication.slave.storage.receipts import SlavedReceiptsStore
from synapse.replication.slave.storage.registration import SlavedRegistrationStore
from synapse.replication.slave.storage.transactions import TransactionStore
from synapse.storage.engines import create_engine
from synapse.storage.presence import UserPresenceState
from synapse.util.async import sleep
from synapse.util.httpresourcetree import create_resource_tree
from synapse.util.logcontext import LoggingContext
from synapse.util.manhole import manhole
from synapse.util.rlimit import change_resource_limit
from synapse.util.versionstring import get_version_string
from synapse import events
from twisted.internet import reactor, defer
from twisted.web.resource import Resource
from daemonize import Daemonize
import sys
import logging
import gc
import ujson as json
logger = logging.getLogger("synapse.app.appservice")
class FederationSenderSlaveStore(
SlavedDeviceInboxStore, TransactionStore, SlavedReceiptsStore, SlavedEventStore,
SlavedRegistrationStore,
):
pass
class FederationSenderServer(HomeServer):
def get_db_conn(self, run_new_connection=True):
# Any param beginning with cp_ is a parameter for adbapi, and should
# not be passed to the database engine.
db_params = {
k: v for k, v in self.db_config.get("args", {}).items()
if not k.startswith("cp_")
}
db_conn = self.database_engine.module.connect(**db_params)
if run_new_connection:
self.database_engine.on_new_connection(db_conn)
return db_conn
def setup(self):
logger.info("Setting up.")
self.datastore = FederationSenderSlaveStore(self.get_db_conn(), self)
logger.info("Finished setting up.")
def _listen_http(self, listener_config):
port = listener_config["port"]
bind_address = listener_config.get("bind_address", "")
site_tag = listener_config.get("tag", port)
resources = {}
for res in listener_config["resources"]:
for name in res["names"]:
if name == "metrics":
resources[METRICS_PREFIX] = MetricsResource(self)
root_resource = create_resource_tree(resources, Resource())
reactor.listenTCP(
port,
SynapseSite(
"synapse.access.http.%s" % (site_tag,),
site_tag,
listener_config,
root_resource,
),
interface=bind_address
)
logger.info("Synapse federation_sender now listening on port %d", port)
def start_listening(self, listeners):
for listener in listeners:
if listener["type"] == "http":
self._listen_http(listener)
elif listener["type"] == "manhole":
reactor.listenTCP(
listener["port"],
manhole(
username="matrix",
password="rabbithole",
globals={"hs": self},
),
interface=listener.get("bind_address", '127.0.0.1')
)
else:
logger.warn("Unrecognized listener type: %s", listener["type"])
@defer.inlineCallbacks
def replicate(self):
http_client = self.get_simple_http_client()
store = self.get_datastore()
replication_url = self.config.worker_replication_url
send_handler = FederationSenderHandler(self)
send_handler.on_start()
while True:
try:
args = store.stream_positions()
args.update((yield send_handler.stream_positions()))
args["timeout"] = 30000
result = yield http_client.get_json(replication_url, args=args)
yield store.process_replication(result)
yield send_handler.process_replication(result)
except:
logger.exception("Error replicating from %r", replication_url)
yield sleep(30)
def start(config_options):
try:
config = HomeServerConfig.load_config(
"Synapse federation sender", config_options
)
except ConfigError as e:
sys.stderr.write("\n" + e.message + "\n")
sys.exit(1)
assert config.worker_app == "synapse.app.federation_sender"
setup_logging(config.worker_log_config, config.worker_log_file)
events.USE_FROZEN_DICTS = config.use_frozen_dicts
database_engine = create_engine(config.database_config)
if config.send_federation:
sys.stderr.write(
"\nThe send_federation must be disabled in the main synapse process"
"\nbefore they can be run in a separate worker."
"\nPlease add ``send_federation: false`` to the main config"
"\n"
)
sys.exit(1)
# Force the pushers to start since they will be disabled in the main config
config.send_federation = True
tls_server_context_factory = context_factory.ServerContextFactory(config)
ps = FederationSenderServer(
config.server_name,
db_config=config.database_config,
tls_server_context_factory=tls_server_context_factory,
config=config,
version_string="Synapse/" + get_version_string(synapse),
database_engine=database_engine,
)
ps.setup()
ps.start_listening(config.worker_listeners)
def run():
with LoggingContext("run"):
logger.info("Running")
change_resource_limit(config.soft_file_limit)
if config.gc_thresholds:
gc.set_threshold(*config.gc_thresholds)
reactor.run()
def start():
ps.replicate()
ps.get_datastore().start_profiling()
ps.get_state_handler().start_caching()
reactor.callWhenRunning(start)
if config.worker_daemonize:
daemon = Daemonize(
app="synapse-federation-sender",
pid=config.worker_pid_file,
action=run,
auto_close_fds=False,
verbose=True,
logger=logger,
)
daemon.start()
else:
run()
class FederationSenderHandler(object):
"""Processes the replication stream and forwards the appropriate entries
to the federation sender.
"""
def __init__(self, hs):
self.store = hs.get_datastore()
self.federation_sender = hs.get_federation_sender()
self._room_serials = {}
self._room_typing = {}
def on_start(self):
# There may be some events that are persisted but haven't been sent,
# so send them now.
self.federation_sender.notify_new_events(
self.store.get_room_max_stream_ordering()
)
@defer.inlineCallbacks
def stream_positions(self):
stream_id = yield self.store.get_federation_out_pos("federation")
defer.returnValue({
"federation": stream_id,
# Ack stuff we've "processed", this should only be called from
# one process.
"federation_ack": stream_id,
})
@defer.inlineCallbacks
def process_replication(self, result):
# The federation stream contains things that we want to send out, e.g.
# presence, typing, etc.
fed_stream = result.get("federation")
if fed_stream:
latest_id = int(fed_stream["position"])
# The federation stream containis a bunch of different types of
# rows that need to be handled differently. We parse the rows, put
# them into the appropriate collection and then send them off.
presence_to_send = {}
keyed_edus = {}
edus = {}
failures = {}
device_destinations = set()
# Parse the rows in the stream
for row in fed_stream["rows"]:
position, typ, content_js = row
content = json.loads(content_js)
if typ == send_queue.PRESENCE_TYPE:
destination = content["destination"]
state = UserPresenceState.from_dict(content["state"])
presence_to_send.setdefault(destination, []).append(state)
elif typ == send_queue.KEYED_EDU_TYPE:
key = content["key"]
edu = Edu(**content["edu"])
keyed_edus.setdefault(
edu.destination, {}
)[(edu.destination, tuple(key))] = edu
elif typ == send_queue.EDU_TYPE:
edu = Edu(**content)
edus.setdefault(edu.destination, []).append(edu)
elif typ == send_queue.FAILURE_TYPE:
destination = content["destination"]
failure = content["failure"]
failures.setdefault(destination, []).append(failure)
elif typ == send_queue.DEVICE_MESSAGE_TYPE:
device_destinations.add(content["destination"])
else:
raise Exception("Unrecognised federation type: %r", typ)
# We've finished collecting, send everything off
for destination, states in presence_to_send.items():
self.federation_sender.send_presence(destination, states)
for destination, edu_map in keyed_edus.items():
for key, edu in edu_map.items():
self.federation_sender.send_edu(
edu.destination, edu.edu_type, edu.content, key=key,
)
for destination, edu_list in edus.items():
for edu in edu_list:
self.federation_sender.send_edu(
edu.destination, edu.edu_type, edu.content, key=None,
)
for destination, failure_list in failures.items():
for failure in failure_list:
self.federation_sender.send_failure(destination, failure)
for destination in device_destinations:
self.federation_sender.send_device_messages(destination)
# Record where we are in the stream.
yield self.store.update_federation_out_pos(
"federation", latest_id
)
# We also need to poke the federation sender when new events happen
event_stream = result.get("events")
if event_stream:
latest_pos = event_stream["position"]
self.federation_sender.notify_new_events(latest_pos)
if __name__ == '__main__':
with LoggingContext("main"):
start(sys.argv[1:])

View File

@@ -16,10 +16,14 @@
import synapse
import gc
import contextlib
import logging
import os
import re
import resource
import subprocess
import sys
import time
from synapse.config._base import ConfigError
from synapse.python_dependencies import (
@@ -33,11 +37,18 @@ from synapse.storage.prepare_database import UpgradeDatabaseException, prepare_d
from synapse.server import HomeServer
from twisted.conch.manhole import ColoredManhole
from twisted.conch.insults import insults
from twisted.conch import manhole_ssh
from twisted.cred import checkers, portal
from twisted.internet import reactor, task, defer
from twisted.application import service
from twisted.web.resource import Resource, EncodingResourceWrapper
from twisted.web.static import File
from twisted.web.server import GzipEncoderFactory
from twisted.web.server import Site, GzipEncoderFactory, Request
from synapse.http.server import RootRedirect
from synapse.rest.media.v0.content_repository import ContentRepoResource
from synapse.rest.media.v1.media_repository import MediaRepositoryResource
@@ -51,18 +62,10 @@ from synapse.api.urls import (
from synapse.config.homeserver import HomeServerConfig
from synapse.crypto import context_factory
from synapse.util.logcontext import LoggingContext
from synapse.metrics import register_memory_metrics, get_metrics_for
from synapse.metrics.resource import MetricsResource, METRICS_PREFIX
from synapse.replication.resource import ReplicationResource, REPLICATION_PREFIX
from synapse.federation.transport.server import TransportLayerServer
from synapse.util.rlimit import change_resource_limit
from synapse.util.versionstring import get_version_string
from synapse.util.httpresourcetree import create_resource_tree
from synapse.util.manhole import manhole
from synapse.http.site import SynapseSite
from synapse import events
from daemonize import Daemonize
@@ -70,6 +73,9 @@ from daemonize import Daemonize
logger = logging.getLogger("synapse.app.homeserver")
ACCESS_TOKEN_RE = re.compile(r'(\?.*access(_|%5[Ff])token=)[^&]*(.*)$')
def gz_wrap(r):
return EncodingResourceWrapper(r, [GzipEncoderFactory()])
@@ -148,7 +154,7 @@ class SynapseHomeServer(HomeServer):
MEDIA_PREFIX: media_repo,
LEGACY_MEDIA_PREFIX: media_repo,
CONTENT_REPO_PREFIX: ContentRepoResource(
self, self.config.uploads_path
self, self.config.uploads_path, self.auth, self.content_addr
),
})
@@ -167,12 +173,7 @@ class SynapseHomeServer(HomeServer):
if name == "replication":
resources[REPLICATION_PREFIX] = ReplicationResource(self)
if WEB_CLIENT_PREFIX in resources:
root_resource = RootRedirect(WEB_CLIENT_PREFIX)
else:
root_resource = Resource()
root_resource = create_resource_tree(resources, root_resource)
root_resource = create_resource_tree(resources)
if tls:
reactor.listenSSL(
port,
@@ -205,13 +206,24 @@ class SynapseHomeServer(HomeServer):
if listener["type"] == "http":
self._listener_http(config, listener)
elif listener["type"] == "manhole":
checker = checkers.InMemoryUsernamePasswordDatabaseDontUse(
matrix="rabbithole"
)
rlm = manhole_ssh.TerminalRealm()
rlm.chainedProtocolFactory = lambda: insults.ServerProtocol(
ColoredManhole,
{
"__name__": "__console__",
"hs": self,
}
)
f = manhole_ssh.ConchFactory(portal.Portal(rlm, [checker]))
reactor.listenTCP(
listener["port"],
manhole(
username="matrix",
password="rabbithole",
globals={"hs": self},
),
f,
interface=listener.get("bind_address", '127.0.0.1')
)
else:
@@ -257,6 +269,86 @@ def quit_with_error(error_string):
sys.exit(1)
def get_version_string():
try:
null = open(os.devnull, 'w')
cwd = os.path.dirname(os.path.abspath(__file__))
try:
git_branch = subprocess.check_output(
['git', 'rev-parse', '--abbrev-ref', 'HEAD'],
stderr=null,
cwd=cwd,
).strip()
git_branch = "b=" + git_branch
except subprocess.CalledProcessError:
git_branch = ""
try:
git_tag = subprocess.check_output(
['git', 'describe', '--exact-match'],
stderr=null,
cwd=cwd,
).strip()
git_tag = "t=" + git_tag
except subprocess.CalledProcessError:
git_tag = ""
try:
git_commit = subprocess.check_output(
['git', 'rev-parse', '--short', 'HEAD'],
stderr=null,
cwd=cwd,
).strip()
except subprocess.CalledProcessError:
git_commit = ""
try:
dirty_string = "-this_is_a_dirty_checkout"
is_dirty = subprocess.check_output(
['git', 'describe', '--dirty=' + dirty_string],
stderr=null,
cwd=cwd,
).strip().endswith(dirty_string)
git_dirty = "dirty" if is_dirty else ""
except subprocess.CalledProcessError:
git_dirty = ""
if git_branch or git_tag or git_commit or git_dirty:
git_version = ",".join(
s for s in
(git_branch, git_tag, git_commit, git_dirty,)
if s
)
return (
"Synapse/%s (%s)" % (
synapse.__version__, git_version,
)
).encode("ascii")
except Exception as e:
logger.info("Failed to check for git repository: %s", e)
return ("Synapse/%s" % (synapse.__version__,)).encode("ascii")
def change_resource_limit(soft_file_no):
try:
soft, hard = resource.getrlimit(resource.RLIMIT_NOFILE)
if not soft_file_no:
soft_file_no = hard
resource.setrlimit(resource.RLIMIT_NOFILE, (soft_file_no, hard))
logger.info("Set file limit to: %d", soft_file_no)
resource.setrlimit(
resource.RLIMIT_CORE, (resource.RLIM_INFINITY, resource.RLIM_INFINITY)
)
except (ValueError, resource.error) as e:
logger.warn("Failed to set file or core limit: %s", e)
def setup(config_options):
"""
Args:
@@ -267,9 +359,10 @@ def setup(config_options):
HomeServer
"""
try:
config = HomeServerConfig.load_or_generate_config(
config = HomeServerConfig.load_config(
"Synapse Homeserver",
config_options,
generate_section="Homeserver"
)
except ConfigError as e:
sys.stderr.write("\n" + e.message + "\n")
@@ -285,7 +378,7 @@ def setup(config_options):
# check any extra requirements we have now we have a config
check_requirements(config)
version_string = "Synapse/" + get_version_string(synapse)
version_string = get_version_string()
logger.info("Server hostname: %s", config.server_name)
logger.info("Server version: %s", version_string)
@@ -302,6 +395,7 @@ def setup(config_options):
db_config=config.database_config,
tls_server_context_factory=tls_server_context_factory,
config=config,
content_addr=config.content_addr,
version_string=version_string,
database_engine=database_engine,
)
@@ -336,8 +430,6 @@ def setup(config_options):
hs.get_datastore().start_doing_background_updates()
hs.get_replication_layer().start_get_pdu_cache()
register_memory_metrics(hs)
reactor.callWhenRunning(start)
return hs
@@ -353,13 +445,215 @@ class SynapseService(service.Service):
def startService(self):
hs = setup(self.config)
change_resource_limit(hs.config.soft_file_limit)
if hs.config.gc_thresholds:
gc.set_threshold(*hs.config.gc_thresholds)
def stopService(self):
return self._port.stopListening()
class SynapseRequest(Request):
def __init__(self, site, *args, **kw):
Request.__init__(self, *args, **kw)
self.site = site
self.authenticated_entity = None
self.start_time = 0
def __repr__(self):
# We overwrite this so that we don't log ``access_token``
return '<%s at 0x%x method=%s uri=%s clientproto=%s site=%s>' % (
self.__class__.__name__,
id(self),
self.method,
self.get_redacted_uri(),
self.clientproto,
self.site.site_tag,
)
def get_redacted_uri(self):
return ACCESS_TOKEN_RE.sub(
r'\1<redacted>\3',
self.uri
)
def get_user_agent(self):
return self.requestHeaders.getRawHeaders("User-Agent", [None])[-1]
def started_processing(self):
self.site.access_logger.info(
"%s - %s - Received request: %s %s",
self.getClientIP(),
self.site.site_tag,
self.method,
self.get_redacted_uri()
)
self.start_time = int(time.time() * 1000)
def finished_processing(self):
try:
context = LoggingContext.current_context()
ru_utime, ru_stime = context.get_resource_usage()
db_txn_count = context.db_txn_count
db_txn_duration = context.db_txn_duration
except:
ru_utime, ru_stime = (0, 0)
db_txn_count, db_txn_duration = (0, 0)
self.site.access_logger.info(
"%s - %s - {%s}"
" Processed request: %dms (%dms, %dms) (%dms/%d)"
" %sB %s \"%s %s %s\" \"%s\"",
self.getClientIP(),
self.site.site_tag,
self.authenticated_entity,
int(time.time() * 1000) - self.start_time,
int(ru_utime * 1000),
int(ru_stime * 1000),
int(db_txn_duration * 1000),
int(db_txn_count),
self.sentLength,
self.code,
self.method,
self.get_redacted_uri(),
self.clientproto,
self.get_user_agent(),
)
@contextlib.contextmanager
def processing(self):
self.started_processing()
yield
self.finished_processing()
class XForwardedForRequest(SynapseRequest):
def __init__(self, *args, **kw):
SynapseRequest.__init__(self, *args, **kw)
"""
Add a layer on top of another request that only uses the value of an
X-Forwarded-For header as the result of C{getClientIP}.
"""
def getClientIP(self):
"""
@return: The client address (the first address) in the value of the
I{X-Forwarded-For header}. If the header is not present, return
C{b"-"}.
"""
return self.requestHeaders.getRawHeaders(
b"x-forwarded-for", [b"-"])[0].split(b",")[0].strip()
class SynapseRequestFactory(object):
def __init__(self, site, x_forwarded_for):
self.site = site
self.x_forwarded_for = x_forwarded_for
def __call__(self, *args, **kwargs):
if self.x_forwarded_for:
return XForwardedForRequest(self.site, *args, **kwargs)
else:
return SynapseRequest(self.site, *args, **kwargs)
class SynapseSite(Site):
"""
Subclass of a twisted http Site that does access logging with python's
standard logging
"""
def __init__(self, logger_name, site_tag, config, resource, *args, **kwargs):
Site.__init__(self, resource, *args, **kwargs)
self.site_tag = site_tag
proxied = config.get("x_forwarded", False)
self.requestFactory = SynapseRequestFactory(self, proxied)
self.access_logger = logging.getLogger(logger_name)
def log(self, request):
pass
def create_resource_tree(desired_tree, redirect_root_to_web_client=True):
"""Create the resource tree for this Home Server.
This in unduly complicated because Twisted does not support putting
child resources more than 1 level deep at a time.
Args:
web_client (bool): True to enable the web client.
redirect_root_to_web_client (bool): True to redirect '/' to the
location of the web client. This does nothing if web_client is not
True.
"""
if redirect_root_to_web_client and WEB_CLIENT_PREFIX in desired_tree:
root_resource = RootRedirect(WEB_CLIENT_PREFIX)
else:
root_resource = Resource()
# ideally we'd just use getChild and putChild but getChild doesn't work
# unless you give it a Request object IN ADDITION to the name :/ So
# instead, we'll store a copy of this mapping so we can actually add
# extra resources to existing nodes. See self._resource_id for the key.
resource_mappings = {}
for full_path, res in desired_tree.items():
logger.info("Attaching %s to path %s", res, full_path)
last_resource = root_resource
for path_seg in full_path.split('/')[1:-1]:
if path_seg not in last_resource.listNames():
# resource doesn't exist, so make a "dummy resource"
child_resource = Resource()
last_resource.putChild(path_seg, child_resource)
res_id = _resource_id(last_resource, path_seg)
resource_mappings[res_id] = child_resource
last_resource = child_resource
else:
# we have an existing Resource, use that instead.
res_id = _resource_id(last_resource, path_seg)
last_resource = resource_mappings[res_id]
# ===========================
# now attach the actual desired resource
last_path_seg = full_path.split('/')[-1]
# if there is already a resource here, thieve its children and
# replace it
res_id = _resource_id(last_resource, last_path_seg)
if res_id in resource_mappings:
# there is a dummy resource at this path already, which needs
# to be replaced with the desired resource.
existing_dummy_resource = resource_mappings[res_id]
for child_name in existing_dummy_resource.listNames():
child_res_id = _resource_id(
existing_dummy_resource, child_name
)
child_resource = resource_mappings[child_res_id]
# steal the children
res.putChild(child_name, child_resource)
# finally, insert the desired resource in the right place
last_resource.putChild(last_path_seg, res)
res_id = _resource_id(last_resource, last_path_seg)
resource_mappings[res_id] = res
return root_resource
def _resource_id(resource, path_seg):
"""Construct an arbitrary resource ID so you can retrieve the mapping
later.
If you want to represent resource A putChild resource B with path C,
the mapping should looks like _resource_id(A,C) = B.
Args:
resource (Resource): The *parent* Resourceb
path_seg (str): The name of the child Resource to be attached.
Returns:
str: A unique string which can be a key to the child Resource.
"""
return "%s-%s" % (resource, path_seg)
def run(hs):
PROFILE_SYNAPSE = False
if PROFILE_SYNAPSE:
@@ -385,8 +679,6 @@ def run(hs):
start_time = hs.get_clock().time()
stats = {}
@defer.inlineCallbacks
def phone_stats_home():
logger.info("Gathering stats for reporting")
@@ -395,10 +687,7 @@ def run(hs):
if uptime < 0:
uptime = 0
# If the stats directory is empty then this is the first time we've
# reported stats.
first_time = not stats
stats = {}
stats["homeserver"] = hs.config.server_name
stats["timestamp"] = now
stats["uptime_seconds"] = uptime
@@ -411,25 +700,6 @@ def run(hs):
daily_messages = yield hs.get_datastore().count_daily_messages()
if daily_messages is not None:
stats["daily_messages"] = daily_messages
else:
stats.pop("daily_messages", None)
if first_time:
# Add callbacks to report the synapse stats as metrics whenever
# prometheus requests them, typically every 30s.
# As some of the stats are expensive to calculate we only update
# them when synapse phones home to matrix.org every 24 hours.
metrics = get_metrics_for("synapse.usage")
metrics.add_callback("timestamp", lambda: stats["timestamp"])
metrics.add_callback("uptime_seconds", lambda: stats["uptime_seconds"])
metrics.add_callback("total_users", lambda: stats["total_users"])
metrics.add_callback("total_room_count", lambda: stats["total_room_count"])
metrics.add_callback(
"daily_active_users", lambda: stats["daily_active_users"]
)
metrics.add_callback(
"daily_messages", lambda: stats.get("daily_messages", 0)
)
logger.info("Reporting stats to matrix.org: %s" % (stats,))
try:
@@ -450,8 +720,6 @@ def run(hs):
# sys.settrace(logcontext_tracer)
with LoggingContext("run"):
change_resource_limit(hs.config.soft_file_limit)
if hs.config.gc_thresholds:
gc.set_threshold(*hs.config.gc_thresholds)
reactor.run()
if hs.config.daemonize:

View File

@@ -1,217 +0,0 @@
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Copyright 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import synapse
from synapse.config._base import ConfigError
from synapse.config.homeserver import HomeServerConfig
from synapse.config.logger import setup_logging
from synapse.http.site import SynapseSite
from synapse.metrics.resource import MetricsResource, METRICS_PREFIX
from synapse.replication.slave.storage._base import BaseSlavedStore
from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
from synapse.replication.slave.storage.registration import SlavedRegistrationStore
from synapse.rest.media.v0.content_repository import ContentRepoResource
from synapse.rest.media.v1.media_repository import MediaRepositoryResource
from synapse.server import HomeServer
from synapse.storage.client_ips import ClientIpStore
from synapse.storage.engines import create_engine
from synapse.storage.media_repository import MediaRepositoryStore
from synapse.util.async import sleep
from synapse.util.httpresourcetree import create_resource_tree
from synapse.util.logcontext import LoggingContext
from synapse.util.manhole import manhole
from synapse.util.rlimit import change_resource_limit
from synapse.util.versionstring import get_version_string
from synapse.api.urls import (
CONTENT_REPO_PREFIX, LEGACY_MEDIA_PREFIX, MEDIA_PREFIX
)
from synapse.crypto import context_factory
from synapse import events
from twisted.internet import reactor, defer
from twisted.web.resource import Resource
from daemonize import Daemonize
import sys
import logging
import gc
logger = logging.getLogger("synapse.app.media_repository")
class MediaRepositorySlavedStore(
SlavedApplicationServiceStore,
SlavedRegistrationStore,
BaseSlavedStore,
MediaRepositoryStore,
ClientIpStore,
):
pass
class MediaRepositoryServer(HomeServer):
def get_db_conn(self, run_new_connection=True):
# Any param beginning with cp_ is a parameter for adbapi, and should
# not be passed to the database engine.
db_params = {
k: v for k, v in self.db_config.get("args", {}).items()
if not k.startswith("cp_")
}
db_conn = self.database_engine.module.connect(**db_params)
if run_new_connection:
self.database_engine.on_new_connection(db_conn)
return db_conn
def setup(self):
logger.info("Setting up.")
self.datastore = MediaRepositorySlavedStore(self.get_db_conn(), self)
logger.info("Finished setting up.")
def _listen_http(self, listener_config):
port = listener_config["port"]
bind_address = listener_config.get("bind_address", "")
site_tag = listener_config.get("tag", port)
resources = {}
for res in listener_config["resources"]:
for name in res["names"]:
if name == "metrics":
resources[METRICS_PREFIX] = MetricsResource(self)
elif name == "media":
media_repo = MediaRepositoryResource(self)
resources.update({
MEDIA_PREFIX: media_repo,
LEGACY_MEDIA_PREFIX: media_repo,
CONTENT_REPO_PREFIX: ContentRepoResource(
self, self.config.uploads_path
),
})
root_resource = create_resource_tree(resources, Resource())
reactor.listenTCP(
port,
SynapseSite(
"synapse.access.http.%s" % (site_tag,),
site_tag,
listener_config,
root_resource,
),
interface=bind_address
)
logger.info("Synapse media repository now listening on port %d", port)
def start_listening(self, listeners):
for listener in listeners:
if listener["type"] == "http":
self._listen_http(listener)
elif listener["type"] == "manhole":
reactor.listenTCP(
listener["port"],
manhole(
username="matrix",
password="rabbithole",
globals={"hs": self},
),
interface=listener.get("bind_address", '127.0.0.1')
)
else:
logger.warn("Unrecognized listener type: %s", listener["type"])
@defer.inlineCallbacks
def replicate(self):
http_client = self.get_simple_http_client()
store = self.get_datastore()
replication_url = self.config.worker_replication_url
while True:
try:
args = store.stream_positions()
args["timeout"] = 30000
result = yield http_client.get_json(replication_url, args=args)
yield store.process_replication(result)
except:
logger.exception("Error replicating from %r", replication_url)
yield sleep(5)
def start(config_options):
try:
config = HomeServerConfig.load_config(
"Synapse media repository", config_options
)
except ConfigError as e:
sys.stderr.write("\n" + e.message + "\n")
sys.exit(1)
assert config.worker_app == "synapse.app.media_repository"
setup_logging(config.worker_log_config, config.worker_log_file)
events.USE_FROZEN_DICTS = config.use_frozen_dicts
database_engine = create_engine(config.database_config)
tls_server_context_factory = context_factory.ServerContextFactory(config)
ss = MediaRepositoryServer(
config.server_name,
db_config=config.database_config,
tls_server_context_factory=tls_server_context_factory,
config=config,
version_string="Synapse/" + get_version_string(synapse),
database_engine=database_engine,
)
ss.setup()
ss.get_handlers()
ss.start_listening(config.worker_listeners)
def run():
with LoggingContext("run"):
logger.info("Running")
change_resource_limit(config.soft_file_limit)
if config.gc_thresholds:
gc.set_threshold(*config.gc_thresholds)
reactor.run()
def start():
ss.get_state_handler().start_caching()
ss.get_datastore().start_profiling()
ss.replicate()
reactor.callWhenRunning(start)
if config.worker_daemonize:
daemon = Daemonize(
app="synapse-media-repository",
pid=config.worker_pid_file,
action=run,
auto_close_fds=False,
verbose=True,
logger=logger,
)
daemon.start()
else:
run()
if __name__ == '__main__':
with LoggingContext("main"):
start(sys.argv[1:])

View File

@@ -1,303 +0,0 @@
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Copyright 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import synapse
from synapse.server import HomeServer
from synapse.config._base import ConfigError
from synapse.config.logger import setup_logging
from synapse.config.homeserver import HomeServerConfig
from synapse.http.site import SynapseSite
from synapse.metrics.resource import MetricsResource, METRICS_PREFIX
from synapse.storage.roommember import RoomMemberStore
from synapse.replication.slave.storage.events import SlavedEventStore
from synapse.replication.slave.storage.pushers import SlavedPusherStore
from synapse.replication.slave.storage.receipts import SlavedReceiptsStore
from synapse.replication.slave.storage.account_data import SlavedAccountDataStore
from synapse.storage.engines import create_engine
from synapse.storage import DataStore
from synapse.util.async import sleep
from synapse.util.httpresourcetree import create_resource_tree
from synapse.util.logcontext import LoggingContext, preserve_fn
from synapse.util.manhole import manhole
from synapse.util.rlimit import change_resource_limit
from synapse.util.versionstring import get_version_string
from synapse import events
from twisted.internet import reactor, defer
from twisted.web.resource import Resource
from daemonize import Daemonize
import sys
import logging
import gc
logger = logging.getLogger("synapse.app.pusher")
class PusherSlaveStore(
SlavedEventStore, SlavedPusherStore, SlavedReceiptsStore,
SlavedAccountDataStore
):
update_pusher_last_stream_ordering_and_success = (
DataStore.update_pusher_last_stream_ordering_and_success.__func__
)
update_pusher_failing_since = (
DataStore.update_pusher_failing_since.__func__
)
update_pusher_last_stream_ordering = (
DataStore.update_pusher_last_stream_ordering.__func__
)
get_throttle_params_by_room = (
DataStore.get_throttle_params_by_room.__func__
)
set_throttle_params = (
DataStore.set_throttle_params.__func__
)
get_time_of_last_push_action_before = (
DataStore.get_time_of_last_push_action_before.__func__
)
get_profile_displayname = (
DataStore.get_profile_displayname.__func__
)
who_forgot_in_room = (
RoomMemberStore.__dict__["who_forgot_in_room"]
)
class PusherServer(HomeServer):
def get_db_conn(self, run_new_connection=True):
# Any param beginning with cp_ is a parameter for adbapi, and should
# not be passed to the database engine.
db_params = {
k: v for k, v in self.db_config.get("args", {}).items()
if not k.startswith("cp_")
}
db_conn = self.database_engine.module.connect(**db_params)
if run_new_connection:
self.database_engine.on_new_connection(db_conn)
return db_conn
def setup(self):
logger.info("Setting up.")
self.datastore = PusherSlaveStore(self.get_db_conn(), self)
logger.info("Finished setting up.")
def remove_pusher(self, app_id, push_key, user_id):
http_client = self.get_simple_http_client()
replication_url = self.config.worker_replication_url
url = replication_url + "/remove_pushers"
return http_client.post_json_get_json(url, {
"remove": [{
"app_id": app_id,
"push_key": push_key,
"user_id": user_id,
}]
})
def _listen_http(self, listener_config):
port = listener_config["port"]
bind_address = listener_config.get("bind_address", "")
site_tag = listener_config.get("tag", port)
resources = {}
for res in listener_config["resources"]:
for name in res["names"]:
if name == "metrics":
resources[METRICS_PREFIX] = MetricsResource(self)
root_resource = create_resource_tree(resources, Resource())
reactor.listenTCP(
port,
SynapseSite(
"synapse.access.http.%s" % (site_tag,),
site_tag,
listener_config,
root_resource,
),
interface=bind_address
)
logger.info("Synapse pusher now listening on port %d", port)
def start_listening(self, listeners):
for listener in listeners:
if listener["type"] == "http":
self._listen_http(listener)
elif listener["type"] == "manhole":
reactor.listenTCP(
listener["port"],
manhole(
username="matrix",
password="rabbithole",
globals={"hs": self},
),
interface=listener.get("bind_address", '127.0.0.1')
)
else:
logger.warn("Unrecognized listener type: %s", listener["type"])
@defer.inlineCallbacks
def replicate(self):
http_client = self.get_simple_http_client()
store = self.get_datastore()
replication_url = self.config.worker_replication_url
pusher_pool = self.get_pusherpool()
def stop_pusher(user_id, app_id, pushkey):
key = "%s:%s" % (app_id, pushkey)
pushers_for_user = pusher_pool.pushers.get(user_id, {})
pusher = pushers_for_user.pop(key, None)
if pusher is None:
return
logger.info("Stopping pusher %r / %r", user_id, key)
pusher.on_stop()
def start_pusher(user_id, app_id, pushkey):
key = "%s:%s" % (app_id, pushkey)
logger.info("Starting pusher %r / %r", user_id, key)
return pusher_pool._refresh_pusher(app_id, pushkey, user_id)
@defer.inlineCallbacks
def poke_pushers(results):
pushers_rows = set(
map(tuple, results.get("pushers", {}).get("rows", []))
)
deleted_pushers_rows = set(
map(tuple, results.get("deleted_pushers", {}).get("rows", []))
)
for row in sorted(pushers_rows | deleted_pushers_rows):
if row in deleted_pushers_rows:
user_id, app_id, pushkey = row[1:4]
stop_pusher(user_id, app_id, pushkey)
elif row in pushers_rows:
user_id = row[1]
app_id = row[5]
pushkey = row[8]
yield start_pusher(user_id, app_id, pushkey)
stream = results.get("events")
if stream and stream["rows"]:
min_stream_id = stream["rows"][0][0]
max_stream_id = stream["position"]
preserve_fn(pusher_pool.on_new_notifications)(
min_stream_id, max_stream_id
)
stream = results.get("receipts")
if stream and stream["rows"]:
rows = stream["rows"]
affected_room_ids = set(row[1] for row in rows)
min_stream_id = rows[0][0]
max_stream_id = stream["position"]
preserve_fn(pusher_pool.on_new_receipts)(
min_stream_id, max_stream_id, affected_room_ids
)
while True:
try:
args = store.stream_positions()
args["timeout"] = 30000
result = yield http_client.get_json(replication_url, args=args)
yield store.process_replication(result)
poke_pushers(result)
except:
logger.exception("Error replicating from %r", replication_url)
yield sleep(30)
def start(config_options):
try:
config = HomeServerConfig.load_config(
"Synapse pusher", config_options
)
except ConfigError as e:
sys.stderr.write("\n" + e.message + "\n")
sys.exit(1)
assert config.worker_app == "synapse.app.pusher"
setup_logging(config.worker_log_config, config.worker_log_file)
events.USE_FROZEN_DICTS = config.use_frozen_dicts
if config.start_pushers:
sys.stderr.write(
"\nThe pushers must be disabled in the main synapse process"
"\nbefore they can be run in a separate worker."
"\nPlease add ``start_pushers: false`` to the main config"
"\n"
)
sys.exit(1)
# Force the pushers to start since they will be disabled in the main config
config.start_pushers = True
database_engine = create_engine(config.database_config)
ps = PusherServer(
config.server_name,
db_config=config.database_config,
config=config,
version_string="Synapse/" + get_version_string(synapse),
database_engine=database_engine,
)
ps.setup()
ps.start_listening(config.worker_listeners)
def run():
with LoggingContext("run"):
logger.info("Running")
change_resource_limit(config.soft_file_limit)
if config.gc_thresholds:
gc.set_threshold(*config.gc_thresholds)
reactor.run()
def start():
ps.replicate()
ps.get_pusherpool().start()
ps.get_datastore().start_profiling()
ps.get_state_handler().start_caching()
reactor.callWhenRunning(start)
if config.worker_daemonize:
daemon = Daemonize(
app="synapse-pusher",
pid=config.worker_pid_file,
action=run,
auto_close_fds=False,
verbose=True,
logger=logger,
)
daemon.start()
else:
run()
if __name__ == '__main__':
with LoggingContext("main"):
ps = start(sys.argv[1:])

View File

@@ -1,496 +0,0 @@
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Copyright 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import synapse
from synapse.api.constants import EventTypes, PresenceState
from synapse.config._base import ConfigError
from synapse.config.homeserver import HomeServerConfig
from synapse.config.logger import setup_logging
from synapse.events import FrozenEvent
from synapse.handlers.presence import PresenceHandler
from synapse.http.site import SynapseSite
from synapse.http.server import JsonResource
from synapse.metrics.resource import MetricsResource, METRICS_PREFIX
from synapse.rest.client.v2_alpha import sync
from synapse.rest.client.v1 import events
from synapse.rest.client.v1.room import RoomInitialSyncRestServlet
from synapse.rest.client.v1.initial_sync import InitialSyncRestServlet
from synapse.replication.slave.storage._base import BaseSlavedStore
from synapse.replication.slave.storage.events import SlavedEventStore
from synapse.replication.slave.storage.receipts import SlavedReceiptsStore
from synapse.replication.slave.storage.account_data import SlavedAccountDataStore
from synapse.replication.slave.storage.appservice import SlavedApplicationServiceStore
from synapse.replication.slave.storage.registration import SlavedRegistrationStore
from synapse.replication.slave.storage.filtering import SlavedFilteringStore
from synapse.replication.slave.storage.push_rule import SlavedPushRuleStore
from synapse.replication.slave.storage.presence import SlavedPresenceStore
from synapse.replication.slave.storage.deviceinbox import SlavedDeviceInboxStore
from synapse.replication.slave.storage.room import RoomStore
from synapse.server import HomeServer
from synapse.storage.client_ips import ClientIpStore
from synapse.storage.engines import create_engine
from synapse.storage.presence import PresenceStore, UserPresenceState
from synapse.storage.roommember import RoomMemberStore
from synapse.util.async import sleep
from synapse.util.httpresourcetree import create_resource_tree
from synapse.util.logcontext import LoggingContext, preserve_fn
from synapse.util.manhole import manhole
from synapse.util.rlimit import change_resource_limit
from synapse.util.stringutils import random_string
from synapse.util.versionstring import get_version_string
from twisted.internet import reactor, defer
from twisted.web.resource import Resource
from daemonize import Daemonize
import sys
import logging
import contextlib
import gc
import ujson as json
logger = logging.getLogger("synapse.app.synchrotron")
class SynchrotronSlavedStore(
SlavedPushRuleStore,
SlavedEventStore,
SlavedReceiptsStore,
SlavedAccountDataStore,
SlavedApplicationServiceStore,
SlavedRegistrationStore,
SlavedFilteringStore,
SlavedPresenceStore,
SlavedDeviceInboxStore,
RoomStore,
BaseSlavedStore,
ClientIpStore, # After BaseSlavedStore because the constructor is different
):
who_forgot_in_room = (
RoomMemberStore.__dict__["who_forgot_in_room"]
)
# XXX: This is a bit broken because we don't persist the accepted list in a
# way that can be replicated. This means that we don't have a way to
# invalidate the cache correctly.
get_presence_list_accepted = PresenceStore.__dict__[
"get_presence_list_accepted"
]
get_presence_list_observers_accepted = PresenceStore.__dict__[
"get_presence_list_observers_accepted"
]
UPDATE_SYNCING_USERS_MS = 10 * 1000
class SynchrotronPresence(object):
def __init__(self, hs):
self.is_mine_id = hs.is_mine_id
self.http_client = hs.get_simple_http_client()
self.store = hs.get_datastore()
self.user_to_num_current_syncs = {}
self.syncing_users_url = hs.config.worker_replication_url + "/syncing_users"
self.clock = hs.get_clock()
self.notifier = hs.get_notifier()
active_presence = self.store.take_presence_startup_info()
self.user_to_current_state = {
state.user_id: state
for state in active_presence
}
self.process_id = random_string(16)
logger.info("Presence process_id is %r", self.process_id)
self._sending_sync = False
self._need_to_send_sync = False
self.clock.looping_call(
self._send_syncing_users_regularly,
UPDATE_SYNCING_USERS_MS,
)
reactor.addSystemEventTrigger("before", "shutdown", self._on_shutdown)
def set_state(self, user, state, ignore_status_msg=False):
# TODO Hows this supposed to work?
pass
get_states = PresenceHandler.get_states.__func__
get_state = PresenceHandler.get_state.__func__
_get_interested_parties = PresenceHandler._get_interested_parties.__func__
current_state_for_users = PresenceHandler.current_state_for_users.__func__
@defer.inlineCallbacks
def user_syncing(self, user_id, affect_presence):
if affect_presence:
curr_sync = self.user_to_num_current_syncs.get(user_id, 0)
self.user_to_num_current_syncs[user_id] = curr_sync + 1
prev_states = yield self.current_state_for_users([user_id])
if prev_states[user_id].state == PresenceState.OFFLINE:
# TODO: Don't block the sync request on this HTTP hit.
yield self._send_syncing_users_now()
def _end():
# We check that the user_id is in user_to_num_current_syncs because
# user_to_num_current_syncs may have been cleared if we are
# shutting down.
if affect_presence and user_id in self.user_to_num_current_syncs:
self.user_to_num_current_syncs[user_id] -= 1
@contextlib.contextmanager
def _user_syncing():
try:
yield
finally:
_end()
defer.returnValue(_user_syncing())
@defer.inlineCallbacks
def _on_shutdown(self):
# When the synchrotron is shutdown tell the master to clear the in
# progress syncs for this process
self.user_to_num_current_syncs.clear()
yield self._send_syncing_users_now()
def _send_syncing_users_regularly(self):
# Only send an update if we aren't in the middle of sending one.
if not self._sending_sync:
preserve_fn(self._send_syncing_users_now)()
@defer.inlineCallbacks
def _send_syncing_users_now(self):
if self._sending_sync:
# We don't want to race with sending another update.
# Instead we wait for that update to finish and send another
# update afterwards.
self._need_to_send_sync = True
return
# Flag that we are sending an update.
self._sending_sync = True
yield self.http_client.post_json_get_json(self.syncing_users_url, {
"process_id": self.process_id,
"syncing_users": [
user_id for user_id, count in self.user_to_num_current_syncs.items()
if count > 0
],
})
# Unset the flag as we are no longer sending an update.
self._sending_sync = False
if self._need_to_send_sync:
# If something happened while we were sending the update then
# we might need to send another update.
# TODO: Check if the update that was sent matches the current state
# as we only need to send an update if they are different.
self._need_to_send_sync = False
yield self._send_syncing_users_now()
@defer.inlineCallbacks
def notify_from_replication(self, states, stream_id):
parties = yield self._get_interested_parties(
states, calculate_remote_hosts=False
)
room_ids_to_states, users_to_states, _ = parties
self.notifier.on_new_event(
"presence_key", stream_id, rooms=room_ids_to_states.keys(),
users=users_to_states.keys()
)
@defer.inlineCallbacks
def process_replication(self, result):
stream = result.get("presence", {"rows": []})
states = []
for row in stream["rows"]:
(
position, user_id, state, last_active_ts,
last_federation_update_ts, last_user_sync_ts, status_msg,
currently_active
) = row
state = UserPresenceState(
user_id, state, last_active_ts,
last_federation_update_ts, last_user_sync_ts, status_msg,
currently_active
)
self.user_to_current_state[user_id] = state
states.append(state)
if states and "position" in stream:
stream_id = int(stream["position"])
yield self.notify_from_replication(states, stream_id)
class SynchrotronTyping(object):
def __init__(self, hs):
self._latest_room_serial = 0
self._room_serials = {}
self._room_typing = {}
def stream_positions(self):
# We must update this typing token from the response of the previous
# sync. In particular, the stream id may "reset" back to zero/a low
# value which we *must* use for the next replication request.
return {"typing": self._latest_room_serial}
def process_replication(self, result):
stream = result.get("typing")
if stream:
self._latest_room_serial = int(stream["position"])
for row in stream["rows"]:
position, room_id, typing_json = row
typing = json.loads(typing_json)
self._room_serials[room_id] = position
self._room_typing[room_id] = typing
class SynchrotronApplicationService(object):
def notify_interested_services(self, event):
pass
class SynchrotronServer(HomeServer):
def get_db_conn(self, run_new_connection=True):
# Any param beginning with cp_ is a parameter for adbapi, and should
# not be passed to the database engine.
db_params = {
k: v for k, v in self.db_config.get("args", {}).items()
if not k.startswith("cp_")
}
db_conn = self.database_engine.module.connect(**db_params)
if run_new_connection:
self.database_engine.on_new_connection(db_conn)
return db_conn
def setup(self):
logger.info("Setting up.")
self.datastore = SynchrotronSlavedStore(self.get_db_conn(), self)
logger.info("Finished setting up.")
def _listen_http(self, listener_config):
port = listener_config["port"]
bind_address = listener_config.get("bind_address", "")
site_tag = listener_config.get("tag", port)
resources = {}
for res in listener_config["resources"]:
for name in res["names"]:
if name == "metrics":
resources[METRICS_PREFIX] = MetricsResource(self)
elif name == "client":
resource = JsonResource(self, canonical_json=False)
sync.register_servlets(self, resource)
events.register_servlets(self, resource)
InitialSyncRestServlet(self).register(resource)
RoomInitialSyncRestServlet(self).register(resource)
resources.update({
"/_matrix/client/r0": resource,
"/_matrix/client/unstable": resource,
"/_matrix/client/v2_alpha": resource,
"/_matrix/client/api/v1": resource,
})
root_resource = create_resource_tree(resources, Resource())
reactor.listenTCP(
port,
SynapseSite(
"synapse.access.http.%s" % (site_tag,),
site_tag,
listener_config,
root_resource,
),
interface=bind_address
)
logger.info("Synapse synchrotron now listening on port %d", port)
def start_listening(self, listeners):
for listener in listeners:
if listener["type"] == "http":
self._listen_http(listener)
elif listener["type"] == "manhole":
reactor.listenTCP(
listener["port"],
manhole(
username="matrix",
password="rabbithole",
globals={"hs": self},
),
interface=listener.get("bind_address", '127.0.0.1')
)
else:
logger.warn("Unrecognized listener type: %s", listener["type"])
@defer.inlineCallbacks
def replicate(self):
http_client = self.get_simple_http_client()
store = self.get_datastore()
replication_url = self.config.worker_replication_url
notifier = self.get_notifier()
presence_handler = self.get_presence_handler()
typing_handler = self.get_typing_handler()
def notify_from_stream(
result, stream_name, stream_key, room=None, user=None
):
stream = result.get(stream_name)
if stream:
position_index = stream["field_names"].index("position")
if room:
room_index = stream["field_names"].index(room)
if user:
user_index = stream["field_names"].index(user)
users = ()
rooms = ()
for row in stream["rows"]:
position = row[position_index]
if user:
users = (row[user_index],)
if room:
rooms = (row[room_index],)
notifier.on_new_event(
stream_key, position, users=users, rooms=rooms
)
def notify(result):
stream = result.get("events")
if stream:
max_position = stream["position"]
for row in stream["rows"]:
position = row[0]
internal = json.loads(row[1])
event_json = json.loads(row[2])
event = FrozenEvent(event_json, internal_metadata_dict=internal)
extra_users = ()
if event.type == EventTypes.Member:
extra_users = (event.state_key,)
notifier.on_new_room_event(
event, position, max_position, extra_users
)
notify_from_stream(
result, "push_rules", "push_rules_key", user="user_id"
)
notify_from_stream(
result, "user_account_data", "account_data_key", user="user_id"
)
notify_from_stream(
result, "room_account_data", "account_data_key", user="user_id"
)
notify_from_stream(
result, "tag_account_data", "account_data_key", user="user_id"
)
notify_from_stream(
result, "receipts", "receipt_key", room="room_id"
)
notify_from_stream(
result, "typing", "typing_key", room="room_id"
)
notify_from_stream(
result, "to_device", "to_device_key", user="user_id"
)
while True:
try:
args = store.stream_positions()
args.update(typing_handler.stream_positions())
args["timeout"] = 30000
result = yield http_client.get_json(replication_url, args=args)
yield store.process_replication(result)
typing_handler.process_replication(result)
yield presence_handler.process_replication(result)
notify(result)
except:
logger.exception("Error replicating from %r", replication_url)
yield sleep(5)
def build_presence_handler(self):
return SynchrotronPresence(self)
def build_typing_handler(self):
return SynchrotronTyping(self)
def start(config_options):
try:
config = HomeServerConfig.load_config(
"Synapse synchrotron", config_options
)
except ConfigError as e:
sys.stderr.write("\n" + e.message + "\n")
sys.exit(1)
assert config.worker_app == "synapse.app.synchrotron"
setup_logging(config.worker_log_config, config.worker_log_file)
synapse.events.USE_FROZEN_DICTS = config.use_frozen_dicts
database_engine = create_engine(config.database_config)
ss = SynchrotronServer(
config.server_name,
db_config=config.database_config,
config=config,
version_string="Synapse/" + get_version_string(synapse),
database_engine=database_engine,
application_service_handler=SynchrotronApplicationService(),
)
ss.setup()
ss.start_listening(config.worker_listeners)
def run():
with LoggingContext("run"):
logger.info("Running")
change_resource_limit(config.soft_file_limit)
if config.gc_thresholds:
gc.set_threshold(*config.gc_thresholds)
reactor.run()
def start():
ss.get_datastore().start_profiling()
ss.replicate()
ss.get_state_handler().start_caching()
reactor.callWhenRunning(start)
if config.worker_daemonize:
daemon = Daemonize(
app="synapse-synchrotron",
pid=config.worker_pid_file,
action=run,
auto_close_fds=False,
verbose=True,
logger=logger,
)
daemon.start()
else:
run()
if __name__ == '__main__':
with LoggingContext("main"):
start(sys.argv[1:])

View File

@@ -14,198 +14,70 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import argparse
import collections
import glob
import sys
import os
import os.path
import signal
import subprocess
import sys
import signal
import yaml
SYNAPSE = [sys.executable, "-B", "-m", "synapse.app.homeserver"]
SYNAPSE = ["python", "-B", "-m", "synapse.app.homeserver"]
GREEN = "\x1b[1;32m"
RED = "\x1b[1;31m"
NORMAL = "\x1b[m"
def write(message, colour=NORMAL, stream=sys.stdout):
if colour == NORMAL:
stream.write(message + "\n")
else:
stream.write(colour + message + NORMAL + "\n")
def start(configfile):
write("Starting ...")
print ("Starting ...")
args = SYNAPSE
args.extend(["--daemonize", "-c", configfile])
try:
subprocess.check_call(args)
write("started synapse.app.homeserver(%r)" % (configfile,), colour=GREEN)
print (GREEN + "started" + NORMAL)
except subprocess.CalledProcessError as e:
write(
"error starting (exit code: %d); see above for logs" % e.returncode,
colour=RED,
print (
RED +
"error starting (exit code: %d); see above for logs" % e.returncode +
NORMAL
)
def start_worker(app, configfile, worker_configfile):
args = [
"python", "-B",
"-m", app,
"-c", configfile,
"-c", worker_configfile
]
try:
subprocess.check_call(args)
write("started %s(%r)" % (app, worker_configfile), colour=GREEN)
except subprocess.CalledProcessError as e:
write(
"error starting %s(%r) (exit code: %d); see above for logs" % (
app, worker_configfile, e.returncode,
),
colour=RED,
)
def stop(pidfile, app):
def stop(pidfile):
if os.path.exists(pidfile):
pid = int(open(pidfile).read())
os.kill(pid, signal.SIGTERM)
write("stopped %s" % (app,), colour=GREEN)
Worker = collections.namedtuple("Worker", [
"app", "configfile", "pidfile", "cache_factor"
])
print (GREEN + "stopped" + NORMAL)
def main():
parser = argparse.ArgumentParser()
parser.add_argument(
"action",
choices=["start", "stop", "restart"],
help="whether to start, stop or restart the synapse",
)
parser.add_argument(
"configfile",
nargs="?",
default="homeserver.yaml",
help="the homeserver config file, defaults to homserver.yaml",
)
parser.add_argument(
"-w", "--worker",
metavar="WORKERCONFIG",
help="start or stop a single worker",
)
parser.add_argument(
"-a", "--all-processes",
metavar="WORKERCONFIGDIR",
help="start or stop all the workers in the given directory"
" and the main synapse process",
)
options = parser.parse_args()
if options.worker and options.all_processes:
write(
'Cannot use "--worker" with "--all-processes"',
stream=sys.stderr
)
sys.exit(1)
configfile = options.configfile
configfile = sys.argv[2] if len(sys.argv) == 3 else "homeserver.yaml"
if not os.path.exists(configfile):
write(
sys.stderr.write(
"No config file found\n"
"To generate a config file, run '%s -c %s --generate-config"
" --server-name=<server name>'\n" % (
" ".join(SYNAPSE), options.configfile
),
stream=sys.stderr,
" ".join(SYNAPSE), configfile
)
)
sys.exit(1)
with open(configfile) as stream:
config = yaml.load(stream)
config = yaml.load(open(configfile))
pidfile = config["pid_file"]
cache_factor = config.get("synctl_cache_factor")
start_stop_synapse = True
if cache_factor:
os.environ["SYNAPSE_CACHE_FACTOR"] = str(cache_factor)
worker_configfiles = []
if options.worker:
start_stop_synapse = False
worker_configfile = options.worker
if not os.path.exists(worker_configfile):
write(
"No worker config found at %r" % (worker_configfile,),
stream=sys.stderr,
)
sys.exit(1)
worker_configfiles.append(worker_configfile)
if options.all_processes:
worker_configdir = options.all_processes
if not os.path.isdir(worker_configdir):
write(
"No worker config directory found at %r" % (worker_configdir,),
stream=sys.stderr,
)
sys.exit(1)
worker_configfiles.extend(sorted(glob.glob(
os.path.join(worker_configdir, "*.yaml")
)))
workers = []
for worker_configfile in worker_configfiles:
with open(worker_configfile) as stream:
worker_config = yaml.load(stream)
worker_app = worker_config["worker_app"]
worker_pidfile = worker_config["worker_pid_file"]
worker_daemonize = worker_config["worker_daemonize"]
assert worker_daemonize # TODO print something more user friendly
worker_cache_factor = worker_config.get("synctl_cache_factor")
workers.append(Worker(
worker_app, worker_configfile, worker_pidfile, worker_cache_factor,
))
action = options.action
if action == "stop" or action == "restart":
for worker in workers:
stop(worker.pidfile, worker.app)
if start_stop_synapse:
stop(pidfile, "synapse.app.homeserver")
# TODO: Wait for synapse to actually shutdown before starting it again
if action == "start" or action == "restart":
if start_stop_synapse:
start(configfile)
for worker in workers:
if worker.cache_factor:
os.environ["SYNAPSE_CACHE_FACTOR"] = str(worker.cache_factor)
start_worker(worker.app, configfile, worker.configfile)
if cache_factor:
os.environ["SYNAPSE_CACHE_FACTOR"] = str(cache_factor)
else:
os.environ.pop("SYNAPSE_CACHE_FACTOR", None)
action = sys.argv[1] if sys.argv[1:] else "usage"
if action == "start":
start(configfile)
elif action == "stop":
stop(pidfile)
elif action == "restart":
stop(pidfile)
start(configfile)
else:
sys.stderr.write("Usage: %s [start|stop|restart] [configfile]\n" % (sys.argv[0],))
sys.exit(1)
if __name__ == "__main__":

View File

@@ -14,8 +14,6 @@
# limitations under the License.
from synapse.api.constants import EventTypes
from twisted.internet import defer
import logging
import re
@@ -81,7 +79,7 @@ class ApplicationService(object):
NS_LIST = [NS_USERS, NS_ALIASES, NS_ROOMS]
def __init__(self, token, url=None, namespaces=None, hs_token=None,
sender=None, id=None, protocols=None, rate_limited=True):
sender=None, id=None):
self.token = token
self.url = url
self.hs_token = hs_token
@@ -89,17 +87,6 @@ class ApplicationService(object):
self.namespaces = self._check_namespaces(namespaces)
self.id = id
if "|" in self.id:
raise Exception("application service ID cannot contain '|' character")
# .protocols is a publicly visible field
if protocols:
self.protocols = set(protocols)
else:
self.protocols = set()
self.rate_limited = rate_limited
def _check_namespaces(self, namespaces):
# Sanity check that it is of the form:
# {
@@ -151,66 +138,65 @@ class ApplicationService(object):
return regex_obj["exclusive"]
return False
@defer.inlineCallbacks
def _matches_user(self, event, store):
if not event:
defer.returnValue(False)
if self.is_interested_in_user(event.sender):
defer.returnValue(True)
def _matches_user(self, event, member_list):
if (hasattr(event, "sender") and
self.is_interested_in_user(event.sender)):
return True
# also check m.room.member state key
if (event.type == EventTypes.Member and
self.is_interested_in_user(event.state_key)):
defer.returnValue(True)
if not store:
defer.returnValue(False)
member_list = yield store.get_users_in_room(event.room_id)
if (hasattr(event, "type") and event.type == EventTypes.Member
and hasattr(event, "state_key")
and self.is_interested_in_user(event.state_key)):
return True
# check joined member events
for user_id in member_list:
if self.is_interested_in_user(user_id):
defer.returnValue(True)
defer.returnValue(False)
return True
return False
def _matches_room_id(self, event):
if hasattr(event, "room_id"):
return self.is_interested_in_room(event.room_id)
return False
@defer.inlineCallbacks
def _matches_aliases(self, event, store):
if not store or not event:
defer.returnValue(False)
alias_list = yield store.get_aliases_for_room(event.room_id)
def _matches_aliases(self, event, alias_list):
for alias in alias_list:
if self.is_interested_in_alias(alias):
defer.returnValue(True)
defer.returnValue(False)
return True
return False
@defer.inlineCallbacks
def is_interested(self, event, store=None):
def is_interested(self, event, restrict_to=None, aliases_for_event=None,
member_list=None):
"""Check if this service is interested in this event.
Args:
event(Event): The event to check.
store(DataStore)
restrict_to(str): The namespace to restrict regex tests to.
aliases_for_event(list): A list of all the known room aliases for
this event.
member_list(list): A list of all joined user_ids in this room.
Returns:
bool: True if this service would like to know about this event.
"""
# Do cheap checks first
if self._matches_room_id(event):
defer.returnValue(True)
if aliases_for_event is None:
aliases_for_event = []
if member_list is None:
member_list = []
if (yield self._matches_aliases(event, store)):
defer.returnValue(True)
if restrict_to and restrict_to not in ApplicationService.NS_LIST:
# this is a programming error, so fail early and raise a general
# exception
raise Exception("Unexpected restrict_to value: %s". restrict_to)
if (yield self._matches_user(event, store)):
defer.returnValue(True)
defer.returnValue(False)
if not restrict_to:
return (self._matches_user(event, member_list)
or self._matches_aliases(event, aliases_for_event)
or self._matches_room_id(event))
elif restrict_to == ApplicationService.NS_ALIASES:
return self._matches_aliases(event, aliases_for_event)
elif restrict_to == ApplicationService.NS_ROOMS:
return self._matches_room_id(event)
elif restrict_to == ApplicationService.NS_USERS:
return self._matches_user(event, member_list)
def is_interested_in_user(self, user_id):
return (
@@ -230,17 +216,11 @@ class ApplicationService(object):
or user_id == self.sender
)
def is_interested_in_protocol(self, protocol):
return protocol in self.protocols
def is_exclusive_alias(self, alias):
return self._is_exclusive(ApplicationService.NS_ALIASES, alias)
def is_exclusive_room(self, room_id):
return self._is_exclusive(ApplicationService.NS_ROOMS, room_id)
def is_rate_limited(self):
return self.rate_limited
def __str__(self):
return "ApplicationService: %s" % (self.__dict__,)

View File

@@ -14,12 +14,9 @@
# limitations under the License.
from twisted.internet import defer
from synapse.api.constants import ThirdPartyEntityKind
from synapse.api.errors import CodeMessageException
from synapse.http.client import SimpleHttpClient
from synapse.events.utils import serialize_event
from synapse.util.caches.response_cache import ResponseCache
from synapse.types import ThirdPartyInstanceID
import logging
import urllib
@@ -27,42 +24,6 @@ import urllib
logger = logging.getLogger(__name__)
HOUR_IN_MS = 60 * 60 * 1000
APP_SERVICE_PREFIX = "/_matrix/app/unstable"
def _is_valid_3pe_metadata(info):
if "instances" not in info:
return False
if not isinstance(info["instances"], list):
return False
return True
def _is_valid_3pe_result(r, field):
if not isinstance(r, dict):
return False
for k in (field, "protocol"):
if k not in r:
return False
if not isinstance(r[k], str):
return False
if "fields" not in r:
return False
fields = r["fields"]
if not isinstance(fields, dict):
return False
for k in fields.keys():
if not isinstance(fields[k], str):
return False
return True
class ApplicationServiceApi(SimpleHttpClient):
"""This class manages HS -> AS communications, including querying and
pushing.
@@ -72,12 +33,8 @@ class ApplicationServiceApi(SimpleHttpClient):
super(ApplicationServiceApi, self).__init__(hs)
self.clock = hs.get_clock()
self.protocol_meta_cache = ResponseCache(hs, timeout_ms=HOUR_IN_MS)
@defer.inlineCallbacks
def query_user(self, service, user_id):
if service.url is None:
defer.returnValue(False)
uri = service.url + ("/users/%s" % urllib.quote(user_id))
response = None
try:
@@ -97,8 +54,6 @@ class ApplicationServiceApi(SimpleHttpClient):
@defer.inlineCallbacks
def query_alias(self, service, alias):
if service.url is None:
defer.returnValue(False)
uri = service.url + ("/rooms/%s" % urllib.quote(alias))
response = None
try:
@@ -116,91 +71,8 @@ class ApplicationServiceApi(SimpleHttpClient):
logger.warning("query_alias to %s threw exception %s", uri, ex)
defer.returnValue(False)
@defer.inlineCallbacks
def query_3pe(self, service, kind, protocol, fields):
if kind == ThirdPartyEntityKind.USER:
required_field = "userid"
elif kind == ThirdPartyEntityKind.LOCATION:
required_field = "alias"
else:
raise ValueError(
"Unrecognised 'kind' argument %r to query_3pe()", kind
)
if service.url is None:
defer.returnValue([])
uri = "%s%s/thirdparty/%s/%s" % (
service.url,
APP_SERVICE_PREFIX,
kind,
urllib.quote(protocol)
)
try:
response = yield self.get_json(uri, fields)
if not isinstance(response, list):
logger.warning(
"query_3pe to %s returned an invalid response %r",
uri, response
)
defer.returnValue([])
ret = []
for r in response:
if _is_valid_3pe_result(r, field=required_field):
ret.append(r)
else:
logger.warning(
"query_3pe to %s returned an invalid result %r",
uri, r
)
defer.returnValue(ret)
except Exception as ex:
logger.warning("query_3pe to %s threw exception %s", uri, ex)
defer.returnValue([])
def get_3pe_protocol(self, service, protocol):
if service.url is None:
defer.returnValue({})
@defer.inlineCallbacks
def _get():
uri = "%s%s/thirdparty/protocol/%s" % (
service.url,
APP_SERVICE_PREFIX,
urllib.quote(protocol)
)
try:
info = yield self.get_json(uri, {})
if not _is_valid_3pe_metadata(info):
logger.warning("query_3pe_protocol to %s did not return a"
" valid result", uri)
defer.returnValue(None)
for instance in info.get("instances", []):
network_id = instance.get("network_id", None)
if network_id is not None:
instance["instance_id"] = ThirdPartyInstanceID(
service.id, network_id,
).to_string()
defer.returnValue(info)
except Exception as ex:
logger.warning("query_3pe_protocol to %s threw exception %s",
uri, ex)
defer.returnValue(None)
key = (service.id, protocol)
return self.protocol_meta_cache.get(key) or (
self.protocol_meta_cache.set(key, _get())
)
@defer.inlineCallbacks
def push_bulk(self, service, events, txn_id=None):
if service.url is None:
defer.returnValue(True)
events = self._serialize(events)
if txn_id is None:

View File

@@ -48,35 +48,32 @@ UP & quit +---------- YES SUCCESS
This is all tied together by the AppServiceScheduler which DIs the required
components.
"""
from twisted.internet import defer
from synapse.appservice import ApplicationServiceState
from synapse.util.logcontext import preserve_fn
from synapse.util.metrics import Measure
from twisted.internet import defer
import logging
logger = logging.getLogger(__name__)
class ApplicationServiceScheduler(object):
class AppServiceScheduler(object):
""" Public facing API for this module. Does the required DI to tie the
components together. This also serves as the "event_pool", which in this
case is a simple array.
"""
def __init__(self, hs):
self.clock = hs.get_clock()
self.store = hs.get_datastore()
self.as_api = hs.get_application_service_api()
def __init__(self, clock, store, as_api):
self.clock = clock
self.store = store
self.as_api = as_api
def create_recoverer(service, callback):
return _Recoverer(self.clock, self.store, self.as_api, service, callback)
return _Recoverer(clock, store, as_api, service, callback)
self.txn_ctrl = _TransactionController(
self.clock, self.store, self.as_api, create_recoverer
clock, store, as_api, create_recoverer
)
self.queuer = _ServiceQueuer(self.txn_ctrl, self.clock)
self.queuer = _ServiceQueuer(self.txn_ctrl)
@defer.inlineCallbacks
def start(self):
@@ -97,36 +94,38 @@ class _ServiceQueuer(object):
this schedules any other events in the queue to run.
"""
def __init__(self, txn_ctrl, clock):
def __init__(self, txn_ctrl):
self.queued_events = {} # dict of {service_id: [events]}
self.requests_in_flight = set()
self.pending_requests = {} # dict of {service_id: Deferred}
self.txn_ctrl = txn_ctrl
self.clock = clock
def enqueue(self, service, event):
# if this service isn't being sent something
self.queued_events.setdefault(service.id, []).append(event)
preserve_fn(self._send_request)(service)
if not self.pending_requests.get(service.id):
self._send_request(service, [event])
else:
# add to queue for this service
if service.id not in self.queued_events:
self.queued_events[service.id] = []
self.queued_events[service.id].append(event)
@defer.inlineCallbacks
def _send_request(self, service):
if service.id in self.requests_in_flight:
return
def _send_request(self, service, events):
# send request and add callbacks
d = self.txn_ctrl.send(service, events)
d.addBoth(self._on_request_finish)
d.addErrback(self._on_request_fail)
self.pending_requests[service.id] = d
self.requests_in_flight.add(service.id)
try:
while True:
events = self.queued_events.pop(service.id, [])
if not events:
return
def _on_request_finish(self, service):
self.pending_requests[service.id] = None
# if there are queued events, then send them.
if (service.id in self.queued_events
and len(self.queued_events[service.id]) > 0):
self._send_request(service, self.queued_events[service.id])
self.queued_events[service.id] = []
with Measure(self.clock, "servicequeuer.send"):
try:
yield self.txn_ctrl.send(service, events)
except:
logger.exception("AS request failed")
finally:
self.requests_in_flight.discard(service.id)
def _on_request_fail(self, err):
logger.error("AS request failed: %s", err)
class _TransactionController(object):
@@ -150,12 +149,14 @@ class _TransactionController(object):
if service_is_up:
sent = yield txn.send(self.as_api)
if sent:
yield txn.complete(self.store)
txn.complete(self.store)
else:
preserve_fn(self._start_recoverer)(service)
self._start_recoverer(service)
except Exception as e:
logger.exception(e)
preserve_fn(self._start_recoverer)(service)
self._start_recoverer(service)
# request has finished
defer.returnValue(service)
@defer.inlineCallbacks
def on_recovered(self, recoverer):

View File

@@ -64,12 +64,11 @@ class Config(object):
if isinstance(value, int) or isinstance(value, long):
return value
second = 1000
minute = 60 * second
hour = 60 * minute
hour = 60 * 60 * second
day = 24 * hour
week = 7 * day
year = 365 * day
sizes = {"s": second, "m": minute, "h": hour, "d": day, "w": week, "y": year}
sizes = {"s": second, "h": hour, "d": day, "w": week, "y": year}
size = 1
suffix = value[-1]
if suffix in sizes:
@@ -158,40 +157,9 @@ class Config(object):
return default_config, config
@classmethod
def load_config(cls, description, argv):
config_parser = argparse.ArgumentParser(
description=description,
)
config_parser.add_argument(
"-c", "--config-path",
action="append",
metavar="CONFIG_FILE",
help="Specify config file. Can be given multiple times and"
" may specify directories containing *.yaml files."
)
config_parser.add_argument(
"--keys-directory",
metavar="DIRECTORY",
help="Where files such as certs and signing keys are stored when"
" their location is given explicitly in the config."
" Defaults to the directory containing the last config file",
)
config_args = config_parser.parse_args(argv)
config_files = find_config_files(search_paths=config_args.config_path)
def load_config(cls, description, argv, generate_section=None):
obj = cls()
obj.read_config_files(
config_files,
keys_directory=config_args.keys_directory,
generate_keys=False,
)
return obj
@classmethod
def load_or_generate_config(cls, description, argv):
config_parser = argparse.ArgumentParser(add_help=False)
config_parser.add_argument(
"-c", "--config-path",
@@ -208,7 +176,7 @@ class Config(object):
config_parser.add_argument(
"--report-stats",
action="store",
help="Whether the generated config reports anonymized usage statistics",
help="Stuff",
choices=["yes", "no"]
)
config_parser.add_argument(
@@ -229,11 +197,36 @@ class Config(object):
)
config_args, remaining_args = config_parser.parse_known_args(argv)
config_files = find_config_files(search_paths=config_args.config_path)
generate_keys = config_args.generate_keys
obj = cls()
config_files = []
if config_args.config_path:
for config_path in config_args.config_path:
if os.path.isdir(config_path):
# We accept specifying directories as config paths, we search
# inside that directory for all files matching *.yaml, and then
# we apply them in *sorted* order.
files = []
for entry in os.listdir(config_path):
entry_path = os.path.join(config_path, entry)
if not os.path.isfile(entry_path):
print (
"Found subdirectory in config directory: %r. IGNORING."
) % (entry_path, )
continue
if not entry.endswith(".yaml"):
print (
"Found file in config directory that does not"
" end in '.yaml': %r. IGNORING."
) % (entry_path, )
continue
files.append(entry_path)
config_files.extend(sorted(files))
else:
config_files.append(config_path)
if config_args.generate_config:
if config_args.report_stats is None:
@@ -306,43 +299,28 @@ class Config(object):
" -c CONFIG-FILE\""
)
obj.read_config_files(
config_files,
keys_directory=config_args.keys_directory,
generate_keys=generate_keys,
)
if generate_keys:
return None
obj.invoke_all("read_arguments", args)
return obj
def read_config_files(self, config_files, keys_directory=None,
generate_keys=False):
if not keys_directory:
keys_directory = os.path.dirname(config_files[-1])
config_dir_path = os.path.abspath(keys_directory)
if config_args.keys_directory:
config_dir_path = config_args.keys_directory
else:
config_dir_path = os.path.dirname(config_args.config_path[-1])
config_dir_path = os.path.abspath(config_dir_path)
specified_config = {}
for config_file in config_files:
yaml_config = self.read_config_file(config_file)
yaml_config = cls.read_config_file(config_file)
specified_config.update(yaml_config)
if "server_name" not in specified_config:
raise ConfigError(MISSING_SERVER_NAME)
server_name = specified_config["server_name"]
_, config = self.generate_config(
_, config = obj.generate_config(
config_dir_path=config_dir_path,
server_name=server_name,
is_generating_file=False,
)
config.pop("log_config")
config.update(specified_config)
if "report_stats" not in config:
raise ConfigError(
MISSING_REPORT_STATS_CONFIG_INSTRUCTIONS + "\n" +
@@ -350,51 +328,11 @@ class Config(object):
)
if generate_keys:
self.invoke_all("generate_files", config)
obj.invoke_all("generate_files", config)
return
self.invoke_all("read_config", config)
obj.invoke_all("read_config", config)
obj.invoke_all("read_arguments", args)
def find_config_files(search_paths):
"""Finds config files using a list of search paths. If a path is a file
then that file path is added to the list. If a search path is a directory
then all the "*.yaml" files in that directory are added to the list in
sorted order.
Args:
search_paths(list(str)): A list of paths to search.
Returns:
list(str): A list of file paths.
"""
config_files = []
if search_paths:
for config_path in search_paths:
if os.path.isdir(config_path):
# We accept specifying directories as config paths, we search
# inside that directory for all files matching *.yaml, and then
# we apply them in *sorted* order.
files = []
for entry in os.listdir(config_path):
entry_path = os.path.join(config_path, entry)
if not os.path.isfile(entry_path):
print (
"Found subdirectory in config directory: %r. IGNORING."
) % (entry_path, )
continue
if not entry.endswith(".yaml"):
print (
"Found file in config directory that does not"
" end in '.yaml': %r. IGNORING."
) % (entry_path, )
continue
files.append(entry_path)
config_files.extend(sorted(files))
else:
config_files.append(config_path)
return config_files
return obj

View File

@@ -12,153 +12,16 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from ._base import Config, ConfigError
from synapse.appservice import ApplicationService
from synapse.types import UserID
import urllib
import yaml
import logging
logger = logging.getLogger(__name__)
from ._base import Config
class AppServiceConfig(Config):
def read_config(self, config):
self.app_service_config_files = config.get("app_service_config_files", [])
self.notify_appservices = config.get("notify_appservices", True)
def default_config(cls, **kwargs):
return """\
# A list of application service config file to use
app_service_config_files: []
"""
def load_appservices(hostname, config_files):
"""Returns a list of Application Services from the config files."""
if not isinstance(config_files, list):
logger.warning(
"Expected %s to be a list of AS config files.", config_files
)
return []
# Dicts of value -> filename
seen_as_tokens = {}
seen_ids = {}
appservices = []
for config_file in config_files:
try:
with open(config_file, 'r') as f:
appservice = _load_appservice(
hostname, yaml.load(f), config_file
)
if appservice.id in seen_ids:
raise ConfigError(
"Cannot reuse ID across application services: "
"%s (files: %s, %s)" % (
appservice.id, config_file, seen_ids[appservice.id],
)
)
seen_ids[appservice.id] = config_file
if appservice.token in seen_as_tokens:
raise ConfigError(
"Cannot reuse as_token across application services: "
"%s (files: %s, %s)" % (
appservice.token,
config_file,
seen_as_tokens[appservice.token],
)
)
seen_as_tokens[appservice.token] = config_file
logger.info("Loaded application service: %s", appservice)
appservices.append(appservice)
except Exception as e:
logger.error("Failed to load appservice from '%s'", config_file)
logger.exception(e)
raise
return appservices
def _load_appservice(hostname, as_info, config_filename):
required_string_fields = [
"id", "as_token", "hs_token", "sender_localpart"
]
for field in required_string_fields:
if not isinstance(as_info.get(field), basestring):
raise KeyError("Required string field: '%s' (%s)" % (
field, config_filename,
))
# 'url' must either be a string or explicitly null, not missing
# to avoid accidentally turning off push for ASes.
if (not isinstance(as_info.get("url"), basestring) and
as_info.get("url", "") is not None):
raise KeyError(
"Required string field or explicit null: 'url' (%s)" % (config_filename,)
)
localpart = as_info["sender_localpart"]
if urllib.quote(localpart) != localpart:
raise ValueError(
"sender_localpart needs characters which are not URL encoded."
)
user = UserID(localpart, hostname)
user_id = user.to_string()
# Rate limiting for users of this AS is on by default (excludes sender)
rate_limited = True
if isinstance(as_info.get("rate_limited"), bool):
rate_limited = as_info.get("rate_limited")
# namespace checks
if not isinstance(as_info.get("namespaces"), dict):
raise KeyError("Requires 'namespaces' object.")
for ns in ApplicationService.NS_LIST:
# specific namespaces are optional
if ns in as_info["namespaces"]:
# expect a list of dicts with exclusive and regex keys
for regex_obj in as_info["namespaces"][ns]:
if not isinstance(regex_obj, dict):
raise ValueError(
"Expected namespace entry in %s to be an object,"
" but got %s", ns, regex_obj
)
if not isinstance(regex_obj.get("regex"), basestring):
raise ValueError(
"Missing/bad type 'regex' key in %s", regex_obj
)
if not isinstance(regex_obj.get("exclusive"), bool):
raise ValueError(
"Missing/bad type 'exclusive' key in %s", regex_obj
)
# protocols check
protocols = as_info.get("protocols")
if protocols:
# Because strings are lists in python
if isinstance(protocols, str) or not isinstance(protocols, list):
raise KeyError("Optional 'protocols' must be a list if present.")
for p in protocols:
if not isinstance(p, str):
raise KeyError("Bad value for 'protocols' item")
if as_info["url"] is None:
logger.info(
"(%s) Explicitly empty 'url' provided. This application service"
" will not receive events or queries.",
config_filename,
)
return ApplicationService(
token=as_info["as_token"],
url=as_info["url"],
namespaces=as_info["namespaces"],
hs_token=as_info["hs_token"],
sender=user_id,
id=as_info["id"],
protocols=protocols,
rate_limited=rate_limited
)

View File

@@ -27,7 +27,6 @@ class CaptchaConfig(Config):
def default_config(self, **kwargs):
return """\
## Captcha ##
# See docs/CAPTCHA_SETUP for full details of configuring this.
# This Home Server's ReCAPTCHA public key.
recaptcha_public_key: "YOUR_PUBLIC_KEY"

View File

@@ -1,98 +0,0 @@
# -*- coding: utf-8 -*-
# Copyright 2015, 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# This file can't be called email.py because if it is, we cannot:
import email.utils
from ._base import Config
class EmailConfig(Config):
def read_config(self, config):
self.email_enable_notifs = False
email_config = config.get("email", {})
self.email_enable_notifs = email_config.get("enable_notifs", False)
if self.email_enable_notifs:
# make sure we can import the required deps
import jinja2
import bleach
# prevent unused warnings
jinja2
bleach
required = [
"smtp_host",
"smtp_port",
"notif_from",
"template_dir",
"notif_template_html",
"notif_template_text",
]
missing = []
for k in required:
if k not in email_config:
missing.append(k)
if (len(missing) > 0):
raise RuntimeError(
"email.enable_notifs is True but required keys are missing: %s" %
(", ".join(["email." + k for k in missing]),)
)
if config.get("public_baseurl") is None:
raise RuntimeError(
"email.enable_notifs is True but no public_baseurl is set"
)
self.email_smtp_host = email_config["smtp_host"]
self.email_smtp_port = email_config["smtp_port"]
self.email_notif_from = email_config["notif_from"]
self.email_template_dir = email_config["template_dir"]
self.email_notif_template_html = email_config["notif_template_html"]
self.email_notif_template_text = email_config["notif_template_text"]
self.email_notif_for_new_users = email_config.get(
"notif_for_new_users", True
)
if "app_name" in email_config:
self.email_app_name = email_config["app_name"]
else:
self.email_app_name = "Matrix"
# make sure it's valid
parsed = email.utils.parseaddr(self.email_notif_from)
if parsed[1] == '':
raise RuntimeError("Invalid notif_from address")
else:
self.email_enable_notifs = False
# Not much point setting defaults for the rest: it would be an
# error for them to be used.
def default_config(self, config_dir_path, server_name, **kwargs):
return """
# Enable sending emails for notification events
#email:
# enable_notifs: false
# smtp_host: "localhost"
# smtp_port: 25
# notif_from: "Your Friendly %(app)s Home Server <noreply@example.com>"
# app_name: Matrix
# template_dir: res/templates
# notif_template_html: notif_mail.html
# notif_template_text: notif_mail.txt
# notif_for_new_users: True
"""

View File

@@ -30,17 +30,14 @@ from .saml2 import SAML2Config
from .cas import CasConfig
from .password import PasswordConfig
from .jwt import JWTConfig
from .password_auth_providers import PasswordAuthProviderConfig
from .emailconfig import EmailConfig
from .workers import WorkerConfig
from .ldap import LDAPConfig
class HomeServerConfig(TlsConfig, ServerConfig, DatabaseConfig, LoggingConfig,
RatelimitConfig, ContentRepositoryConfig, CaptchaConfig,
VoipConfig, RegistrationConfig, MetricsConfig, ApiConfig,
AppServiceConfig, KeyConfig, SAML2Config, CasConfig,
JWTConfig, PasswordConfig, EmailConfig,
WorkerConfig, PasswordAuthProviderConfig,):
JWTConfig, LDAPConfig, PasswordConfig,):
pass

View File

@@ -13,16 +13,7 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from ._base import Config, ConfigError
MISSING_JWT = (
"""Missing jwt library. This is required for jwt login.
Install by running:
pip install pyjwt
"""
)
from ._base import Config
class JWTConfig(Config):
@@ -32,12 +23,6 @@ class JWTConfig(Config):
self.jwt_enabled = jwt_config.get("enabled", False)
self.jwt_secret = jwt_config["secret"]
self.jwt_algorithm = jwt_config["algorithm"]
try:
import jwt
jwt # To stop unused lint.
except ImportError:
raise ConfigError(MISSING_JWT)
else:
self.jwt_enabled = False
self.jwt_secret = None
@@ -45,8 +30,6 @@ class JWTConfig(Config):
def default_config(self, **kwargs):
return """\
# The JWT needs to contain a globally unique "sub" (subject) claim.
#
# jwt_config:
# enabled: true
# secret: "a secret"

View File

@@ -57,8 +57,6 @@ class KeyConfig(Config):
seed = self.signing_key[0].seed
self.macaroon_secret_key = hashlib.sha256(seed)
self.expire_access_token = config.get("expire_access_token", False)
def default_config(self, config_dir_path, server_name, is_generating_file=False,
**kwargs):
base_key_name = os.path.join(config_dir_path, server_name)
@@ -71,9 +69,6 @@ class KeyConfig(Config):
return """\
macaroon_secret_key: "%(macaroon_secret_key)s"
# Used to enable access token expiration.
expire_access_token: False
## Signing Keys ##
# Path to the signing key to sign messages with

52
synapse/config/ldap.py Normal file
View File

@@ -0,0 +1,52 @@
# -*- coding: utf-8 -*-
# Copyright 2015 Niklas Riekenbrauck
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from ._base import Config
class LDAPConfig(Config):
def read_config(self, config):
ldap_config = config.get("ldap_config", None)
if ldap_config:
self.ldap_enabled = ldap_config.get("enabled", False)
self.ldap_server = ldap_config["server"]
self.ldap_port = ldap_config["port"]
self.ldap_tls = ldap_config.get("tls", False)
self.ldap_search_base = ldap_config["search_base"]
self.ldap_search_property = ldap_config["search_property"]
self.ldap_email_property = ldap_config["email_property"]
self.ldap_full_name_property = ldap_config["full_name_property"]
else:
self.ldap_enabled = False
self.ldap_server = None
self.ldap_port = None
self.ldap_tls = False
self.ldap_search_base = None
self.ldap_search_property = None
self.ldap_email_property = None
self.ldap_full_name_property = None
def default_config(self, **kwargs):
return """\
# ldap_config:
# enabled: true
# server: "ldap://localhost"
# port: 389
# tls: false
# search_base: "ou=Users,dc=example,dc=com"
# search_property: "cn"
# email_property: "email"
# full_name_property: "givenName"
"""

View File

@@ -15,7 +15,7 @@
from ._base import Config
from synapse.util.logcontext import LoggingContextFilter
from twisted.logger import globalLogBeginner, STDLibLogObserver
from twisted.python.log import PythonLoggingObserver
import logging
import logging.config
import yaml
@@ -50,7 +50,6 @@ handlers:
console:
class: logging.StreamHandler
formatter: precise
filters: [context]
loggers:
synapse:
@@ -127,68 +126,54 @@ class LoggingConfig(Config):
)
def setup_logging(self):
setup_logging(self.log_config, self.log_file, self.verbosity)
log_format = (
"%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(request)s"
" - %(message)s"
)
if self.log_config is None:
level = logging.INFO
level_for_storage = logging.INFO
if self.verbosity:
level = logging.DEBUG
if self.verbosity > 1:
level_for_storage = logging.DEBUG
def setup_logging(log_config=None, log_file=None, verbosity=None):
log_format = (
"%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(request)s"
" - %(message)s"
)
if log_config is None:
# FIXME: we need a logging.WARN for a -q quiet option
logger = logging.getLogger('')
logger.setLevel(level)
level = logging.INFO
level_for_storage = logging.INFO
if verbosity:
level = logging.DEBUG
if verbosity > 1:
level_for_storage = logging.DEBUG
logging.getLogger('synapse.storage').setLevel(level_for_storage)
# FIXME: we need a logging.WARN for a -q quiet option
logger = logging.getLogger('')
logger.setLevel(level)
formatter = logging.Formatter(log_format)
if self.log_file:
# TODO: Customisable file size / backup count
handler = logging.handlers.RotatingFileHandler(
self.log_file, maxBytes=(1000 * 1000 * 100), backupCount=3
)
logging.getLogger('synapse.storage').setLevel(level_for_storage)
def sighup(signum, stack):
logger.info("Closing log file due to SIGHUP")
handler.doRollover()
logger.info("Opened new log file due to SIGHUP")
formatter = logging.Formatter(log_format)
if log_file:
# TODO: Customisable file size / backup count
handler = logging.handlers.RotatingFileHandler(
log_file, maxBytes=(1000 * 1000 * 100), backupCount=3
)
# TODO(paul): obviously this is a terrible mechanism for
# stealing SIGHUP, because it means no other part of synapse
# can use it instead. If we want to catch SIGHUP anywhere
# else as well, I'd suggest we find a nicer way to broadcast
# it around.
if getattr(signal, "SIGHUP"):
signal.signal(signal.SIGHUP, sighup)
else:
handler = logging.StreamHandler()
handler.setFormatter(formatter)
def sighup(signum, stack):
logger.info("Closing log file due to SIGHUP")
handler.doRollover()
logger.info("Opened new log file due to SIGHUP")
handler.addFilter(LoggingContextFilter(request=""))
# TODO(paul): obviously this is a terrible mechanism for
# stealing SIGHUP, because it means no other part of synapse
# can use it instead. If we want to catch SIGHUP anywhere
# else as well, I'd suggest we find a nicer way to broadcast
# it around.
if getattr(signal, "SIGHUP"):
signal.signal(signal.SIGHUP, sighup)
logger.addHandler(handler)
else:
handler = logging.StreamHandler()
handler.setFormatter(formatter)
with open(self.log_config, 'r') as f:
logging.config.dictConfig(yaml.load(f))
handler.addFilter(LoggingContextFilter(request=""))
logger.addHandler(handler)
else:
with open(log_config, 'r') as f:
logging.config.dictConfig(yaml.load(f))
# It's critical to point twisted's internal logging somewhere, otherwise it
# stacks up and leaks kup to 64K object;
# see: https://twistedmatrix.com/trac/ticket/8164
#
# Routing to the python logging framework could be a performance problem if
# the handlers blocked for a long time as python.logging is a blocking API
# see https://twistedmatrix.com/documents/current/core/howto/logger.html
# filed as https://github.com/matrix-org/synapse/issues/1727
#
# However this may not be too much of a problem if we are just writing to a file.
observer = STDLibLogObserver()
globalLogBeginner.beginLoggingTo([observer])
observer = PythonLoggingObserver()
observer.start()

View File

@@ -23,14 +23,10 @@ class PasswordConfig(Config):
def read_config(self, config):
password_config = config.get("password_config", {})
self.password_enabled = password_config.get("enabled", True)
self.password_pepper = password_config.get("pepper", "")
def default_config(self, config_dir_path, server_name, **kwargs):
return """
# Enable password for login.
password_config:
enabled: true
# Uncomment and change to a secret random string for extra security.
# DO NOT CHANGE THIS AFTER INITIAL SETUP!
#pepper: ""
"""

View File

@@ -1,72 +0,0 @@
# -*- coding: utf-8 -*-
# Copyright 2016 Openmarket
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from ._base import Config, ConfigError
import importlib
class PasswordAuthProviderConfig(Config):
def read_config(self, config):
self.password_providers = []
# We want to be backwards compatible with the old `ldap_config`
# param.
ldap_config = config.get("ldap_config", {})
self.ldap_enabled = ldap_config.get("enabled", False)
if self.ldap_enabled:
from ldap_auth_provider import LdapAuthProvider
parsed_config = LdapAuthProvider.parse_config(ldap_config)
self.password_providers.append((LdapAuthProvider, parsed_config))
providers = config.get("password_providers", [])
for provider in providers:
# This is for backwards compat when the ldap auth provider resided
# in this package.
if provider['module'] == "synapse.util.ldap_auth_provider.LdapAuthProvider":
from ldap_auth_provider import LdapAuthProvider
provider_class = LdapAuthProvider
else:
# We need to import the module, and then pick the class out of
# that, so we split based on the last dot.
module, clz = provider['module'].rsplit(".", 1)
module = importlib.import_module(module)
provider_class = getattr(module, clz)
try:
provider_config = provider_class.parse_config(provider["config"])
except Exception as e:
raise ConfigError(
"Failed to parse config for %r: %r" % (provider['module'], e)
)
self.password_providers.append((provider_class, provider_config))
def default_config(self, **kwargs):
return """\
# password_providers:
# - module: "ldap_auth_provider.LdapAuthProvider"
# config:
# enabled: true
# uri: "ldap://ldap.example.com:389"
# start_tls: true
# base: "ou=users,dc=example,dc=com"
# attributes:
# uid: "cn"
# mail: "email"
# name: "givenName"
# #bind_dn:
# #bind_password:
# #filter: "(objectClass=posixAccount)"
"""

View File

@@ -13,25 +13,9 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from ._base import Config, ConfigError
from ._base import Config
from collections import namedtuple
MISSING_NETADDR = (
"Missing netaddr library. This is required for URL preview API."
)
MISSING_LXML = (
"""Missing lxml library. This is required for URL preview API.
Install by running:
pip install lxml
Requires libxslt1-dev system package.
"""
)
ThumbnailRequirement = namedtuple(
"ThumbnailRequirement", ["width", "height", "method", "media_type"]
)
@@ -39,7 +23,7 @@ ThumbnailRequirement = namedtuple(
def parse_thumbnail_requirements(thumbnail_sizes):
""" Takes a list of dictionaries with "width", "height", and "method" keys
and creates a map from image media types to the thumbnail size, thumbnailing
and creates a map from image media types to the thumbnail size, thumnailing
method, and thumbnail media type to precalculate
Args:
@@ -69,44 +53,12 @@ class ContentRepositoryConfig(Config):
def read_config(self, config):
self.max_upload_size = self.parse_size(config["max_upload_size"])
self.max_image_pixels = self.parse_size(config["max_image_pixels"])
self.max_spider_size = self.parse_size(config["max_spider_size"])
self.media_store_path = self.ensure_directory(config["media_store_path"])
self.uploads_path = self.ensure_directory(config["uploads_path"])
self.dynamic_thumbnails = config["dynamic_thumbnails"]
self.thumbnail_requirements = parse_thumbnail_requirements(
config["thumbnail_sizes"]
)
self.url_preview_enabled = config.get("url_preview_enabled", False)
if self.url_preview_enabled:
try:
import lxml
lxml # To stop unused lint.
except ImportError:
raise ConfigError(MISSING_LXML)
try:
from netaddr import IPSet
except ImportError:
raise ConfigError(MISSING_NETADDR)
if "url_preview_ip_range_blacklist" in config:
self.url_preview_ip_range_blacklist = IPSet(
config["url_preview_ip_range_blacklist"]
)
else:
raise ConfigError(
"For security, you must specify an explicit target IP address "
"blacklist in url_preview_ip_range_blacklist for url previewing "
"to work"
)
self.url_preview_ip_range_whitelist = IPSet(
config.get("url_preview_ip_range_whitelist", ())
)
self.url_preview_url_blacklist = config.get(
"url_preview_url_blacklist", ()
)
def default_config(self, **kwargs):
media_store = self.default_path("media_store")
@@ -128,7 +80,7 @@ class ContentRepositoryConfig(Config):
# the resolution requested by the client. If true then whenever
# a new resolution is requested by the client the server will
# generate a new thumbnail. If false the server will pick a thumbnail
# from a precalculated list.
# from a precalcualted list.
dynamic_thumbnails: false
# List of thumbnail to precalculate when an image is uploaded.
@@ -148,73 +100,4 @@ class ContentRepositoryConfig(Config):
- width: 800
height: 600
method: scale
# Is the preview URL API enabled? If enabled, you *must* specify
# an explicit url_preview_ip_range_blacklist of IPs that the spider is
# denied from accessing.
url_preview_enabled: False
# List of IP address CIDR ranges that the URL preview spider is denied
# from accessing. There are no defaults: you must explicitly
# specify a list for URL previewing to work. You should specify any
# internal services in your network that you do not want synapse to try
# to connect to, otherwise anyone in any Matrix room could cause your
# synapse to issue arbitrary GET requests to your internal services,
# causing serious security issues.
#
# url_preview_ip_range_blacklist:
# - '127.0.0.0/8'
# - '10.0.0.0/8'
# - '172.16.0.0/12'
# - '192.168.0.0/16'
# - '100.64.0.0/10'
# - '169.254.0.0/16'
#
# List of IP address CIDR ranges that the URL preview spider is allowed
# to access even if they are specified in url_preview_ip_range_blacklist.
# This is useful for specifying exceptions to wide-ranging blacklisted
# target IP ranges - e.g. for enabling URL previews for a specific private
# website only visible in your network.
#
# url_preview_ip_range_whitelist:
# - '192.168.1.1'
# Optional list of URL matches that the URL preview spider is
# denied from accessing. You should use url_preview_ip_range_blacklist
# in preference to this, otherwise someone could define a public DNS
# entry that points to a private IP address and circumvent the blacklist.
# This is more useful if you know there is an entire shape of URL that
# you know that will never want synapse to try to spider.
#
# Each list entry is a dictionary of url component attributes as returned
# by urlparse.urlsplit as applied to the absolute form of the URL. See
# https://docs.python.org/2/library/urlparse.html#urlparse.urlsplit
# The values of the dictionary are treated as an filename match pattern
# applied to that component of URLs, unless they start with a ^ in which
# case they are treated as a regular expression match. If all the
# specified component matches for a given list item succeed, the URL is
# blacklisted.
#
# url_preview_url_blacklist:
# # blacklist any URL with a username in its URI
# - username: '*'
#
# # blacklist all *.google.com URLs
# - netloc: 'google.com'
# - netloc: '*.google.com'
#
# # blacklist all plain HTTP URLs
# - scheme: 'http'
#
# # blacklist http(s)://www.acme.com/foo
# - netloc: 'www.acme.com'
# path: '/foo'
#
# # blacklist any URL with a literal IPv4 address
# - netloc: '^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+$'
# The largest allowed URL preview spidering size in bytes
max_spider_size: "10M"
""" % locals()

View File

@@ -13,7 +13,7 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from ._base import Config, ConfigError
from ._base import Config
class ServerConfig(Config):
@@ -27,23 +27,10 @@ class ServerConfig(Config):
self.daemonize = config.get("daemonize")
self.print_pidfile = config.get("print_pidfile")
self.user_agent_suffix = config.get("user_agent_suffix")
self.use_frozen_dicts = config.get("use_frozen_dicts", False)
self.public_baseurl = config.get("public_baseurl")
# Whether to send federation traffic out in this process. This only
# applies to some federation traffic, and so shouldn't be used to
# "disable" federation
self.send_federation = config.get("send_federation", True)
if self.public_baseurl is not None:
if self.public_baseurl[-1] != '/':
self.public_baseurl += '/'
self.start_pushers = config.get("start_pushers", True)
self.use_frozen_dicts = config.get("use_frozen_dicts", True)
self.listeners = config.get("listeners", [])
self.gc_thresholds = read_gc_thresholds(config.get("gc_thresholds", None))
bind_port = config.get("bind_port")
if bind_port:
self.listeners = []
@@ -111,6 +98,26 @@ class ServerConfig(Config):
]
})
# Attempt to guess the content_addr for the v0 content repostitory
content_addr = config.get("content_addr")
if not content_addr:
for listener in self.listeners:
if listener["type"] == "http" and not listener.get("tls", False):
unsecure_port = listener["port"]
break
else:
raise RuntimeError("Could not determine 'content_addr'")
host = self.server_name
if ':' not in host:
host = "%s:%d" % (host, unsecure_port)
else:
host = host.split(':')[0]
host = "%s:%d" % (host, unsecure_port)
content_addr = "http://%s" % (host,)
self.content_addr = content_addr
def default_config(self, server_name, **kwargs):
if ":" in server_name:
bind_port = int(server_name.split(":")[1])
@@ -135,17 +142,11 @@ class ServerConfig(Config):
# Whether to serve a web client from the HTTP/HTTPS root resource.
web_client: True
# The public-facing base URL for the client API (not including _matrix/...)
# public_baseurl: https://example.com:8448/
# Set the soft limit on the number of file descriptors synapse can use
# Zero is used to indicate synapse should set the soft limit to the
# hard limit.
soft_file_limit: 0
# The GC threshold parameters to pass to `gc.set_threshold`, if defined
# gc_thresholds: [700, 10, 10]
# List of ports that Synapse should listen on, their purpose and their
# configuration.
listeners:
@@ -227,20 +228,3 @@ class ServerConfig(Config):
type=int,
help="Turn on the twisted telnet manhole"
" service on the given port.")
def read_gc_thresholds(thresholds):
"""Reads the three integer thresholds for garbage collection. Ensures that
the thresholds are integers if thresholds are supplied.
"""
if thresholds is None:
return None
try:
assert len(thresholds) == 3
return (
int(thresholds[0]), int(thresholds[1]), int(thresholds[2]),
)
except:
raise ConfigError(
"Value of `gc_threshold` must be a list of three integers if set"
)

View File

@@ -19,9 +19,6 @@ from OpenSSL import crypto
import subprocess
import os
from hashlib import sha256
from unpaddedbase64 import encode_base64
GENERATE_DH_PARAMS = False
@@ -45,19 +42,6 @@ class TlsConfig(Config):
config.get("tls_dh_params_path"), "tls_dh_params"
)
self.tls_fingerprints = config["tls_fingerprints"]
# Check that our own certificate is included in the list of fingerprints
# and include it if it is not.
x509_certificate_bytes = crypto.dump_certificate(
crypto.FILETYPE_ASN1,
self.tls_certificate
)
sha256_fingerprint = encode_base64(sha256(x509_certificate_bytes).digest())
sha256_fingerprints = set(f["sha256"] for f in self.tls_fingerprints)
if sha256_fingerprint not in sha256_fingerprints:
self.tls_fingerprints.append({u"sha256": sha256_fingerprint})
# This config option applies to non-federation HTTP clients
# (e.g. for talking to recaptcha, identity servers, and such)
# It should never be used in production, and is intended for
@@ -89,28 +73,6 @@ class TlsConfig(Config):
# Don't bind to the https port
no_tls: False
# List of allowed TLS fingerprints for this server to publish along
# with the signing keys for this server. Other matrix servers that
# make HTTPS requests to this server will check that the TLS
# certificates returned by this server match one of the fingerprints.
#
# Synapse automatically adds its the fingerprint of its own certificate
# to the list. So if federation traffic is handle directly by synapse
# then no modification to the list is required.
#
# If synapse is run behind a load balancer that handles the TLS then it
# will be necessary to add the fingerprints of the certificates used by
# the loadbalancers to this list if they are different to the one
# synapse is using.
#
# Homeservers are permitted to cache the list of TLS fingerprints
# returned in the key responses up to the "valid_until_ts" returned in
# key. It may be necessary to publish the fingerprints of a new
# certificate and wait until the "valid_until_ts" of the previous key
# responses have passed before deploying it.
tls_fingerprints: []
# tls_fingerprints: [{"sha256": "<base64_encoded_sha256_fingerprint>"}]
""" % locals()
def read_tls_certificate(self, cert_path):

View File

@@ -1,31 +0,0 @@
# -*- coding: utf-8 -*-
# Copyright 2016 matrix.org
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from ._base import Config
class WorkerConfig(Config):
"""The workers are processes run separately to the main synapse process.
They have their own pid_file and listener configuration. They use the
replication_url to talk to the main synapse process."""
def read_config(self, config):
self.worker_app = config.get("worker_app")
self.worker_listeners = config.get("worker_listeners")
self.worker_daemonize = config.get("worker_daemonize")
self.worker_pid_file = config.get("worker_pid_file")
self.worker_log_file = config.get("worker_log_file")
self.worker_log_config = config.get("worker_log_config")
self.worker_replication_url = config.get("worker_replication_url")

View File

@@ -77,12 +77,10 @@ class SynapseKeyClientProtocol(HTTPClient):
def __init__(self):
self.remote_key = defer.Deferred()
self.host = None
self._peer = None
def connectionMade(self):
self._peer = self.transport.getPeer()
logger.debug("Connected to %s", self._peer)
self.host = self.transport.getHost()
logger.debug("Connected to %s", self.host)
self.sendCommand(b"GET", self.path)
if self.host:
self.sendHeader(b"Host", self.host)
@@ -126,10 +124,7 @@ class SynapseKeyClientProtocol(HTTPClient):
self.timer.cancel()
def on_timeout(self):
logger.debug(
"Timeout waiting for response from %s: %s",
self.host, self._peer,
)
logger.debug("Timeout waiting for response from %s", self.host)
self.errback(IOError("Timeout waiting for response"))
self.transport.abortConnection()
@@ -138,5 +133,4 @@ class SynapseKeyClientFactory(Factory):
def protocol(self):
protocol = SynapseKeyClientProtocol()
protocol.path = self.path
protocol.host = self.host
return protocol

View File

@@ -22,7 +22,6 @@ from synapse.util.logcontext import (
preserve_context_over_deferred, preserve_context_over_fn, PreserveLoggingContext,
preserve_fn
)
from synapse.util.metrics import Measure
from twisted.internet import defer
@@ -45,25 +44,7 @@ import logging
logger = logging.getLogger(__name__)
VerifyKeyRequest = namedtuple("VerifyRequest", (
"server_name", "key_ids", "json_object", "deferred"
))
"""
A request for a verify key to verify a JSON object.
Attributes:
server_name(str): The name of the server to verify against.
key_ids(set(str)): The set of key_ids to that could be used to verify the
JSON object
json_object(dict): The JSON object to verify.
deferred(twisted.internet.defer.Deferred):
A deferred (server_name, key_id, verify_key) tuple that resolves when
a verify key has been fetched
"""
class KeyLookupError(ValueError):
pass
KeyGroup = namedtuple("KeyGroup", ("server_name", "group_id", "key_ids"))
class Keyring(object):
@@ -93,32 +74,39 @@ class Keyring(object):
list of deferreds indicating success or failure to verify each
json object's signature for the given server_name.
"""
verify_requests = []
group_id_to_json = {}
group_id_to_group = {}
group_ids = []
next_group_id = 0
deferreds = {}
for server_name, json_object in server_and_json:
logger.debug("Verifying for %s", server_name)
group_id = next_group_id
next_group_id += 1
group_ids.append(group_id)
key_ids = signature_ids(json_object, server_name)
if not key_ids:
deferred = defer.fail(SynapseError(
deferreds[group_id] = defer.fail(SynapseError(
400,
"Not signed with a supported algorithm",
Codes.UNAUTHORIZED,
))
else:
deferred = defer.Deferred()
deferreds[group_id] = defer.Deferred()
verify_request = VerifyKeyRequest(
server_name, key_ids, json_object, deferred
)
group = KeyGroup(server_name, group_id, key_ids)
verify_requests.append(verify_request)
group_id_to_group[group_id] = group
group_id_to_json[group_id] = json_object
@defer.inlineCallbacks
def handle_key_deferred(verify_request):
server_name = verify_request.server_name
def handle_key_deferred(group, deferred):
server_name = group.server_name
try:
_, key_id, verify_key = yield verify_request.deferred
_, _, key_id, verify_key = yield deferred
except IOError as e:
logger.warn(
"Got IOError when downloading keys for %s: %s %s",
@@ -140,7 +128,7 @@ class Keyring(object):
Codes.UNAUTHORIZED,
)
json_object = verify_request.json_object
json_object = group_id_to_json[group.group_id]
try:
verify_signed_json(json_object, server_name, verify_key)
@@ -169,34 +157,36 @@ class Keyring(object):
# Actually start fetching keys.
wait_on_deferred.addBoth(
lambda _: self.get_server_verify_keys(verify_requests)
lambda _: self.get_server_verify_keys(group_id_to_group, deferreds)
)
# When we've finished fetching all the keys for a given server_name,
# resolve the deferred passed to `wait_for_previous_lookups` so that
# any lookups waiting will proceed.
server_to_request_ids = {}
server_to_gids = {}
def remove_deferreds(res, server_name, verify_request):
request_id = id(verify_request)
server_to_request_ids[server_name].discard(request_id)
if not server_to_request_ids[server_name]:
def remove_deferreds(res, server_name, group_id):
server_to_gids[server_name].discard(group_id)
if not server_to_gids[server_name]:
d = server_to_deferred.pop(server_name, None)
if d:
d.callback(None)
return res
for verify_request in verify_requests:
server_name = verify_request.server_name
request_id = id(verify_request)
server_to_request_ids.setdefault(server_name, set()).add(request_id)
deferred.addBoth(remove_deferreds, server_name, verify_request)
for g_id, deferred in deferreds.items():
server_name = group_id_to_group[g_id].server_name
server_to_gids.setdefault(server_name, set()).add(g_id)
deferred.addBoth(remove_deferreds, server_name, g_id)
# Pass those keys to handle_key_deferred so that the json object
# signatures can be verified
return [
preserve_context_over_fn(handle_key_deferred, verify_request)
for verify_request in verify_requests
preserve_context_over_fn(
handle_key_deferred,
group_id_to_group[g_id],
deferreds[g_id],
)
for g_id in group_ids
]
@defer.inlineCallbacks
@@ -230,7 +220,7 @@ class Keyring(object):
d.addBoth(rm, server_name)
def get_server_verify_keys(self, verify_requests):
def get_server_verify_keys(self, group_id_to_group, group_id_to_deferred):
"""Takes a dict of KeyGroups and tries to find at least one key for
each group.
"""
@@ -244,79 +234,76 @@ class Keyring(object):
@defer.inlineCallbacks
def do_iterations():
with Measure(self.clock, "get_server_verify_keys"):
merged_results = {}
merged_results = {}
missing_keys = {}
for verify_request in verify_requests:
missing_keys.setdefault(verify_request.server_name, set()).update(
verify_request.key_ids
missing_keys = {}
for group in group_id_to_group.values():
missing_keys.setdefault(group.server_name, set()).update(
group.key_ids
)
for fn in key_fetch_fns:
results = yield fn(missing_keys.items())
merged_results.update(results)
# We now need to figure out which groups we have keys for
# and which we don't
missing_groups = {}
for group in group_id_to_group.values():
for key_id in group.key_ids:
if key_id in merged_results[group.server_name]:
with PreserveLoggingContext():
group_id_to_deferred[group.group_id].callback((
group.group_id,
group.server_name,
key_id,
merged_results[group.server_name][key_id],
))
break
else:
missing_groups.setdefault(
group.server_name, []
).append(group)
if not missing_groups:
break
missing_keys = {
server_name: set(
key_id for group in groups for key_id in group.key_ids
)
for server_name, groups in missing_groups.items()
}
for fn in key_fetch_fns:
results = yield fn(missing_keys.items())
merged_results.update(results)
# We now need to figure out which verify requests we have keys
# for and which we don't
missing_keys = {}
requests_missing_keys = []
for verify_request in verify_requests:
server_name = verify_request.server_name
result_keys = merged_results[server_name]
if verify_request.deferred.called:
# We've already called this deferred, which probably
# means that we've already found a key for it.
continue
for key_id in verify_request.key_ids:
if key_id in result_keys:
with PreserveLoggingContext():
verify_request.deferred.callback((
server_name,
key_id,
result_keys[key_id],
))
break
else:
# The else block is only reached if the loop above
# doesn't break.
missing_keys.setdefault(server_name, set()).update(
verify_request.key_ids
)
requests_missing_keys.append(verify_request)
if not missing_keys:
break
for verify_request in requests_missing_keys.values():
verify_request.deferred.errback(SynapseError(
401,
"No key for %s with id %s" % (
verify_request.server_name, verify_request.key_ids,
),
Codes.UNAUTHORIZED,
))
for group in missing_groups.values():
group_id_to_deferred[group.group_id].errback(SynapseError(
401,
"No key for %s with id %s" % (
group.server_name, group.key_ids,
),
Codes.UNAUTHORIZED,
))
def on_err(err):
for verify_request in verify_requests:
if not verify_request.deferred.called:
verify_request.deferred.errback(err)
for deferred in group_id_to_deferred.values():
if not deferred.called:
deferred.errback(err)
do_iterations().addErrback(on_err)
return group_id_to_deferred
@defer.inlineCallbacks
def get_keys_from_store(self, server_name_and_key_ids):
res = yield preserve_context_over_deferred(defer.gatherResults(
res = yield defer.gatherResults(
[
preserve_fn(self.store.get_server_verify_keys)(
self.store.get_server_verify_keys(
server_name, key_ids
).addCallback(lambda ks, server: (server, ks), server_name)
for server_name, key_ids in server_name_and_key_ids
],
consumeErrors=True,
)).addErrback(unwrapFirstError)
).addErrback(unwrapFirstError)
defer.returnValue(dict(res))
@@ -337,13 +324,13 @@ class Keyring(object):
)
defer.returnValue({})
results = yield preserve_context_over_deferred(defer.gatherResults(
results = yield defer.gatherResults(
[
preserve_fn(get_key)(p_name, p_keys)
get_key(p_name, p_keys)
for p_name, p_keys in self.perspective_servers.items()
],
consumeErrors=True,
)).addErrback(unwrapFirstError)
).addErrback(unwrapFirstError)
union_of_keys = {}
for result in results:
@@ -369,7 +356,7 @@ class Keyring(object):
)
except Exception as e:
logger.info(
"Unable to get key %r for %r directly: %s %s",
"Unable to getting key %r for %r directly: %s %s",
key_ids, server_name,
type(e).__name__, str(e.message),
)
@@ -383,13 +370,13 @@ class Keyring(object):
defer.returnValue(keys)
results = yield preserve_context_over_deferred(defer.gatherResults(
results = yield defer.gatherResults(
[
preserve_fn(get_key)(server_name, key_ids)
get_key(server_name, key_ids)
for server_name, key_ids in server_name_and_key_ids
],
consumeErrors=True,
)).addErrback(unwrapFirstError)
).addErrback(unwrapFirstError)
merged = {}
for result in results:
@@ -431,7 +418,7 @@ class Keyring(object):
for response in responses:
if (u"signatures" not in response
or perspective_name not in response[u"signatures"]):
raise KeyLookupError(
raise ValueError(
"Key response not signed by perspective server"
" %r" % (perspective_name,)
)
@@ -454,21 +441,21 @@ class Keyring(object):
list(response[u"signatures"][perspective_name]),
list(perspective_keys)
)
raise KeyLookupError(
raise ValueError(
"Response not signed with a known key for perspective"
" server %r" % (perspective_name,)
)
processed_response = yield self.process_v2_response(
perspective_name, response, only_from_server=False
perspective_name, response
)
for server_name, response_keys in processed_response.items():
keys.setdefault(server_name, {}).update(response_keys)
yield preserve_context_over_deferred(defer.gatherResults(
yield defer.gatherResults(
[
preserve_fn(self.store_keys)(
self.store_keys(
server_name=server_name,
from_server=perspective_name,
verify_keys=response_keys,
@@ -476,7 +463,7 @@ class Keyring(object):
for server_name, response_keys in keys.items()
],
consumeErrors=True
)).addErrback(unwrapFirstError)
).addErrback(unwrapFirstError)
defer.returnValue(keys)
@@ -497,10 +484,10 @@ class Keyring(object):
if (u"signatures" not in response
or server_name not in response[u"signatures"]):
raise KeyLookupError("Key response not signed by remote server")
raise ValueError("Key response not signed by remote server")
if "tls_fingerprints" not in response:
raise KeyLookupError("Key response missing TLS fingerprints")
raise ValueError("Key response missing TLS fingerprints")
certificate_bytes = crypto.dump_certificate(
crypto.FILETYPE_ASN1, tls_certificate
@@ -514,7 +501,7 @@ class Keyring(object):
response_sha256_fingerprints.add(fingerprint[u"sha256"])
if sha256_fingerprint_b64 not in response_sha256_fingerprints:
raise KeyLookupError("TLS certificate not allowed by fingerprints")
raise ValueError("TLS certificate not allowed by fingerprints")
response_keys = yield self.process_v2_response(
from_server=server_name,
@@ -524,7 +511,7 @@ class Keyring(object):
keys.update(response_keys)
yield preserve_context_over_deferred(defer.gatherResults(
yield defer.gatherResults(
[
preserve_fn(self.store_keys)(
server_name=key_server_name,
@@ -534,13 +521,13 @@ class Keyring(object):
for key_server_name, verify_keys in keys.items()
],
consumeErrors=True
)).addErrback(unwrapFirstError)
).addErrback(unwrapFirstError)
defer.returnValue(keys)
@defer.inlineCallbacks
def process_v2_response(self, from_server, response_json,
requested_ids=[], only_from_server=True):
requested_ids=[]):
time_now_ms = self.clock.time_msec()
response_keys = {}
verify_keys = {}
@@ -564,16 +551,9 @@ class Keyring(object):
results = {}
server_name = response_json["server_name"]
if only_from_server:
if server_name != from_server:
raise KeyLookupError(
"Expected a response for server %r not %r" % (
from_server, server_name
)
)
for key_id in response_json["signatures"].get(server_name, {}):
if key_id not in response_json["verify_keys"]:
raise KeyLookupError(
raise ValueError(
"Key response must include verification keys for all"
" signatures"
)
@@ -600,7 +580,7 @@ class Keyring(object):
response_keys.update(verify_keys)
response_keys.update(old_verify_keys)
yield preserve_context_over_deferred(defer.gatherResults(
yield defer.gatherResults(
[
preserve_fn(self.store.store_server_keys_json)(
server_name=server_name,
@@ -613,7 +593,7 @@ class Keyring(object):
for key_id in updated_key_ids
],
consumeErrors=True,
)).addErrback(unwrapFirstError)
).addErrback(unwrapFirstError)
results[server_name] = response_keys
@@ -641,15 +621,15 @@ class Keyring(object):
if ("signatures" not in response
or server_name not in response["signatures"]):
raise KeyLookupError("Key response not signed by remote server")
raise ValueError("Key response not signed by remote server")
if "tls_certificate" not in response:
raise KeyLookupError("Key response missing TLS certificate")
raise ValueError("Key response missing TLS certificate")
tls_certificate_b64 = response["tls_certificate"]
if encode_base64(x509_certificate_bytes) != tls_certificate_b64:
raise KeyLookupError("TLS certificate doesn't match")
raise ValueError("TLS certificate doesn't match")
# Cache the result in the datastore.
@@ -665,7 +645,7 @@ class Keyring(object):
for key_id in response["signatures"][server_name]:
if key_id not in response["verify_keys"]:
raise KeyLookupError(
raise ValueError(
"Key response must include verification keys for all"
" signatures"
)
@@ -702,7 +682,7 @@ class Keyring(object):
A deferred that completes when the keys are stored.
"""
# TODO(markjh): Store whether the keys have expired.
yield preserve_context_over_deferred(defer.gatherResults(
yield defer.gatherResults(
[
preserve_fn(self.store.store_server_verify_key)(
server_name, server_name, key.time_added, key
@@ -710,4 +690,4 @@ class Keyring(object):
for key_id, key in verify_keys.items()
],
consumeErrors=True,
)).addErrback(unwrapFirstError)
).addErrback(unwrapFirstError)

View File

@@ -99,7 +99,7 @@ class EventBase(object):
return d
def get(self, key, default=None):
def get(self, key, default):
return self._event_dict.get(key, default)
def get_internal_metadata_dict(self):

View File

@@ -15,30 +15,9 @@
class EventContext(object):
__slots__ = [
"current_state_ids",
"prev_state_ids",
"state_group",
"rejected",
"push_actions",
"prev_group",
"delta_ids",
"prev_state_events",
]
def __init__(self):
# The current state including the current event
self.current_state_ids = None
# The current state excluding the current event
self.prev_state_ids = None
def __init__(self, current_state=None):
self.current_state = current_state
self.state_group = None
self.rejected = False
self.push_actions = []
# A previously persisted state group and a delta between that
# and this state.
self.prev_group = None
self.delta_ids = None
self.prev_state_events = None

View File

@@ -16,17 +16,6 @@
from synapse.api.constants import EventTypes
from . import EventBase
from frozendict import frozendict
import re
# Split strings on "." but not "\." This uses a negative lookbehind assertion for '\'
# (?<!stuff) matches if the current position in the string is not preceded
# by a match for 'stuff'.
# TODO: This is fast, but fails to handle "foo\\.bar" which should be treated as
# the literal fields "foo\" and "bar" but will instead be treated as "foo\\.bar"
SPLIT_FIELD_REGEX = re.compile(r'(?<!\\)\.')
def prune_event(event):
""" Returns a pruned version of the given event, which removes all keys we
@@ -99,8 +88,6 @@ def prune_event(event):
if "age_ts" in event.unsigned:
allowed_fields["unsigned"]["age_ts"] = event.unsigned["age_ts"]
if "replaces_state" in event.unsigned:
allowed_fields["unsigned"]["replaces_state"] = event.unsigned["replaces_state"]
return type(event)(
allowed_fields,
@@ -108,83 +95,6 @@ def prune_event(event):
)
def _copy_field(src, dst, field):
"""Copy the field in 'src' to 'dst'.
For example, if src={"foo":{"bar":5}} and dst={}, and field=["foo","bar"]
then dst={"foo":{"bar":5}}.
Args:
src(dict): The dict to read from.
dst(dict): The dict to modify.
field(list<str>): List of keys to drill down to in 'src'.
"""
if len(field) == 0: # this should be impossible
return
if len(field) == 1: # common case e.g. 'origin_server_ts'
if field[0] in src:
dst[field[0]] = src[field[0]]
return
# Else is a nested field e.g. 'content.body'
# Pop the last field as that's the key to move across and we need the
# parent dict in order to access the data. Drill down to the right dict.
key_to_move = field.pop(-1)
sub_dict = src
for sub_field in field: # e.g. sub_field => "content"
if sub_field in sub_dict and type(sub_dict[sub_field]) in [dict, frozendict]:
sub_dict = sub_dict[sub_field]
else:
return
if key_to_move not in sub_dict:
return
# Insert the key into the output dictionary, creating nested objects
# as required. We couldn't do this any earlier or else we'd need to delete
# the empty objects if the key didn't exist.
sub_out_dict = dst
for sub_field in field:
sub_out_dict = sub_out_dict.setdefault(sub_field, {})
sub_out_dict[key_to_move] = sub_dict[key_to_move]
def only_fields(dictionary, fields):
"""Return a new dict with only the fields in 'dictionary' which are present
in 'fields'.
If there are no event fields specified then all fields are included.
The entries may include '.' charaters to indicate sub-fields.
So ['content.body'] will include the 'body' field of the 'content' object.
A literal '.' character in a field name may be escaped using a '\'.
Args:
dictionary(dict): The dictionary to read from.
fields(list<str>): A list of fields to copy over. Only shallow refs are
taken.
Returns:
dict: A new dictionary with only the given fields. If fields was empty,
the same dictionary is returned.
"""
if len(fields) == 0:
return dictionary
# for each field, convert it:
# ["content.body.thing\.with\.dots"] => [["content", "body", "thing\.with\.dots"]]
split_fields = [SPLIT_FIELD_REGEX.split(f) for f in fields]
# for each element of the output array of arrays:
# remove escaping so we can use the right key names.
split_fields[:] = [
[f.replace(r'\.', r'.') for f in field_array] for field_array in split_fields
]
output = {}
for field_array in split_fields:
_copy_field(dictionary, output, field_array)
return output
def format_event_raw(d):
return d
@@ -225,7 +135,7 @@ def format_event_for_client_v2_without_room_id(d):
def serialize_event(e, time_now_ms, as_client_event=True,
event_format=format_event_for_client_v1,
token_id=None, only_event_fields=None):
token_id=None):
# FIXME(erikj): To handle the case of presence events and the like
if not isinstance(e, EventBase):
return e
@@ -252,12 +162,6 @@ def serialize_event(e, time_now_ms, as_client_event=True,
d["unsigned"]["transaction_id"] = txn_id
if as_client_event:
d = event_format(d)
if only_event_fields:
if (not isinstance(only_event_fields, list) or
not all(isinstance(f, basestring) for f in only_event_fields)):
raise TypeError("only_event_fields must be a list of strings")
d = only_fields(d, only_event_fields)
return d
return event_format(d)
else:
return d

View File

@@ -17,9 +17,10 @@
"""
from .replication import ReplicationLayer
from .transport.client import TransportLayerClient
def initialize_http_replication(hs):
transport = hs.get_federation_transport_client()
def initialize_http_replication(homeserver):
transport = TransportLayerClient(homeserver)
return ReplicationLayer(hs, transport)
return ReplicationLayer(homeserver, transport)

View File

@@ -23,7 +23,6 @@ from synapse.crypto.event_signing import check_event_content_hash
from synapse.api.errors import SynapseError
from synapse.util import unwrapFirstError
from synapse.util.logcontext import preserve_fn, preserve_context_over_deferred
import logging
@@ -32,9 +31,6 @@ logger = logging.getLogger(__name__)
class FederationBase(object):
def __init__(self, hs):
pass
@defer.inlineCallbacks
def _check_sigs_and_hash_and_fetch(self, origin, pdus, outlier=False,
include_none=False):
@@ -103,10 +99,10 @@ class FederationBase(object):
warn, pdu
)
valid_pdus = yield preserve_context_over_deferred(defer.gatherResults(
valid_pdus = yield defer.gatherResults(
deferreds,
consumeErrors=True
)).addErrback(unwrapFirstError)
).addErrback(unwrapFirstError)
if include_none:
defer.returnValue(valid_pdus)
@@ -130,7 +126,7 @@ class FederationBase(object):
for pdu in pdus
]
deferreds = preserve_fn(self.keyring.verify_json_objects_for_server)([
deferreds = self.keyring.verify_json_objects_for_server([
(p.origin, p.get_pdu_json())
for p in redacted_pdus
])

View File

@@ -18,6 +18,7 @@ from twisted.internet import defer
from .federation_base import FederationBase
from synapse.api.constants import Membership
from .units import Edu
from synapse.api.errors import (
CodeMessageException, HttpResponseException, SynapseError,
@@ -25,7 +26,6 @@ from synapse.api.errors import (
from synapse.util import unwrapFirstError
from synapse.util.caches.expiringcache import ExpiringCache
from synapse.util.logutils import log_function
from synapse.util.logcontext import preserve_fn, preserve_context_over_deferred
from synapse.events import FrozenEvent
import synapse.metrics
@@ -43,37 +43,14 @@ logger = logging.getLogger(__name__)
# synapse.federation.federation_client is a silly name
metrics = synapse.metrics.get_metrics_for("synapse.federation.client")
sent_pdus_destination_dist = metrics.register_distribution("sent_pdu_destinations")
sent_edus_counter = metrics.register_counter("sent_edus")
sent_queries_counter = metrics.register_counter("sent_queries", labels=["type"])
PDU_RETRY_TIME_MS = 1 * 60 * 1000
class FederationClient(FederationBase):
def __init__(self, hs):
super(FederationClient, self).__init__(hs)
self.pdu_destination_tried = {}
self._clock.looping_call(
self._clear_tried_cache, 60 * 1000,
)
self.state = hs.get_state_handler()
def _clear_tried_cache(self):
"""Clear pdu_destination_tried cache"""
now = self._clock.time_msec()
old_dict = self.pdu_destination_tried
self.pdu_destination_tried = {}
for event_id, destination_dict in old_dict.items():
destination_dict = {
dest: time
for dest, time in destination_dict.items()
if time + PDU_RETRY_TIME_MS > now
}
if destination_dict:
self.pdu_destination_tried[event_id] = destination_dict
def start_get_pdu_cache(self):
self._get_pdu_cache = ExpiringCache(
@@ -86,6 +63,55 @@ class FederationClient(FederationBase):
self._get_pdu_cache.start()
@log_function
def send_pdu(self, pdu, destinations):
"""Informs the replication layer about a new PDU generated within the
home server that should be transmitted to others.
TODO: Figure out when we should actually resolve the deferred.
Args:
pdu (Pdu): The new Pdu.
Returns:
Deferred: Completes when we have successfully processed the PDU
and replicated it to any interested remote home servers.
"""
order = self._order
self._order += 1
sent_pdus_destination_dist.inc_by(len(destinations))
logger.debug("[%s] transaction_layer.enqueue_pdu... ", pdu.event_id)
# TODO, add errback, etc.
self._transaction_queue.enqueue_pdu(pdu, destinations, order)
logger.debug(
"[%s] transaction_layer.enqueue_pdu... done",
pdu.event_id
)
@log_function
def send_edu(self, destination, edu_type, content):
edu = Edu(
origin=self.server_name,
destination=destination,
edu_type=edu_type,
content=content,
)
sent_edus_counter.inc()
# TODO, add errback, etc.
self._transaction_queue.enqueue_edu(edu)
return defer.succeed(None)
@log_function
def send_failure(self, failure, destination):
self._transaction_queue.enqueue_failure(failure, destination)
return defer.succeed(None)
@log_function
def make_query(self, destination, query_type, args,
retry_on_dns_fail=False):
@@ -110,7 +136,7 @@ class FederationClient(FederationBase):
)
@log_function
def query_client_keys(self, destination, content, timeout):
def query_client_keys(self, destination, content):
"""Query device keys for a device hosted on a remote server.
Args:
@@ -122,12 +148,10 @@ class FederationClient(FederationBase):
response
"""
sent_queries_counter.inc("client_device_keys")
return self.transport_layer.query_client_keys(
destination, content, timeout
)
return self.transport_layer.query_client_keys(destination, content)
@log_function
def claim_client_keys(self, destination, content, timeout):
def claim_client_keys(self, destination, content):
"""Claims one-time keys for a device hosted on a remote server.
Args:
@@ -139,9 +163,7 @@ class FederationClient(FederationBase):
response
"""
sent_queries_counter.inc("client_one_time_keys")
return self.transport_layer.claim_client_keys(
destination, content, timeout
)
return self.transport_layer.claim_client_keys(destination, content)
@defer.inlineCallbacks
@log_function
@@ -176,10 +198,10 @@ class FederationClient(FederationBase):
]
# FIXME: We should handle signature failures more gracefully.
pdus[:] = yield preserve_context_over_deferred(defer.gatherResults(
pdus[:] = yield defer.gatherResults(
self._check_sigs_and_hashes(pdus),
consumeErrors=True,
)).addErrback(unwrapFirstError)
).addErrback(unwrapFirstError)
defer.returnValue(pdus)
@@ -211,19 +233,12 @@ class FederationClient(FederationBase):
# TODO: Rate limit the number of times we try and get the same event.
if self._get_pdu_cache:
ev = self._get_pdu_cache.get(event_id)
if ev:
defer.returnValue(ev)
e = self._get_pdu_cache.get(event_id)
if e:
defer.returnValue(e)
pdu_attempts = self.pdu_destination_tried.setdefault(event_id, {})
signed_pdu = None
pdu = None
for destination in destinations:
now = self._clock.time_msec()
last_attempt = pdu_attempts.get(destination, 0)
if last_attempt + PDU_RETRY_TIME_MS > now:
continue
try:
limiter = yield get_retry_limiter(
destination,
@@ -247,33 +262,39 @@ class FederationClient(FederationBase):
pdu = pdu_list[0]
# Check signatures are correct.
signed_pdu = yield self._check_sigs_and_hashes([pdu])[0]
pdu = yield self._check_sigs_and_hashes([pdu])[0]
break
pdu_attempts[destination] = now
except SynapseError as e:
except SynapseError:
logger.info(
"Failed to get PDU %s from %s because %s",
event_id, destination, e,
)
continue
except CodeMessageException as e:
if 400 <= e.code < 500:
raise
logger.info(
"Failed to get PDU %s from %s because %s",
event_id, destination, e,
)
continue
except NotRetryingDestination as e:
logger.info(e.message)
continue
except Exception as e:
pdu_attempts[destination] = now
logger.info(
"Failed to get PDU %s from %s because %s",
event_id, destination, e,
)
continue
if self._get_pdu_cache is not None and signed_pdu:
self._get_pdu_cache[event_id] = signed_pdu
if self._get_pdu_cache is not None and pdu:
self._get_pdu_cache[event_id] = pdu
defer.returnValue(signed_pdu)
defer.returnValue(pdu)
@defer.inlineCallbacks
@log_function
@@ -290,42 +311,6 @@ class FederationClient(FederationBase):
Deferred: Results in a list of PDUs.
"""
try:
# First we try and ask for just the IDs, as thats far quicker if
# we have most of the state and auth_chain already.
# However, this may 404 if the other side has an old synapse.
result = yield self.transport_layer.get_room_state_ids(
destination, room_id, event_id=event_id,
)
state_event_ids = result["pdu_ids"]
auth_event_ids = result.get("auth_chain_ids", [])
fetched_events, failed_to_fetch = yield self.get_events(
[destination], room_id, set(state_event_ids + auth_event_ids)
)
if failed_to_fetch:
logger.warn("Failed to get %r", failed_to_fetch)
event_map = {
ev.event_id: ev for ev in fetched_events
}
pdus = [event_map[e_id] for e_id in state_event_ids if e_id in event_map]
auth_chain = [
event_map[e_id] for e_id in auth_event_ids if e_id in event_map
]
auth_chain.sort(key=lambda e: e.depth)
defer.returnValue((pdus, auth_chain))
except HttpResponseException as e:
if e.code == 400 or e.code == 404:
logger.info("Failed to use get_room_state_ids API, falling back")
else:
raise e
result = yield self.transport_layer.get_room_state(
destination, room_id, event_id=event_id,
)
@@ -339,95 +324,18 @@ class FederationClient(FederationBase):
for p in result.get("auth_chain", [])
]
seen_events = yield self.store.get_events([
ev.event_id for ev in itertools.chain(pdus, auth_chain)
])
signed_pdus = yield self._check_sigs_and_hash_and_fetch(
destination,
[p for p in pdus if p.event_id not in seen_events],
outlier=True
)
signed_pdus.extend(
seen_events[p.event_id] for p in pdus if p.event_id in seen_events
destination, pdus, outlier=True
)
signed_auth = yield self._check_sigs_and_hash_and_fetch(
destination,
[p for p in auth_chain if p.event_id not in seen_events],
outlier=True
)
signed_auth.extend(
seen_events[p.event_id] for p in auth_chain if p.event_id in seen_events
destination, auth_chain, outlier=True
)
signed_auth.sort(key=lambda e: e.depth)
defer.returnValue((signed_pdus, signed_auth))
@defer.inlineCallbacks
def get_events(self, destinations, room_id, event_ids, return_local=True):
"""Fetch events from some remote destinations, checking if we already
have them.
Args:
destinations (list)
room_id (str)
event_ids (list)
return_local (bool): Whether to include events we already have in
the DB in the returned list of events
Returns:
Deferred: A deferred resolving to a 2-tuple where the first is a list of
events and the second is a list of event ids that we failed to fetch.
"""
if return_local:
seen_events = yield self.store.get_events(event_ids, allow_rejected=True)
signed_events = seen_events.values()
else:
seen_events = yield self.store.have_events(event_ids)
signed_events = []
failed_to_fetch = set()
missing_events = set(event_ids)
for k in seen_events:
missing_events.discard(k)
if not missing_events:
defer.returnValue((signed_events, failed_to_fetch))
def random_server_list():
srvs = list(destinations)
random.shuffle(srvs)
return srvs
batch_size = 20
missing_events = list(missing_events)
for i in xrange(0, len(missing_events), batch_size):
batch = set(missing_events[i:i + batch_size])
deferreds = [
preserve_fn(self.get_pdu)(
destinations=random_server_list(),
event_id=e_id,
)
for e_id in batch
]
res = yield preserve_context_over_deferred(
defer.DeferredList(deferreds, consumeErrors=True)
)
for success, result in res:
if success and result:
signed_events.append(result)
batch.discard(result.event_id)
# We removed all events we successfully fetched from `batch`
failed_to_fetch.update(batch)
defer.returnValue((signed_events, failed_to_fetch))
@defer.inlineCallbacks
@log_function
def get_event_auth(self, destination, room_id, event_id):
@@ -503,19 +411,14 @@ class FederationClient(FederationBase):
(destination, self.event_from_pdu_json(pdu_dict))
)
break
except CodeMessageException as e:
if not 500 <= e.code < 600:
raise
else:
logger.warn(
"Failed to make_%s via %s: %s",
membership, destination, e.message
)
except CodeMessageException:
raise
except Exception as e:
logger.warn(
"Failed to make_%s via %s: %s",
membership, destination, e.message
)
raise
raise RuntimeError("Failed to send to any server.")
@@ -587,14 +490,8 @@ class FederationClient(FederationBase):
"auth_chain": signed_auth,
"origin": destination,
})
except CodeMessageException as e:
if not 500 <= e.code < 600:
raise
else:
logger.exception(
"Failed to send_join via %s: %s",
destination, e.message
)
except CodeMessageException:
raise
except Exception as e:
logger.exception(
"Failed to send_join via %s: %s",
@@ -653,18 +550,6 @@ class FederationClient(FederationBase):
raise RuntimeError("Failed to send to any server.")
def get_public_rooms(self, destination, limit=None, since_token=None,
search_filter=None, include_all_networks=False,
third_party_instance_id=None):
if destination == self.server_name:
return
return self.transport_layer.get_public_rooms(
destination, limit, since_token, search_filter,
include_all_networks=include_all_networks,
third_party_instance_id=third_party_instance_id,
)
@defer.inlineCallbacks
def query_auth(self, destination, room_id, event_id, local_auth):
"""
@@ -740,6 +625,8 @@ class FederationClient(FederationBase):
signed_events = yield self._check_sigs_and_hash_and_fetch(
destination, events, outlier=False
)
have_gotten_all_from_destination = True
except HttpResponseException as e:
if not e.code == 400:
raise
@@ -747,6 +634,69 @@ class FederationClient(FederationBase):
# We are probably hitting an old server that doesn't support
# get_missing_events
signed_events = []
have_gotten_all_from_destination = False
if len(signed_events) >= limit:
defer.returnValue(signed_events)
servers = yield self.store.get_joined_hosts_for_room(room_id)
servers = set(servers)
servers.discard(self.server_name)
failed_to_fetch = set()
while len(signed_events) < limit:
# Are we missing any?
seen_events = set(earliest_events_ids)
seen_events.update(e.event_id for e in signed_events if e)
missing_events = {}
for e in itertools.chain(latest_events, signed_events):
if e.depth > min_depth:
missing_events.update({
e_id: e.depth for e_id, _ in e.prev_events
if e_id not in seen_events
and e_id not in failed_to_fetch
})
if not missing_events:
break
have_seen = yield self.store.have_events(missing_events)
for k in have_seen:
missing_events.pop(k, None)
if not missing_events:
break
# Okay, we haven't gotten everything yet. Lets get them.
ordered_missing = sorted(missing_events.items(), key=lambda x: x[0])
if have_gotten_all_from_destination:
servers.discard(destination)
def random_server_list():
srvs = list(servers)
random.shuffle(srvs)
return srvs
deferreds = [
self.get_pdu(
destinations=random_server_list(),
event_id=e_id,
)
for e_id, depth in ordered_missing[:limit - len(signed_events)]
]
res = yield defer.DeferredList(deferreds, consumeErrors=True)
for (result, val), (e_id, _) in zip(res, ordered_missing):
if result and val:
signed_events.append(val)
else:
failed_to_fetch.add(e_id)
defer.returnValue(signed_events)

View File

@@ -19,13 +19,11 @@ from twisted.internet import defer
from .federation_base import FederationBase
from .units import Transaction, Edu
from synapse.util.async import Linearizer
from synapse.util.logutils import log_function
from synapse.util.caches.response_cache import ResponseCache
from synapse.events import FrozenEvent
import synapse.metrics
from synapse.api.errors import AuthError, FederationError, SynapseError
from synapse.api.errors import FederationError, SynapseError
from synapse.crypto.event_signing import compute_event_signature
@@ -46,18 +44,6 @@ received_queries_counter = metrics.register_counter("received_queries", labels=[
class FederationServer(FederationBase):
def __init__(self, hs):
super(FederationServer, self).__init__(hs)
self.auth = hs.get_auth()
self._room_pdu_linearizer = Linearizer()
self._server_linearizer = Linearizer()
# We cache responses to state queries, as they take a while and often
# come in waves.
self._state_resp_cache = ResponseCache(hs, timeout_ms=30000)
def set_handler(self, handler):
"""Sets the handler that the replication layer will use to communicate
receipt of new PDUs from other home servers. The required methods are
@@ -97,14 +83,11 @@ class FederationServer(FederationBase):
@defer.inlineCallbacks
@log_function
def on_backfill_request(self, origin, room_id, versions, limit):
with (yield self._server_linearizer.queue((origin, room_id))):
pdus = yield self.handler.on_backfill_request(
origin, room_id, versions, limit
)
pdus = yield self.handler.on_backfill_request(
origin, room_id, versions, limit
)
res = self._transaction_from_pdus(pdus).get_dict()
defer.returnValue((200, res))
defer.returnValue((200, self._transaction_from_pdus(pdus).get_dict()))
@defer.inlineCallbacks
@log_function
@@ -188,64 +171,22 @@ class FederationServer(FederationBase):
except SynapseError as e:
logger.info("Failed to handle edu %r: %r", edu_type, e)
except Exception as e:
logger.exception("Failed to handle edu %r", edu_type)
logger.exception("Failed to handle edu %r", edu_type, e)
else:
logger.warn("Received EDU of type %s with no handler", edu_type)
@defer.inlineCallbacks
@log_function
def on_context_state_request(self, origin, room_id, event_id):
if not event_id:
raise NotImplementedError("Specify an event")
if event_id:
pdus = yield self.handler.get_state_for_pdu(
origin, room_id, event_id,
)
auth_chain = yield self.store.get_auth_chain(
[pdu.event_id for pdu in pdus]
)
in_room = yield self.auth.check_host_in_room(room_id, origin)
if not in_room:
raise AuthError(403, "Host not in room.")
result = self._state_resp_cache.get((room_id, event_id))
if not result:
with (yield self._server_linearizer.queue((origin, room_id))):
resp = yield self._state_resp_cache.set(
(room_id, event_id),
self._on_context_state_request_compute(room_id, event_id)
)
else:
resp = yield result
defer.returnValue((200, resp))
@defer.inlineCallbacks
def on_state_ids_request(self, origin, room_id, event_id):
if not event_id:
raise NotImplementedError("Specify an event")
in_room = yield self.auth.check_host_in_room(room_id, origin)
if not in_room:
raise AuthError(403, "Host not in room.")
state_ids = yield self.handler.get_state_ids_for_pdu(
room_id, event_id,
)
auth_chain_ids = yield self.store.get_auth_chain_ids(state_ids)
defer.returnValue((200, {
"pdu_ids": state_ids,
"auth_chain_ids": auth_chain_ids,
}))
@defer.inlineCallbacks
def _on_context_state_request_compute(self, room_id, event_id):
pdus = yield self.handler.get_state_for_pdu(
room_id, event_id,
)
auth_chain = yield self.store.get_auth_chain(
[pdu.event_id for pdu in pdus]
)
for event in auth_chain:
# We sign these again because there was a bug where we
# incorrectly signed things the first time round
if self.hs.is_mine_id(event.event_id):
for event in auth_chain:
event.signatures.update(
compute_event_signature(
event,
@@ -253,11 +194,13 @@ class FederationServer(FederationBase):
self.hs.config.signing_key[0]
)
)
else:
raise NotImplementedError("Specify an event")
defer.returnValue({
defer.returnValue((200, {
"pdus": [pdu.get_pdu_json() for pdu in pdus],
"auth_chain": [pdu.get_pdu_json() for pdu in auth_chain],
})
}))
@defer.inlineCallbacks
@log_function
@@ -331,16 +274,14 @@ class FederationServer(FederationBase):
@defer.inlineCallbacks
def on_event_auth(self, origin, room_id, event_id):
with (yield self._server_linearizer.queue((origin, room_id))):
time_now = self._clock.time_msec()
auth_pdus = yield self.handler.on_event_auth(event_id)
res = {
"auth_chain": [a.get_pdu_json(time_now) for a in auth_pdus],
}
defer.returnValue((200, res))
time_now = self._clock.time_msec()
auth_pdus = yield self.handler.on_event_auth(event_id)
defer.returnValue((200, {
"auth_chain": [a.get_pdu_json(time_now) for a in auth_pdus],
}))
@defer.inlineCallbacks
def on_query_auth_request(self, origin, content, room_id, event_id):
def on_query_auth_request(self, origin, content, event_id):
"""
Content is a dict with keys::
auth_chain (list): A list of events that give the auth chain.
@@ -359,41 +300,58 @@ class FederationServer(FederationBase):
Returns:
Deferred: Results in `dict` with the same format as `content`
"""
with (yield self._server_linearizer.queue((origin, room_id))):
auth_chain = [
self.event_from_pdu_json(e)
for e in content["auth_chain"]
]
auth_chain = [
self.event_from_pdu_json(e)
for e in content["auth_chain"]
]
signed_auth = yield self._check_sigs_and_hash_and_fetch(
origin, auth_chain, outlier=True
)
signed_auth = yield self._check_sigs_and_hash_and_fetch(
origin, auth_chain, outlier=True
)
ret = yield self.handler.on_query_auth(
origin,
event_id,
signed_auth,
content.get("rejects", []),
content.get("missing", []),
)
ret = yield self.handler.on_query_auth(
origin,
event_id,
signed_auth,
content.get("rejects", []),
content.get("missing", []),
)
time_now = self._clock.time_msec()
send_content = {
"auth_chain": [
e.get_pdu_json(time_now)
for e in ret["auth_chain"]
],
"rejects": ret.get("rejects", []),
"missing": ret.get("missing", []),
}
time_now = self._clock.time_msec()
send_content = {
"auth_chain": [
e.get_pdu_json(time_now)
for e in ret["auth_chain"]
],
"rejects": ret.get("rejects", []),
"missing": ret.get("missing", []),
}
defer.returnValue(
(200, send_content)
)
@defer.inlineCallbacks
@log_function
def on_query_client_keys(self, origin, content):
return self.on_query_request("client_keys", content)
query = []
for user_id, device_ids in content.get("device_keys", {}).items():
if not device_ids:
query.append((user_id, None))
else:
for device_id in device_ids:
query.append((user_id, device_id))
results = yield self.store.get_e2e_device_keys(query)
json_result = {}
for user_id, device_keys in results.items():
for device_id, json_bytes in device_keys.items():
json_result.setdefault(user_id, {})[device_id] = json.loads(
json_bytes
)
defer.returnValue({"device_keys": json_result})
@defer.inlineCallbacks
@log_function
@@ -419,34 +377,16 @@ class FederationServer(FederationBase):
@log_function
def on_get_missing_events(self, origin, room_id, earliest_events,
latest_events, limit, min_depth):
with (yield self._server_linearizer.queue((origin, room_id))):
logger.info(
"on_get_missing_events: earliest_events: %r, latest_events: %r,"
" limit: %d, min_depth: %d",
earliest_events, latest_events, limit, min_depth
)
missing_events = yield self.handler.on_get_missing_events(
origin, room_id, earliest_events, latest_events, limit, min_depth
)
missing_events = yield self.handler.on_get_missing_events(
origin, room_id, earliest_events, latest_events, limit, min_depth
)
if len(missing_events) < 5:
logger.info(
"Returning %d events: %r", len(missing_events), missing_events
)
else:
logger.info("Returning %d events", len(missing_events))
time_now = self._clock.time_msec()
time_now = self._clock.time_msec()
defer.returnValue({
"events": [ev.get_pdu_json(time_now) for ev in missing_events],
})
@log_function
def on_openid_userinfo(self, token):
ts_now_ms = self._clock.time_msec()
return self.store.get_user_id_for_open_id_token(token, ts_now_ms)
@log_function
def _get_persisted_pdu(self, origin, event_id, do_auth=True):
""" Get a PDU from the database with given origin and id.
@@ -536,68 +476,42 @@ class FederationServer(FederationBase):
pdu.internal_metadata.outlier = True
elif min_depth and pdu.depth > min_depth:
if get_missing and prevs - seen:
# If we're missing stuff, ensure we only fetch stuff one
# at a time.
logger.info(
"Acquiring lock for room %r to fetch %d missing events: %r...",
pdu.room_id, len(prevs - seen), list(prevs - seen)[:5],
latest = yield self.store.get_latest_event_ids_in_room(
pdu.room_id
)
with (yield self._room_pdu_linearizer.queue(pdu.room_id)):
logger.info(
"Acquired lock for room %r to fetch %d missing events",
pdu.room_id, len(prevs - seen),
# We add the prev events that we have seen to the latest
# list to ensure the remote server doesn't give them to us
latest = set(latest)
latest |= seen
missing_events = yield self.get_missing_events(
origin,
pdu.room_id,
earliest_events_ids=list(latest),
latest_events=[pdu],
limit=10,
min_depth=min_depth,
)
# We want to sort these by depth so we process them and
# tell clients about them in order.
missing_events.sort(key=lambda x: x.depth)
for e in missing_events:
yield self._handle_new_pdu(
origin,
e,
get_missing=False
)
# We recalculate seen, since it may have changed.
have_seen = yield self.store.have_events(prevs)
seen = set(have_seen.keys())
if prevs - seen:
latest = yield self.store.get_latest_event_ids_in_room(
pdu.room_id
)
# We add the prev events that we have seen to the latest
# list to ensure the remote server doesn't give them to us
latest = set(latest)
latest |= seen
logger.info(
"Missing %d events for room %r: %r...",
len(prevs - seen), pdu.room_id, list(prevs - seen)[:5]
)
missing_events = yield self.get_missing_events(
origin,
pdu.room_id,
earliest_events_ids=list(latest),
latest_events=[pdu],
limit=10,
min_depth=min_depth,
)
# We want to sort these by depth so we process them and
# tell clients about them in order.
missing_events.sort(key=lambda x: x.depth)
for e in missing_events:
yield self._handle_new_pdu(
origin,
e,
get_missing=False
)
have_seen = yield self.store.have_events(
[ev for ev, _ in pdu.prev_events]
)
have_seen = yield self.store.have_events(
[ev for ev, _ in pdu.prev_events]
)
prevs = {e_id for e_id, _ in pdu.prev_events}
seen = set(have_seen.keys())
if prevs - seen:
logger.info(
"Still missing %d events for room %r: %r...",
len(prevs - seen), pdu.room_id, list(prevs - seen)[:5]
)
fetch_state = True
if fetch_state:
@@ -612,7 +526,7 @@ class FederationServer(FederationBase):
origin, pdu.room_id, pdu.event_id,
)
except:
logger.exception("Failed to get state for event: %s", pdu.event_id)
logger.warn("Failed to get state for event: %s", pdu.event_id)
yield self.handler.on_receive_pdu(
origin,

View File

@@ -20,6 +20,8 @@ a given transport.
from .federation_client import FederationClient
from .federation_server import FederationServer
from .transaction_queue import TransactionQueue
from .persistence import TransactionActions
import logging
@@ -64,10 +66,11 @@ class ReplicationLayer(FederationClient, FederationServer):
self._clock = hs.get_clock()
self.transaction_actions = TransactionActions(self.store)
self._transaction_queue = TransactionQueue(hs, transport_layer)
self._order = 0
self.hs = hs
super(ReplicationLayer, self).__init__(hs)
def __str__(self):
return "<ReplicationLayer(%s)>" % self.server_name

View File

@@ -1,298 +0,0 @@
# -*- coding: utf-8 -*-
# Copyright 2014-2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""A federation sender that forwards things to be sent across replication to
a worker process.
It assumes there is a single worker process feeding off of it.
Each row in the replication stream consists of a type and some json, where the
types indicate whether they are presence, or edus, etc.
Ephemeral or non-event data are queued up in-memory. When the worker requests
updates since a particular point, all in-memory data since before that point is
dropped. We also expire things in the queue after 5 minutes, to ensure that a
dead worker doesn't cause the queues to grow limitlessly.
Events are replicated via a separate events stream.
"""
from .units import Edu
from synapse.util.metrics import Measure
import synapse.metrics
from blist import sorteddict
import ujson
metrics = synapse.metrics.get_metrics_for(__name__)
PRESENCE_TYPE = "p"
KEYED_EDU_TYPE = "k"
EDU_TYPE = "e"
FAILURE_TYPE = "f"
DEVICE_MESSAGE_TYPE = "d"
class FederationRemoteSendQueue(object):
"""A drop in replacement for TransactionQueue"""
def __init__(self, hs):
self.server_name = hs.hostname
self.clock = hs.get_clock()
self.presence_map = {}
self.presence_changed = sorteddict()
self.keyed_edu = {}
self.keyed_edu_changed = sorteddict()
self.edus = sorteddict()
self.failures = sorteddict()
self.device_messages = sorteddict()
self.pos = 1
self.pos_time = sorteddict()
# EVERYTHING IS SAD. In particular, python only makes new scopes when
# we make a new function, so we need to make a new function so the inner
# lambda binds to the queue rather than to the name of the queue which
# changes. ARGH.
def register(name, queue):
metrics.register_callback(
queue_name + "_size",
lambda: len(queue),
)
for queue_name in [
"presence_map", "presence_changed", "keyed_edu", "keyed_edu_changed",
"edus", "failures", "device_messages", "pos_time",
]:
register(queue_name, getattr(self, queue_name))
self.clock.looping_call(self._clear_queue, 30 * 1000)
def _next_pos(self):
pos = self.pos
self.pos += 1
self.pos_time[self.clock.time_msec()] = pos
return pos
def _clear_queue(self):
"""Clear the queues for anything older than N minutes"""
FIVE_MINUTES_AGO = 5 * 60 * 1000
now = self.clock.time_msec()
keys = self.pos_time.keys()
time = keys.bisect_left(now - FIVE_MINUTES_AGO)
if not keys[:time]:
return
position_to_delete = max(keys[:time])
for key in keys[:time]:
del self.pos_time[key]
self._clear_queue_before_pos(position_to_delete)
def _clear_queue_before_pos(self, position_to_delete):
"""Clear all the queues from before a given position"""
with Measure(self.clock, "send_queue._clear"):
# Delete things out of presence maps
keys = self.presence_changed.keys()
i = keys.bisect_left(position_to_delete)
for key in keys[:i]:
del self.presence_changed[key]
user_ids = set(
user_id for uids in self.presence_changed.values() for _, user_id in uids
)
to_del = [
user_id for user_id in self.presence_map if user_id not in user_ids
]
for user_id in to_del:
del self.presence_map[user_id]
# Delete things out of keyed edus
keys = self.keyed_edu_changed.keys()
i = keys.bisect_left(position_to_delete)
for key in keys[:i]:
del self.keyed_edu_changed[key]
live_keys = set()
for edu_key in self.keyed_edu_changed.values():
live_keys.add(edu_key)
to_del = [edu_key for edu_key in self.keyed_edu if edu_key not in live_keys]
for edu_key in to_del:
del self.keyed_edu[edu_key]
# Delete things out of edu map
keys = self.edus.keys()
i = keys.bisect_left(position_to_delete)
for key in keys[:i]:
del self.edus[key]
# Delete things out of failure map
keys = self.failures.keys()
i = keys.bisect_left(position_to_delete)
for key in keys[:i]:
del self.failures[key]
# Delete things out of device map
keys = self.device_messages.keys()
i = keys.bisect_left(position_to_delete)
for key in keys[:i]:
del self.device_messages[key]
def notify_new_events(self, current_id):
"""As per TransactionQueue"""
# We don't need to replicate this as it gets sent down a different
# stream.
pass
def send_edu(self, destination, edu_type, content, key=None):
"""As per TransactionQueue"""
pos = self._next_pos()
edu = Edu(
origin=self.server_name,
destination=destination,
edu_type=edu_type,
content=content,
)
if key:
assert isinstance(key, tuple)
self.keyed_edu[(destination, key)] = edu
self.keyed_edu_changed[pos] = (destination, key)
else:
self.edus[pos] = edu
def send_presence(self, destination, states):
"""As per TransactionQueue"""
pos = self._next_pos()
self.presence_map.update({
state.user_id: state
for state in states
})
self.presence_changed[pos] = [
(destination, state.user_id) for state in states
]
def send_failure(self, failure, destination):
"""As per TransactionQueue"""
pos = self._next_pos()
self.failures[pos] = (destination, str(failure))
def send_device_messages(self, destination):
"""As per TransactionQueue"""
pos = self._next_pos()
self.device_messages[pos] = destination
def get_current_token(self):
return self.pos - 1
def get_replication_rows(self, token, limit, federation_ack=None):
"""
Args:
token (int)
limit (int)
federation_ack (int): Optional. The position where the worker is
explicitly acknowledged it has handled. Allows us to drop
data from before that point
"""
# TODO: Handle limit.
# To handle restarts where we wrap around
if token > self.pos:
token = -1
rows = []
# There should be only one reader, so lets delete everything its
# acknowledged its seen.
if federation_ack:
self._clear_queue_before_pos(federation_ack)
# Fetch changed presence
keys = self.presence_changed.keys()
i = keys.bisect_right(token)
dest_user_ids = set(
(pos, dest_user_id)
for pos in keys[i:]
for dest_user_id in self.presence_changed[pos]
)
for (key, (dest, user_id)) in dest_user_ids:
rows.append((key, PRESENCE_TYPE, ujson.dumps({
"destination": dest,
"state": self.presence_map[user_id].as_dict(),
})))
# Fetch changes keyed edus
keys = self.keyed_edu_changed.keys()
i = keys.bisect_right(token)
keyed_edus = set((k, self.keyed_edu_changed[k]) for k in keys[i:])
for (pos, (destination, edu_key)) in keyed_edus:
rows.append(
(pos, KEYED_EDU_TYPE, ujson.dumps({
"key": edu_key,
"edu": self.keyed_edu[(destination, edu_key)].get_internal_dict(),
}))
)
# Fetch changed edus
keys = self.edus.keys()
i = keys.bisect_right(token)
edus = set((k, self.edus[k]) for k in keys[i:])
for (pos, edu) in edus:
rows.append((pos, EDU_TYPE, ujson.dumps(edu.get_internal_dict())))
# Fetch changed failures
keys = self.failures.keys()
i = keys.bisect_right(token)
failures = set((k, self.failures[k]) for k in keys[i:])
for (pos, (destination, failure)) in failures:
rows.append((pos, FAILURE_TYPE, ujson.dumps({
"destination": destination,
"failure": failure,
})))
# Fetch changed device messages
keys = self.device_messages.keys()
i = keys.bisect_right(token)
device_messages = set((k, self.device_messages[k]) for k in keys[i:])
for (pos, destination) in device_messages:
rows.append((pos, DEVICE_MESSAGE_TYPE, ujson.dumps({
"destination": destination,
})))
# Sort rows based on pos
rows.sort()
return rows

View File

@@ -17,18 +17,14 @@
from twisted.internet import defer
from .persistence import TransactionActions
from .units import Transaction, Edu
from .units import Transaction
from synapse.api.constants import EventTypes, Membership
from synapse.api.errors import HttpResponseException
from synapse.util.async import run_on_reactor
from synapse.util.logcontext import preserve_context_over_fn
from synapse.util.logutils import log_function
from synapse.util.logcontext import PreserveLoggingContext
from synapse.util.retryutils import (
get_retry_limiter, NotRetryingDestination,
)
from synapse.util.metrics import measure_func
from synapse.types import get_domain_from_id
from synapse.handlers.presence import format_user_presence_state
import synapse.metrics
import logging
@@ -38,12 +34,6 @@ logger = logging.getLogger(__name__)
metrics = synapse.metrics.get_metrics_for(__name__)
client_metrics = synapse.metrics.get_metrics_for("synapse.federation.client")
sent_pdus_destination_dist = client_metrics.register_distribution(
"sent_pdu_destinations"
)
sent_edus_counter = client_metrics.register_counter("sent_edus")
class TransactionQueue(object):
"""This class makes sure we only have one transaction in flight at
@@ -52,16 +42,15 @@ class TransactionQueue(object):
It batches pending PDUs into single transactions.
"""
def __init__(self, hs):
def __init__(self, hs, transport_layer):
self.server_name = hs.hostname
self.store = hs.get_datastore()
self.state = hs.get_state_handler()
self.transaction_actions = TransactionActions(self.store)
self.transport_layer = hs.get_federation_transport_client()
self.transport_layer = transport_layer
self.clock = hs.get_clock()
self._clock = hs.get_clock()
# Is a mapping from destinations -> deferreds. Used to keep track
# of which destinations have transactions in flight and when they are
@@ -79,35 +68,20 @@ class TransactionQueue(object):
# destination -> list of tuple(edu, deferred)
self.pending_edus_by_dest = edus = {}
# Presence needs to be separate as we send single aggragate EDUs
self.pending_presence_by_dest = presence = {}
self.pending_edus_keyed_by_dest = edus_keyed = {}
metrics.register_callback(
"pending_pdus",
lambda: sum(map(len, pdus.values())),
)
metrics.register_callback(
"pending_edus",
lambda: (
sum(map(len, edus.values()))
+ sum(map(len, presence.values()))
+ sum(map(len, edus_keyed.values()))
),
lambda: sum(map(len, edus.values())),
)
# destination -> list of tuple(failure, deferred)
self.pending_failures_by_dest = {}
self.last_device_stream_id_by_dest = {}
# HACK to get unique tx id
self._next_txn_id = int(self.clock.time_msec())
self._order = 1
self._is_processing = False
self._last_poked_id = -1
self._next_txn_id = int(self._clock.time_msec())
def can_send_to(self, destination):
"""Can we send messages to the given server?
@@ -129,61 +103,11 @@ class TransactionQueue(object):
else:
return not destination.startswith("localhost")
@defer.inlineCallbacks
def notify_new_events(self, current_id):
"""This gets called when we have some new events we might want to
send out to other servers.
"""
self._last_poked_id = max(current_id, self._last_poked_id)
if self._is_processing:
return
try:
self._is_processing = True
while True:
last_token = yield self.store.get_federation_out_pos("events")
next_token, events = yield self.store.get_all_new_events_stream(
last_token, self._last_poked_id, limit=20,
)
logger.debug("Handling %s -> %s", last_token, next_token)
if not events and next_token >= self._last_poked_id:
break
for event in events:
users_in_room = yield self.state.get_current_user_in_room(
event.room_id, latest_event_ids=[event.event_id],
)
destinations = set(
get_domain_from_id(user_id) for user_id in users_in_room
)
if event.type == EventTypes.Member:
if event.content["membership"] == Membership.JOIN:
destinations.add(get_domain_from_id(event.state_key))
logger.debug("Sending %s to %r", event, destinations)
self._send_pdu(event, destinations)
yield self.store.update_federation_out_pos(
"events", next_token
)
finally:
self._is_processing = False
def _send_pdu(self, pdu, destinations):
def enqueue_pdu(self, pdu, destinations, order):
# We loop through all destinations to see whether we already have
# a transaction in progress. If we do, stick it in the pending_pdus
# table and we'll get back to it later.
order = self._order
self._order += 1
destinations = set(destinations)
destinations = set(
dest for dest in destinations if self.can_send_to(dest)
@@ -194,83 +118,86 @@ class TransactionQueue(object):
if not destinations:
return
sent_pdus_destination_dist.inc_by(len(destinations))
deferreds = []
for destination in destinations:
deferred = defer.Deferred()
self.pending_pdus_by_dest.setdefault(destination, []).append(
(pdu, order)
(pdu, deferred, order)
)
preserve_context_over_fn(
self._attempt_new_transaction, destination
)
def chain(failure):
if not deferred.called:
deferred.errback(failure)
def send_presence(self, destination, states):
if not self.can_send_to(destination):
return
def log_failure(f):
logger.warn("Failed to send pdu to %s: %s", destination, f.value)
self.pending_presence_by_dest.setdefault(destination, {}).update({
state.user_id: state for state in states
})
deferred.addErrback(log_failure)
preserve_context_over_fn(
self._attempt_new_transaction, destination
)
with PreserveLoggingContext():
self._attempt_new_transaction(destination).addErrback(chain)
def send_edu(self, destination, edu_type, content, key=None):
edu = Edu(
origin=self.server_name,
destination=destination,
edu_type=edu_type,
content=content,
)
deferreds.append(deferred)
# NO inlineCallbacks
def enqueue_edu(self, edu):
destination = edu.destination
if not self.can_send_to(destination):
return
sent_edus_counter.inc()
if key:
self.pending_edus_keyed_by_dest.setdefault(
destination, {}
)[(edu.edu_type, key)] = edu
else:
self.pending_edus_by_dest.setdefault(destination, []).append(edu)
preserve_context_over_fn(
self._attempt_new_transaction, destination
deferred = defer.Deferred()
self.pending_edus_by_dest.setdefault(destination, []).append(
(edu, deferred)
)
def send_failure(self, failure, destination):
def chain(failure):
if not deferred.called:
deferred.errback(failure)
def log_failure(f):
logger.warn("Failed to send edu to %s: %s", destination, f.value)
deferred.addErrback(log_failure)
with PreserveLoggingContext():
self._attempt_new_transaction(destination).addErrback(chain)
return deferred
@defer.inlineCallbacks
def enqueue_failure(self, failure, destination):
if destination == self.server_name or destination == "localhost":
return
deferred = defer.Deferred()
if not self.can_send_to(destination):
return
self.pending_failures_by_dest.setdefault(
destination, []
).append(failure)
preserve_context_over_fn(
self._attempt_new_transaction, destination
).append(
(failure, deferred)
)
def send_device_messages(self, destination):
if destination == self.server_name or destination == "localhost":
return
def chain(f):
if not deferred.called:
deferred.errback(f)
if not self.can_send_to(destination):
return
def log_failure(f):
logger.warn("Failed to send failure to %s: %s", destination, f.value)
preserve_context_over_fn(
self._attempt_new_transaction, destination
)
deferred.addErrback(log_failure)
def get_current_token(self):
return 0
with PreserveLoggingContext():
self._attempt_new_transaction(destination).addErrback(chain)
yield deferred
@defer.inlineCallbacks
@log_function
def _attempt_new_transaction(self, destination):
# list of (pending_pdu, deferred, order)
if destination in self.pending_transactions:
@@ -284,128 +211,55 @@ class TransactionQueue(object):
)
return
pending_pdus = self.pending_pdus_by_dest.pop(destination, [])
pending_edus = self.pending_edus_by_dest.pop(destination, [])
pending_failures = self.pending_failures_by_dest.pop(destination, [])
if pending_pdus:
logger.debug("TX [%s] len(pending_pdus_by_dest[dest]) = %d",
destination, len(pending_pdus))
if not pending_pdus and not pending_edus and not pending_failures:
logger.debug("TX [%s] Nothing to send", destination)
return
try:
self.pending_transactions[destination] = 1
yield run_on_reactor()
while True:
pending_pdus = self.pending_pdus_by_dest.pop(destination, [])
pending_edus = self.pending_edus_by_dest.pop(destination, [])
pending_presence = self.pending_presence_by_dest.pop(destination, {})
pending_failures = self.pending_failures_by_dest.pop(destination, [])
pending_edus.extend(
self.pending_edus_keyed_by_dest.pop(destination, {}).values()
)
limiter = yield get_retry_limiter(
destination,
self.clock,
self.store,
)
device_message_edus, device_stream_id = (
yield self._get_new_device_messages(destination)
)
pending_edus.extend(device_message_edus)
if pending_presence:
pending_edus.append(
Edu(
origin=self.server_name,
destination=destination,
edu_type="m.presence",
content={
"push": [
format_user_presence_state(
presence, self.clock.time_msec()
)
for presence in pending_presence.values()
]
},
)
)
if pending_pdus:
logger.debug("TX [%s] len(pending_pdus_by_dest[dest]) = %d",
destination, len(pending_pdus))
if not pending_pdus and not pending_edus and not pending_failures:
logger.debug("TX [%s] Nothing to send", destination)
self.last_device_stream_id_by_dest[destination] = (
device_stream_id
)
return
success = yield self._send_new_transaction(
destination, pending_pdus, pending_edus, pending_failures,
device_stream_id,
should_delete_from_device_stream=bool(device_message_edus),
limiter=limiter,
)
if not success:
break
except NotRetryingDestination:
logger.info(
"TX [%s] not ready for retry yet - "
"dropping transaction for now",
destination,
)
finally:
# We want to be *very* sure we delete this after we stop processing
self.pending_transactions.pop(destination, None)
@defer.inlineCallbacks
def _get_new_device_messages(self, destination):
last_device_stream_id = self.last_device_stream_id_by_dest.get(destination, 0)
to_device_stream_id = self.store.get_to_device_stream_token()
contents, stream_id = yield self.store.get_new_device_msgs_for_remote(
destination, last_device_stream_id, to_device_stream_id
)
edus = [
Edu(
origin=self.server_name,
destination=destination,
edu_type="m.direct_to_device",
content=content,
)
for content in contents
]
defer.returnValue((edus, stream_id))
@measure_func("_send_new_transaction")
@defer.inlineCallbacks
def _send_new_transaction(self, destination, pending_pdus, pending_edus,
pending_failures, device_stream_id,
should_delete_from_device_stream, limiter):
# Sort based on the order field
pending_pdus.sort(key=lambda t: t[1])
pdus = [x[0] for x in pending_pdus]
edus = pending_edus
failures = [x.get_dict() for x in pending_failures]
success = True
try:
logger.debug("TX [%s] _attempt_new_transaction", destination)
# Sort based on the order field
pending_pdus.sort(key=lambda t: t[2])
pdus = [x[0] for x in pending_pdus]
edus = [x[0] for x in pending_edus]
failures = [x[0].get_dict() for x in pending_failures]
deferreds = [
x[1]
for x in pending_pdus + pending_edus + pending_failures
]
txn_id = str(self._next_txn_id)
limiter = yield get_retry_limiter(
destination,
self._clock,
self.store,
)
logger.debug(
"TX [%s] {%s} Attempting new transaction"
" (pdus: %d, edus: %d, failures: %d)",
destination, txn_id,
len(pdus),
len(edus),
len(failures)
len(pending_pdus),
len(pending_edus),
len(pending_failures)
)
logger.debug("TX [%s] Persisting transaction...", destination)
transaction = Transaction.create_new(
origin_server_ts=int(self.clock.time_msec()),
origin_server_ts=int(self._clock.time_msec()),
transaction_id=txn_id,
origin=self.server_name,
destination=destination,
@@ -424,9 +278,9 @@ class TransactionQueue(object):
" (PDUs: %d, EDUs: %d, failures: %d)",
destination, txn_id,
transaction.transaction_id,
len(pdus),
len(edus),
len(failures),
len(pending_pdus),
len(pending_edus),
len(pending_failures),
)
with limiter:
@@ -436,7 +290,7 @@ class TransactionQueue(object):
# keys work
def json_data_cb():
data = transaction.get_dict()
now = int(self.clock.time_msec())
now = int(self._clock.time_msec())
if "pdus" in data:
for p in data["pdus"]:
if "age_ts" in p:
@@ -462,13 +316,6 @@ class TransactionQueue(object):
code = e.code
response = e.response
if e.code == 429 or 500 <= e.code:
logger.info(
"TX [%s] {%s} got %d response",
destination, txn_id, code
)
raise e
logger.info(
"TX [%s] {%s} got %d response",
destination, txn_id, code
@@ -483,19 +330,28 @@ class TransactionQueue(object):
logger.debug("TX [%s] Marked as delivered", destination)
if code != 200:
for p in pdus:
logger.info(
"Failed to send event %s to %s", p.event_id, destination
)
success = False
else:
# Remove the acknowledged device messages from the database
if should_delete_from_device_stream:
yield self.store.delete_device_msgs_for_remote(
destination, device_stream_id
)
self.last_device_stream_id_by_dest[destination] = device_stream_id
logger.debug("TX [%s] Yielding to callbacks...", destination)
for deferred in deferreds:
if code == 200:
deferred.callback(None)
else:
deferred.errback(RuntimeError("Got status %d" % code))
# Ensures we don't continue until all callbacks on that
# deferred have fired
try:
yield deferred
except:
pass
logger.debug("TX [%s] Yielded to callbacks", destination)
except NotRetryingDestination:
logger.info(
"TX [%s] not ready for retry yet - "
"dropping transaction for now",
destination,
)
except RuntimeError as e:
# We capture this here as there as nothing actually listens
# for this finishing functions deferred.
@@ -504,11 +360,6 @@ class TransactionQueue(object):
destination,
e,
)
success = False
for p in pdus:
logger.info("Failed to send event %s to %s", p.event_id, destination)
except Exception as e:
# We capture this here as there as nothing actually listens
# for this finishing functions deferred.
@@ -518,9 +369,13 @@ class TransactionQueue(object):
e,
)
success = False
for deferred in deferreds:
if not deferred.called:
deferred.errback(e)
for p in pdus:
logger.info("Failed to send event %s to %s", p.event_id, destination)
finally:
# We want to be *very* sure we delete this after we stop processing
self.pending_transactions.pop(destination, None)
defer.returnValue(success)
# Check to see if there is anything else to send.
self._attempt_new_transaction(destination)

View File

@@ -54,28 +54,6 @@ class TransportLayerClient(object):
destination, path=path, args={"event_id": event_id},
)
@log_function
def get_room_state_ids(self, destination, room_id, event_id):
""" Requests all state for a given room from the given server at the
given event. Returns the state's event_id's
Args:
destination (str): The host name of the remote home server we want
to get the state from.
context (str): The name of the context we want the state of
event_id (str): The event we want the context at.
Returns:
Deferred: Results in a dict received from the remote homeserver.
"""
logger.debug("get_room_state_ids dest=%s, room=%s",
destination, room_id)
path = PREFIX + "/state_ids/%s/" % room_id
return self.client.get_json(
destination, path=path, args={"event_id": event_id},
)
@log_function
def get_event(self, destination, event_id, timeout=None):
""" Requests the pdu with give id and origin from the given server.
@@ -201,8 +179,7 @@ class TransportLayerClient(object):
content = yield self.client.get_json(
destination=destination,
path=path,
retry_on_dns_fail=False,
timeout=20000,
retry_on_dns_fail=True,
)
defer.returnValue(content)
@@ -246,33 +223,6 @@ class TransportLayerClient(object):
defer.returnValue(response)
@defer.inlineCallbacks
@log_function
def get_public_rooms(self, remote_server, limit, since_token,
search_filter=None, include_all_networks=False,
third_party_instance_id=None):
path = PREFIX + "/publicRooms"
args = {
"include_all_networks": "true" if include_all_networks else "false",
}
if third_party_instance_id:
args["third_party_instance_id"] = third_party_instance_id,
if limit:
args["limit"] = [str(limit)]
if since_token:
args["since"] = [since_token]
# TODO(erikj): Actually send the search_filter across federation.
response = yield self.client.get_json(
destination=remote_server,
path=path,
args=args,
)
defer.returnValue(response)
@defer.inlineCallbacks
@log_function
def exchange_third_party_invite(self, destination, room_id, event_dict):
@@ -313,7 +263,7 @@ class TransportLayerClient(object):
@defer.inlineCallbacks
@log_function
def query_client_keys(self, destination, query_content, timeout):
def query_client_keys(self, destination, query_content):
"""Query the device keys for a list of user ids hosted on a remote
server.
@@ -342,13 +292,12 @@ class TransportLayerClient(object):
destination=destination,
path=path,
data=query_content,
timeout=timeout,
)
defer.returnValue(content)
@defer.inlineCallbacks
@log_function
def claim_client_keys(self, destination, query_content, timeout):
def claim_client_keys(self, destination, query_content):
"""Claim one-time keys for a list of devices hosted on a remote server.
Request:
@@ -379,7 +328,6 @@ class TransportLayerClient(object):
destination=destination,
path=path,
data=query_content,
timeout=timeout,
)
defer.returnValue(content)

View File

@@ -18,18 +18,13 @@ from twisted.internet import defer
from synapse.api.urls import FEDERATION_PREFIX as PREFIX
from synapse.api.errors import Codes, SynapseError
from synapse.http.server import JsonResource
from synapse.http.servlet import (
parse_json_object_from_request, parse_integer_from_args, parse_string_from_args,
parse_boolean_from_args,
)
from synapse.http.servlet import parse_json_object_from_request
from synapse.util.ratelimitutils import FederationRateLimiter
from synapse.util.versionstring import get_version_string
from synapse.types import ThirdPartyInstanceID
import functools
import logging
import simplejson as json
import re
import synapse
logger = logging.getLogger(__name__)
@@ -42,7 +37,7 @@ class TransportLayerServer(JsonResource):
self.hs = hs
self.clock = hs.get_clock()
super(TransportLayerServer, self).__init__(hs, canonical_json=False)
super(TransportLayerServer, self).__init__(hs)
self.authenticator = Authenticator(hs)
self.ratelimiter = FederationRateLimiter(
@@ -65,16 +60,6 @@ class TransportLayerServer(JsonResource):
)
class AuthenticationError(SynapseError):
"""There was a problem authenticating the request"""
pass
class NoAuthenticationError(AuthenticationError):
"""The request had no authentication information"""
pass
class Authenticator(object):
def __init__(self, hs):
self.keyring = hs.get_keyring()
@@ -82,7 +67,7 @@ class Authenticator(object):
# A method just so we can pass 'self' as the authenticator to the Servlets
@defer.inlineCallbacks
def authenticate_request(self, request, content):
def authenticate_request(self, request):
json_request = {
"method": request.method,
"uri": request.uri,
@@ -90,11 +75,18 @@ class Authenticator(object):
"signatures": {},
}
if content is not None:
json_request["content"] = content
content = None
origin = None
if request.method in ["PUT", "POST"]:
# TODO: Handle other method types? other content types?
try:
content_bytes = request.content.read()
content = json.loads(content_bytes)
json_request["content"] = content
except:
raise SynapseError(400, "Unable to parse JSON", Codes.BAD_JSON)
def parse_auth_header(header_str):
try:
params = auth.split(" ")[1].split(",")
@@ -111,14 +103,14 @@ class Authenticator(object):
sig = strip_quotes(param_dict["sig"])
return (origin, key, sig)
except:
raise AuthenticationError(
raise SynapseError(
400, "Malformed Authorization header", Codes.UNAUTHORIZED
)
auth_headers = request.requestHeaders.getRawHeaders(b"Authorization")
if not auth_headers:
raise NoAuthenticationError(
raise SynapseError(
401, "Missing Authorization headers", Codes.UNAUTHORIZED,
)
@@ -129,7 +121,7 @@ class Authenticator(object):
json_request["signatures"].setdefault(origin, {})[key] = sig
if not json_request["signatures"]:
raise NoAuthenticationError(
raise SynapseError(
401, "Missing Authorization headers", Codes.UNAUTHORIZED,
)
@@ -138,59 +130,38 @@ class Authenticator(object):
logger.info("Request from %s", origin)
request.authenticated_entity = origin
defer.returnValue(origin)
defer.returnValue((origin, content))
class BaseFederationServlet(object):
REQUIRE_AUTH = True
def __init__(self, handler, authenticator, ratelimiter, server_name,
room_list_handler):
def __init__(self, handler, authenticator, ratelimiter, server_name):
self.handler = handler
self.authenticator = authenticator
self.ratelimiter = ratelimiter
self.room_list_handler = room_list_handler
def _wrap(self, func):
def _wrap(self, code):
authenticator = self.authenticator
ratelimiter = self.ratelimiter
@defer.inlineCallbacks
@functools.wraps(func)
def new_func(request, *args, **kwargs):
content = None
if request.method in ["PUT", "POST"]:
# TODO: Handle other method types? other content types?
content = parse_json_object_from_request(request)
@functools.wraps(code)
def new_code(request, *args, **kwargs):
try:
origin = yield authenticator.authenticate_request(request, content)
except NoAuthenticationError:
origin = None
if self.REQUIRE_AUTH:
logger.exception("authenticate_request failed")
raise
(origin, content) = yield authenticator.authenticate_request(request)
with ratelimiter.ratelimit(origin) as d:
yield d
response = yield code(
origin, content, request.args, *args, **kwargs
)
except:
logger.exception("authenticate_request failed")
raise
if origin:
with ratelimiter.ratelimit(origin) as d:
yield d
response = yield func(
origin, content, request.args, *args, **kwargs
)
else:
response = yield func(
origin, content, request.args, *args, **kwargs
)
defer.returnValue(response)
# Extra logic that functools.wraps() doesn't finish
new_func.__self__ = func.__self__
new_code.__self__ = code.__self__
return new_func
return new_code
def register(self, server):
pattern = re.compile("^" + PREFIX + self.PATH + "$")
@@ -298,17 +269,6 @@ class FederationStateServlet(BaseFederationServlet):
)
class FederationStateIdsServlet(BaseFederationServlet):
PATH = "/state_ids/(?P<room_id>[^/]*)/"
def on_GET(self, origin, content, query, room_id):
return self.handler.on_state_ids_request(
origin,
room_id,
query.get("event_id", [None])[0],
)
class FederationBackfillServlet(BaseFederationServlet):
PATH = "/backfill/(?P<context>[^/]*)/"
@@ -363,7 +323,7 @@ class FederationSendLeaveServlet(BaseFederationServlet):
class FederationEventAuthServlet(BaseFederationServlet):
PATH = "/event_auth/(?P<context>[^/]*)/(?P<event_id>[^/]*)"
PATH = "/event_auth(?P<context>[^/]*)/(?P<event_id>[^/]*)"
def on_GET(self, origin, content, query, context, event_id):
return self.handler.on_event_auth(origin, context, event_id)
@@ -405,8 +365,10 @@ class FederationThirdPartyInviteExchangeServlet(BaseFederationServlet):
class FederationClientKeysQueryServlet(BaseFederationServlet):
PATH = "/user/keys/query"
@defer.inlineCallbacks
def on_POST(self, origin, content, query):
return self.handler.on_query_client_keys(origin, content)
response = yield self.handler.on_query_client_keys(origin, content)
defer.returnValue((200, response))
class FederationClientKeysClaimServlet(BaseFederationServlet):
@@ -424,7 +386,7 @@ class FederationQueryAuthServlet(BaseFederationServlet):
@defer.inlineCallbacks
def on_POST(self, origin, content, query, context, event_id):
new_content = yield self.handler.on_query_auth_request(
origin, content, context, event_id
origin, content, event_id
)
defer.returnValue((200, new_content))
@@ -456,10 +418,9 @@ class FederationGetMissingEventsServlet(BaseFederationServlet):
class On3pidBindServlet(BaseFederationServlet):
PATH = "/3pid/onbind"
REQUIRE_AUTH = False
@defer.inlineCallbacks
def on_POST(self, origin, content, query):
def on_POST(self, request):
content = parse_json_object_from_request(request)
if "invites" in content:
last_exception = None
for invite in content["invites"]:
@@ -481,118 +442,10 @@ class On3pidBindServlet(BaseFederationServlet):
raise last_exception
defer.returnValue((200, {}))
class OpenIdUserInfo(BaseFederationServlet):
"""
Exchange a bearer token for information about a user.
The response format should be compatible with:
http://openid.net/specs/openid-connect-core-1_0.html#UserInfoResponse
GET /openid/userinfo?access_token=ABDEFGH HTTP/1.1
HTTP/1.1 200 OK
Content-Type: application/json
{
"sub": "@userpart:example.org",
}
"""
PATH = "/openid/userinfo"
REQUIRE_AUTH = False
@defer.inlineCallbacks
def on_GET(self, origin, content, query):
token = query.get("access_token", [None])[0]
if token is None:
defer.returnValue((401, {
"errcode": "M_MISSING_TOKEN", "error": "Access Token required"
}))
return
user_id = yield self.handler.on_openid_userinfo(token)
if user_id is None:
defer.returnValue((401, {
"errcode": "M_UNKNOWN_TOKEN",
"error": "Access Token unknown or expired"
}))
defer.returnValue((200, {"sub": user_id}))
class PublicRoomList(BaseFederationServlet):
"""
Fetch the public room list for this server.
This API returns information in the same format as /publicRooms on the
client API, but will only ever include local public rooms and hence is
intended for consumption by other home servers.
GET /publicRooms HTTP/1.1
HTTP/1.1 200 OK
Content-Type: application/json
{
"chunk": [
{
"aliases": [
"#test:localhost"
],
"guest_can_join": false,
"name": "test room",
"num_joined_members": 3,
"room_id": "!whkydVegtvatLfXmPN:localhost",
"world_readable": false
}
],
"end": "END",
"start": "START"
}
"""
PATH = "/publicRooms"
@defer.inlineCallbacks
def on_GET(self, origin, content, query):
limit = parse_integer_from_args(query, "limit", 0)
since_token = parse_string_from_args(query, "since", None)
include_all_networks = parse_boolean_from_args(
query, "include_all_networks", False
)
third_party_instance_id = parse_string_from_args(
query, "third_party_instance_id", None
)
if include_all_networks:
network_tuple = None
elif third_party_instance_id:
network_tuple = ThirdPartyInstanceID.from_string(third_party_instance_id)
else:
network_tuple = ThirdPartyInstanceID(None, None)
data = yield self.room_list_handler.get_local_public_room_list(
limit, since_token,
network_tuple=network_tuple
)
defer.returnValue((200, data))
class FederationVersionServlet(BaseFederationServlet):
PATH = "/version"
REQUIRE_AUTH = False
def on_GET(self, origin, content, query):
return defer.succeed((200, {
"server": {
"name": "Synapse",
"version": get_version_string(synapse)
},
}))
# Avoid doing remote HS authorization checks which are done by default by
# BaseFederationServlet.
def _wrap(self, code):
return code
SERVLET_CLASSES = (
@@ -600,7 +453,6 @@ SERVLET_CLASSES = (
FederationPullServlet,
FederationEventServlet,
FederationStateServlet,
FederationStateIdsServlet,
FederationBackfillServlet,
FederationQueryServlet,
FederationMakeJoinServlet,
@@ -616,9 +468,6 @@ SERVLET_CLASSES = (
FederationClientKeysClaimServlet,
FederationThirdPartyInviteExchangeServlet,
On3pidBindServlet,
OpenIdUserInfo,
PublicRoomList,
FederationVersionServlet,
)
@@ -629,5 +478,4 @@ def register_servlets(hs, resource, authenticator, ratelimiter):
authenticator=authenticator,
ratelimiter=ratelimiter,
server_name=hs.hostname,
room_list_handler=hs.get_room_list_handler(),
).register(resource)

View File

@@ -13,37 +13,35 @@
# See the License for the specific language governing permissions and
# limitations under the License.
from synapse.appservice.scheduler import AppServiceScheduler
from synapse.appservice.api import ApplicationServiceApi
from .register import RegistrationHandler
from .room import (
RoomCreationHandler, RoomContextHandler,
RoomCreationHandler, RoomListHandler, RoomContextHandler,
)
from .room_member import RoomMemberHandler
from .message import MessageHandler
from .events import EventStreamHandler, EventHandler
from .federation import FederationHandler
from .profile import ProfileHandler
from .presence import PresenceHandler
from .directory import DirectoryHandler
from .typing import TypingNotificationHandler
from .admin import AdminHandler
from .appservice import ApplicationServicesHandler
from .sync import SyncHandler
from .auth import AuthHandler
from .identity import IdentityHandler
from .receipts import ReceiptsHandler
from .search import SearchHandler
class Handlers(object):
""" Deprecated. A collection of handlers.
""" A collection of all the event handlers.
At some point most of the classes whose name ended "Handler" were
accessed through this class.
However this makes it painful to unit test the handlers and to run cut
down versions of synapse that only use specific handlers because using a
single handler required creating all of the handlers. So some of the
handlers have been lifted out of the Handlers object and are now accessed
directly through the homeserver object itself.
Any new handlers should follow the new pattern of being accessed through
the homeserver object and should not be added to the Handlers object.
The remaining handlers should be moved out of the handlers object.
There's no need to lazily create these; we'll just make them all eagerly
at construction time.
"""
def __init__(self, hs):
@@ -51,10 +49,26 @@ class Handlers(object):
self.message_handler = MessageHandler(hs)
self.room_creation_handler = RoomCreationHandler(hs)
self.room_member_handler = RoomMemberHandler(hs)
self.event_stream_handler = EventStreamHandler(hs)
self.event_handler = EventHandler(hs)
self.federation_handler = FederationHandler(hs)
self.profile_handler = ProfileHandler(hs)
self.presence_handler = PresenceHandler(hs)
self.room_list_handler = RoomListHandler(hs)
self.directory_handler = DirectoryHandler(hs)
self.typing_notification_handler = TypingNotificationHandler(hs)
self.admin_handler = AdminHandler(hs)
self.receipts_handler = ReceiptsHandler(hs)
asapi = ApplicationServiceApi(hs)
self.appservice_handler = ApplicationServicesHandler(
hs, asapi, AppServiceScheduler(
clock=hs.get_clock(),
store=hs.get_datastore(),
as_api=asapi
)
)
self.sync_handler = SyncHandler(hs)
self.auth_handler = AuthHandler(hs)
self.identity_handler = IdentityHandler(hs)
self.search_handler = SearchHandler(hs)
self.room_context_handler = RoomContextHandler(hs)

View File

@@ -13,33 +13,49 @@
# See the License for the specific language governing permissions and
# limitations under the License.
import logging
from twisted.internet import defer
import synapse.types
from synapse.api.errors import LimitExceededError, SynapseError, AuthError
from synapse.crypto.event_signing import add_hashes_and_signatures
from synapse.api.constants import Membership, EventTypes
from synapse.api.errors import LimitExceededError
from synapse.types import UserID
from synapse.types import UserID, RoomAlias, Requester
from synapse.push.action_generator import ActionGenerator
from synapse.util.logcontext import PreserveLoggingContext
import logging
logger = logging.getLogger(__name__)
VISIBILITY_PRIORITY = (
"world_readable",
"shared",
"invited",
"joined",
)
MEMBERSHIP_PRIORITY = (
Membership.JOIN,
Membership.INVITE,
Membership.KNOCK,
Membership.LEAVE,
Membership.BAN,
)
class BaseHandler(object):
"""
Common base class for the event handlers.
Attributes:
store (synapse.storage.DataStore):
store (synapse.storage.events.StateStore):
state_handler (synapse.state.StateHandler):
"""
def __init__(self, hs):
"""
Args:
hs (synapse.server.HomeServer):
"""
self.store = hs.get_datastore()
self.auth = hs.get_auth()
self.notifier = hs.get_notifier()
@@ -49,26 +65,165 @@ class BaseHandler(object):
self.clock = hs.get_clock()
self.hs = hs
self.signing_key = hs.config.signing_key[0]
self.server_name = hs.hostname
self.event_builder_factory = hs.get_event_builder_factory()
@defer.inlineCallbacks
def filter_events_for_clients(self, user_tuples, events, event_id_to_state):
""" Returns dict of user_id -> list of events that user is allowed to
see.
Args:
user_tuples (str, bool): (user id, is_peeking) for each user to be
checked. is_peeking should be true if:
* the user is not currently a member of the room, and:
* the user has not been a member of the room since the
given events
events ([synapse.events.EventBase]): list of events to filter
"""
forgotten = yield defer.gatherResults([
self.store.who_forgot_in_room(
room_id,
)
for room_id in frozenset(e.room_id for e in events)
], consumeErrors=True)
# Set of membership event_ids that have been forgotten
event_id_forgotten = frozenset(
row["event_id"] for rows in forgotten for row in rows
)
def allowed(event, user_id, is_peeking):
"""
Args:
event (synapse.events.EventBase): event to check
user_id (str)
is_peeking (bool)
"""
state = event_id_to_state[event.event_id]
# get the room_visibility at the time of the event.
visibility_event = state.get((EventTypes.RoomHistoryVisibility, ""), None)
if visibility_event:
visibility = visibility_event.content.get("history_visibility", "shared")
else:
visibility = "shared"
if visibility not in VISIBILITY_PRIORITY:
visibility = "shared"
# if it was world_readable, it's easy: everyone can read it
if visibility == "world_readable":
return True
# Always allow history visibility events on boundaries. This is done
# by setting the effective visibility to the least restrictive
# of the old vs new.
if event.type == EventTypes.RoomHistoryVisibility:
prev_content = event.unsigned.get("prev_content", {})
prev_visibility = prev_content.get("history_visibility", None)
if prev_visibility not in VISIBILITY_PRIORITY:
prev_visibility = "shared"
new_priority = VISIBILITY_PRIORITY.index(visibility)
old_priority = VISIBILITY_PRIORITY.index(prev_visibility)
if old_priority < new_priority:
visibility = prev_visibility
# likewise, if the event is the user's own membership event, use
# the 'most joined' membership
membership = None
if event.type == EventTypes.Member and event.state_key == user_id:
membership = event.content.get("membership", None)
if membership not in MEMBERSHIP_PRIORITY:
membership = "leave"
prev_content = event.unsigned.get("prev_content", {})
prev_membership = prev_content.get("membership", None)
if prev_membership not in MEMBERSHIP_PRIORITY:
prev_membership = "leave"
new_priority = MEMBERSHIP_PRIORITY.index(membership)
old_priority = MEMBERSHIP_PRIORITY.index(prev_membership)
if old_priority < new_priority:
membership = prev_membership
# otherwise, get the user's membership at the time of the event.
if membership is None:
membership_event = state.get((EventTypes.Member, user_id), None)
if membership_event:
if membership_event.event_id not in event_id_forgotten:
membership = membership_event.membership
# if the user was a member of the room at the time of the event,
# they can see it.
if membership == Membership.JOIN:
return True
if visibility == "joined":
# we weren't a member at the time of the event, so we can't
# see this event.
return False
elif visibility == "invited":
# user can also see the event if they were *invited* at the time
# of the event.
return membership == Membership.INVITE
else:
# visibility is shared: user can also see the event if they have
# become a member since the event
#
# XXX: if the user has subsequently joined and then left again,
# ideally we would share history up to the point they left. But
# we don't know when they left.
return not is_peeking
defer.returnValue({
user_id: [
event
for event in events
if allowed(event, user_id, is_peeking)
]
for user_id, is_peeking in user_tuples
})
@defer.inlineCallbacks
def _filter_events_for_client(self, user_id, events, is_peeking=False):
"""
Check which events a user is allowed to see
Args:
user_id(str): user id to be checked
events([synapse.events.EventBase]): list of events to be checked
is_peeking(bool): should be True if:
* the user is not currently a member of the room, and:
* the user has not been a member of the room since the given
events
Returns:
[synapse.events.EventBase]
"""
types = (
(EventTypes.RoomHistoryVisibility, ""),
(EventTypes.Member, user_id),
)
event_id_to_state = yield self.store.get_state_for_events(
frozenset(e.event_id for e in events),
types=types
)
res = yield self.filter_events_for_clients(
[(user_id, is_peeking)], events, event_id_to_state
)
defer.returnValue(res.get(user_id, []))
def ratelimit(self, requester):
time_now = self.clock.time()
user_id = requester.user.to_string()
# The AS user itself is never rate limited.
app_service = self.store.get_app_service_by_user_id(user_id)
if app_service is not None:
return # do not ratelimit app service senders
# Disable rate limiting of users belonging to any AS that is configured
# not to be rate limited in its registration file (rate_limited: true|false).
if requester.app_service and not requester.app_service.is_rate_limited():
return
allowed, time_allowed = self.ratelimiter.send_message(
user_id, time_now,
requester.user.to_string(), time_now,
msg_rate_hz=self.hs.config.rc_messages_per_second,
burst_count=self.hs.config.rc_message_burst_count,
)
@@ -78,20 +233,213 @@ class BaseHandler(object):
)
@defer.inlineCallbacks
def maybe_kick_guest_users(self, event, context=None):
def _create_new_client_event(self, builder, prev_event_ids=None):
if prev_event_ids:
prev_events = yield self.store.add_event_hashes(prev_event_ids)
prev_max_depth = yield self.store.get_max_depth_of_events(prev_event_ids)
depth = prev_max_depth + 1
else:
latest_ret = yield self.store.get_latest_event_ids_and_hashes_in_room(
builder.room_id,
)
if latest_ret:
depth = max([d for _, _, d in latest_ret]) + 1
else:
depth = 1
prev_events = [
(event_id, prev_hashes)
for event_id, prev_hashes, _ in latest_ret
]
builder.prev_events = prev_events
builder.depth = depth
state_handler = self.state_handler
context = yield state_handler.compute_event_context(builder)
if builder.is_state():
builder.prev_state = yield self.store.add_event_hashes(
context.prev_state_events
)
yield self.auth.add_auth_events(builder, context)
add_hashes_and_signatures(
builder, self.server_name, self.signing_key
)
event = builder.build()
logger.debug(
"Created event %s with current state: %s",
event.event_id, context.current_state,
)
defer.returnValue(
(event, context,)
)
def is_host_in_room(self, current_state):
room_members = [
(state_key, event.membership)
for ((event_type, state_key), event) in current_state.items()
if event_type == EventTypes.Member
]
if len(room_members) == 0:
# Have we just created the room, and is this about to be the very
# first member event?
create_event = current_state.get(("m.room.create", ""))
if create_event:
return True
for (state_key, membership) in room_members:
if (
UserID.from_string(state_key).domain == self.hs.hostname
and membership == Membership.JOIN
):
return True
return False
@defer.inlineCallbacks
def handle_new_client_event(
self,
requester,
event,
context,
ratelimit=True,
extra_users=[]
):
# We now need to go and hit out to wherever we need to hit out to.
if ratelimit:
self.ratelimit(requester)
self.auth.check(event, auth_events=context.current_state)
yield self.maybe_kick_guest_users(event, context.current_state.values())
if event.type == EventTypes.CanonicalAlias:
# Check the alias is acually valid (at this time at least)
room_alias_str = event.content.get("alias", None)
if room_alias_str:
room_alias = RoomAlias.from_string(room_alias_str)
directory_handler = self.hs.get_handlers().directory_handler
mapping = yield directory_handler.get_association(room_alias)
if mapping["room_id"] != event.room_id:
raise SynapseError(
400,
"Room alias %s does not point to the room" % (
room_alias_str,
)
)
federation_handler = self.hs.get_handlers().federation_handler
if event.type == EventTypes.Member:
if event.content["membership"] == Membership.INVITE:
def is_inviter_member_event(e):
return (
e.type == EventTypes.Member and
e.sender == event.sender
)
event.unsigned["invite_room_state"] = [
{
"type": e.type,
"state_key": e.state_key,
"content": e.content,
"sender": e.sender,
}
for k, e in context.current_state.items()
if e.type in self.hs.config.room_invite_state_types
or is_inviter_member_event(e)
]
invitee = UserID.from_string(event.state_key)
if not self.hs.is_mine(invitee):
# TODO: Can we add signature from remote server in a nicer
# way? If we have been invited by a remote server, we need
# to get them to sign the event.
returned_invite = yield federation_handler.send_invite(
invitee.domain,
event,
)
event.unsigned.pop("room_state", None)
# TODO: Make sure the signatures actually are correct.
event.signatures.update(
returned_invite.signatures
)
if event.type == EventTypes.Redaction:
if self.auth.check_redaction(event, auth_events=context.current_state):
original_event = yield self.store.get_event(
event.redacts,
check_redacted=False,
get_prev_content=False,
allow_rejected=False,
allow_none=False
)
if event.user_id != original_event.user_id:
raise AuthError(
403,
"You don't have permission to redact events"
)
if event.type == EventTypes.Create and context.current_state:
raise AuthError(
403,
"Changing the room create event is forbidden",
)
action_generator = ActionGenerator(self.hs)
yield action_generator.handle_push_actions_for_event(
event, context, self
)
(event_stream_id, max_stream_id) = yield self.store.persist_event(
event, context=context
)
destinations = set()
for k, s in context.current_state.items():
try:
if k[0] == EventTypes.Member:
if s.content["membership"] == Membership.JOIN:
destinations.add(
UserID.from_string(s.state_key).domain
)
except SynapseError:
logger.warn(
"Failed to get destination from event %s", s.event_id
)
with PreserveLoggingContext():
# Don't block waiting on waking up all the listeners.
self.notifier.on_new_room_event(
event, event_stream_id, max_stream_id,
extra_users=extra_users
)
# If invite, remove room_state from unsigned before sending.
event.unsigned.pop("invite_room_state", None)
federation_handler.handle_new_event(
event, destinations=destinations,
)
@defer.inlineCallbacks
def maybe_kick_guest_users(self, event, current_state):
# Technically this function invalidates current_state by changing it.
# Hopefully this isn't that important to the caller.
if event.type == EventTypes.GuestAccess:
guest_access = event.content.get("guest_access", "forbidden")
if guest_access != "can_join":
if context:
current_state = yield self.store.get_events(
context.current_state_ids.values()
)
current_state = current_state.values()
else:
current_state = yield self.store.get_current_state(event.room_id)
logger.info("maybe_kick_guest_users %r", current_state)
yield self.kick_guest_users(current_state)
@defer.inlineCallbacks
@@ -124,8 +472,7 @@ class BaseHandler(object):
# and having homeservers have their own users leave keeps more
# of that decision-making and control local to the guest-having
# homeserver.
requester = synapse.types.create_requester(
target_user, is_guest=True)
requester = Requester(target_user, "", True)
handler = self.hs.get_handlers().room_member_handler
yield handler.update_membership(
requester,

View File

@@ -16,8 +16,8 @@
from twisted.internet import defer
from synapse.api.constants import EventTypes
from synapse.util.metrics import Measure
from synapse.util.logcontext import preserve_fn, preserve_context_over_deferred
from synapse.appservice import ApplicationService
from synapse.types import UserID
import logging
@@ -35,81 +35,47 @@ def log_failure(failure):
)
# NB: Purposefully not inheriting BaseHandler since that contains way too much
# setup code which this handler does not need or use. This makes testing a lot
# easier.
class ApplicationServicesHandler(object):
def __init__(self, hs):
def __init__(self, hs, appservice_api, appservice_scheduler):
self.store = hs.get_datastore()
self.is_mine_id = hs.is_mine_id
self.appservice_api = hs.get_application_service_api()
self.scheduler = hs.get_application_service_scheduler()
self.hs = hs
self.appservice_api = appservice_api
self.scheduler = appservice_scheduler
self.started_scheduler = False
self.clock = hs.get_clock()
self.notify_appservices = hs.config.notify_appservices
self.current_max = 0
self.is_processing = False
@defer.inlineCallbacks
def notify_interested_services(self, current_id):
def notify_interested_services(self, event):
"""Notifies (pushes) all application services interested in this event.
Pushing is done asynchronously, so this method won't block for any
prolonged length of time.
Args:
current_id(int): The current maximum ID.
event(Event): The event to push out to interested services.
"""
services = self.store.get_app_services()
if not services or not self.notify_appservices:
return
# Gather interested services
services = yield self._get_services_for_event(event)
if len(services) == 0:
return # no services need notifying
self.current_max = max(self.current_max, current_id)
if self.is_processing:
return
# Do we know this user exists? If not, poke the user query API for
# all services which match that user regex. This needs to block as these
# user queries need to be made BEFORE pushing the event.
yield self._check_user_exists(event.sender)
if event.type == EventTypes.Member:
yield self._check_user_exists(event.state_key)
with Measure(self.clock, "notify_interested_services"):
self.is_processing = True
try:
upper_bound = self.current_max
limit = 100
while True:
upper_bound, events = yield self.store.get_new_events_for_appservice(
upper_bound, limit
)
if not self.started_scheduler:
self.scheduler.start().addErrback(log_failure)
self.started_scheduler = True
if not events:
break
for event in events:
# Gather interested services
services = yield self._get_services_for_event(event)
if len(services) == 0:
continue # no services need notifying
# Do we know this user exists? If not, poke the user
# query API for all services which match that user regex.
# This needs to block as these user queries need to be
# made BEFORE pushing the event.
yield self._check_user_exists(event.sender)
if event.type == EventTypes.Member:
yield self._check_user_exists(event.state_key)
if not self.started_scheduler:
self.scheduler.start().addErrback(log_failure)
self.started_scheduler = True
# Fork off pushes to these services
for service in services:
preserve_fn(self.scheduler.submit_event_for_as)(
service, event
)
yield self.store.set_appservice_last_pos(upper_bound)
if len(events) < limit:
break
finally:
self.is_processing = False
# Fork off pushes to these services
for service in services:
self.scheduler.submit_event_for_as(service, event)
@defer.inlineCallbacks
def query_user_exists(self, user_id):
@@ -142,12 +108,11 @@ class ApplicationServicesHandler(object):
association can be found.
"""
room_alias_str = room_alias.to_string()
services = self.store.get_app_services()
alias_query_services = [
s for s in services if (
s.is_interested_in_alias(room_alias_str)
)
]
alias_query_services = yield self._get_services_for_event(
event=None,
restrict_to=ApplicationService.NS_ALIASES,
alias_list=[room_alias_str]
)
for alias_service in alias_query_services:
is_known_alias = yield self.appservice_api.query_alias(
alias_service, room_alias_str
@@ -160,97 +125,52 @@ class ApplicationServicesHandler(object):
defer.returnValue(result)
@defer.inlineCallbacks
def query_3pe(self, kind, protocol, fields):
services = yield self._get_services_for_3pn(protocol)
results = yield preserve_context_over_deferred(defer.DeferredList([
preserve_fn(self.appservice_api.query_3pe)(service, kind, protocol, fields)
for service in services
], consumeErrors=True))
ret = []
for (success, result) in results:
if success:
ret.extend(result)
defer.returnValue(ret)
@defer.inlineCallbacks
def get_3pe_protocols(self, only_protocol=None):
services = self.store.get_app_services()
protocols = {}
# Collect up all the individual protocol responses out of the ASes
for s in services:
for p in s.protocols:
if only_protocol is not None and p != only_protocol:
continue
if p not in protocols:
protocols[p] = []
info = yield self.appservice_api.get_3pe_protocol(s, p)
if info is not None:
protocols[p].append(info)
def _merge_instances(infos):
if not infos:
return {}
# Merge the 'instances' lists of multiple results, but just take
# the other fields from the first as they ought to be identical
# copy the result so as not to corrupt the cached one
combined = dict(infos[0])
combined["instances"] = list(combined["instances"])
for info in infos[1:]:
combined["instances"].extend(info["instances"])
return combined
for p in protocols.keys():
protocols[p] = _merge_instances(protocols[p])
defer.returnValue(protocols)
@defer.inlineCallbacks
def _get_services_for_event(self, event):
def _get_services_for_event(self, event, restrict_to="", alias_list=None):
"""Retrieve a list of application services interested in this event.
Args:
event(Event): The event to check. Can be None if alias_list is not.
restrict_to(str): The namespace to restrict regex tests to.
alias_list: A list of aliases to get services for. If None, this
list is obtained from the database.
Returns:
list<ApplicationService>: A list of services interested in this
event based on the service regex.
"""
services = self.store.get_app_services()
member_list = None
if hasattr(event, "room_id"):
# We need to know the aliases associated with this event.room_id,
# if any.
if not alias_list:
alias_list = yield self.store.get_aliases_for_room(
event.room_id
)
# We need to know the members associated with this event.room_id,
# if any.
member_list = yield self.store.get_users_in_room(event.room_id)
services = yield self.store.get_app_services()
interested_list = [
s for s in services if (
yield s.is_interested(event, self.store)
s.is_interested(event, restrict_to, alias_list, member_list)
)
]
defer.returnValue(interested_list)
@defer.inlineCallbacks
def _get_services_for_user(self, user_id):
services = self.store.get_app_services()
services = yield self.store.get_app_services()
interested_list = [
s for s in services if (
s.is_interested_in_user(user_id)
)
]
return defer.succeed(interested_list)
def _get_services_for_3pn(self, protocol):
services = self.store.get_app_services()
interested_list = [
s for s in services if s.is_interested_in_protocol(protocol)
]
return defer.succeed(interested_list)
defer.returnValue(interested_list)
@defer.inlineCallbacks
def _is_unknown_user(self, user_id):
if not self.is_mine_id(user_id):
user = UserID.from_string(user_id)
if not self.hs.is_mine(user):
# we don't know if they are unknown or not since it isn't one of our
# users. We can't poke ASes.
defer.returnValue(False)
@@ -262,7 +182,7 @@ class ApplicationServicesHandler(object):
return
# user not found; could be the AS though, so check.
services = self.store.get_app_services()
services = yield self.store.get_app_services()
service_list = [s for s in services if s.sender == user_id]
defer.returnValue(len(service_list) == 0)

View File

@@ -18,7 +18,7 @@ from twisted.internet import defer
from ._base import BaseHandler
from synapse.api.constants import LoginType
from synapse.types import UserID
from synapse.api.errors import AuthError, LoginError, Codes, StoreError, SynapseError
from synapse.api.errors import AuthError, LoginError, Codes
from synapse.util.async import run_on_reactor
from twisted.web.client import PartialDownloadError
@@ -38,10 +38,6 @@ class AuthHandler(BaseHandler):
SESSION_EXPIRE_MS = 48 * 60 * 60 * 1000
def __init__(self, hs):
"""
Args:
hs (synapse.server.HomeServer):
"""
super(AuthHandler, self).__init__(hs)
self.checkers = {
LoginType.PASSWORD: self._check_password_auth,
@@ -51,20 +47,22 @@ class AuthHandler(BaseHandler):
}
self.bcrypt_rounds = hs.config.bcrypt_rounds
self.sessions = {}
self.INVALID_TOKEN_HTTP_STATUS = 401
account_handler = _AccountHandler(
hs, check_user_exists=self.check_user_exists
)
self.ldap_enabled = hs.config.ldap_enabled
self.ldap_server = hs.config.ldap_server
self.ldap_port = hs.config.ldap_port
self.ldap_tls = hs.config.ldap_tls
self.ldap_search_base = hs.config.ldap_search_base
self.ldap_search_property = hs.config.ldap_search_property
self.ldap_email_property = hs.config.ldap_email_property
self.ldap_full_name_property = hs.config.ldap_full_name_property
self.password_providers = [
module(config=config, account_handler=account_handler)
for module, config in hs.config.password_providers
]
logger.info("Extra password_providers: %r", self.password_providers)
if self.ldap_enabled is True:
import ldap
logger.info("Import ldap version: %s", ldap.__version__)
self.hs = hs # FIXME better possibility to access registrationHandler later?
self.device_handler = hs.get_device_handler()
@defer.inlineCallbacks
def check_auth(self, flows, clientdict, clientip):
@@ -135,47 +133,21 @@ class AuthHandler(BaseHandler):
creds = session['creds']
# check auth type currently being presented
errordict = {}
if 'type' in authdict:
login_type = authdict['type']
if login_type not in self.checkers:
if authdict['type'] not in self.checkers:
raise LoginError(400, "", Codes.UNRECOGNIZED)
try:
result = yield self.checkers[login_type](authdict, clientip)
if result:
creds[login_type] = result
self._save_session(session)
except LoginError, e:
if login_type == LoginType.EMAIL_IDENTITY:
# riot used to have a bug where it would request a new
# validation token (thus sending a new email) each time it
# got a 401 with a 'flows' field.
# (https://github.com/vector-im/vector-web/issues/2447).
#
# Grandfather in the old behaviour for now to avoid
# breaking old riot deployments.
raise e
# this step failed. Merge the error dict into the response
# so that the client can have another go.
errordict = e.error_dict()
result = yield self.checkers[authdict['type']](authdict, clientip)
if result:
creds[authdict['type']] = result
self._save_session(session)
for f in flows:
if len(set(f) - set(creds.keys())) == 0:
# it's very useful to know what args are stored, but this can
# include the password in the case of registering, so only log
# the keys (confusingly, clientdict may contain a password
# param, creds is just what the user authed as for UI auth
# and is not sensitive).
logger.info(
"Auth completed with creds: %r. Client dict has keys: %r",
creds, clientdict.keys()
)
logger.info("Auth completed with creds: %r", creds)
defer.returnValue((True, creds, clientdict, session['id']))
ret = self._auth_dict_for_flows(flows, session)
ret['completed'] = creds.keys()
ret.update(errordict)
defer.returnValue((False, ret, clientdict, session['id']))
@defer.inlineCallbacks
@@ -248,6 +220,7 @@ class AuthHandler(BaseHandler):
sess = self._get_session_info(session_id)
return sess.setdefault('serverdict', {}).get(key, default)
@defer.inlineCallbacks
def _check_password_auth(self, authdict, _):
if "user" not in authdict or "password" not in authdict:
raise LoginError(400, "", Codes.MISSING_PARAM)
@@ -257,7 +230,11 @@ class AuthHandler(BaseHandler):
if not user_id.startswith('@'):
user_id = UserID.create(user_id, self.hs.hostname).to_string()
return self._check_password(user_id, password)
if not (yield self._check_password(user_id, password)):
logger.warn("Failed password login for user %s", user_id)
raise LoginError(403, "", errcode=Codes.FORBIDDEN)
defer.returnValue(user_id)
@defer.inlineCallbacks
def _check_recaptcha(self, authdict, clientip):
@@ -293,17 +270,8 @@ class AuthHandler(BaseHandler):
data = pde.response
resp_body = simplejson.loads(data)
if 'success' in resp_body:
# Note that we do NOT check the hostname here: we explicitly
# intend the CAPTCHA to be presented by whatever client the
# user is using, we just care that they have completed a CAPTCHA.
logger.info(
"%s reCAPTCHA from hostname %s",
"Successful" if resp_body['success'] else "Failed",
resp_body.get('hostname')
)
if resp_body['success']:
defer.returnValue(True)
if 'success' in resp_body and resp_body['success']:
defer.returnValue(True)
raise LoginError(401, "", errcode=Codes.UNAUTHORIZED)
@defer.inlineCallbacks
@@ -370,205 +338,198 @@ class AuthHandler(BaseHandler):
return self.sessions[session_id]
def validate_password_login(self, user_id, password):
@defer.inlineCallbacks
def login_with_password(self, user_id, password):
"""
Authenticates the user with their username and password.
Used only by the v1 login API.
Args:
user_id (str): complete @user:id
user_id (str): User ID
password (str): Password
Returns:
defer.Deferred: (str) canonical user id
Raises:
StoreError if there was a problem accessing the database
LoginError if there was an authentication problem.
"""
return self._check_password(user_id, password)
@defer.inlineCallbacks
def get_access_token_for_user_id(self, user_id, device_id=None,
initial_display_name=None):
"""
Creates a new access token for the user with the given user ID.
The user is assumed to have been authenticated by some other
machanism (e.g. CAS), and the user_id converted to the canonical case.
The device will be recorded in the table if it is not there already.
Args:
user_id (str): canonical User ID
device_id (str|None): the device ID to associate with the tokens.
None to leave the tokens unassociated with a device (deprecated:
we should always have a device ID)
initial_display_name (str): display name to associate with the
device if it needs re-registering
Returns:
A tuple of:
The user's ID.
The access token for the user's session.
The refresh token for the user's session.
Raises:
StoreError if there was a problem storing the token.
LoginError if there was an authentication problem.
"""
logger.info("Logging in user %s on device %s", user_id, device_id)
access_token = yield self.issue_access_token(user_id, device_id)
# the device *should* have been registered before we got here; however,
# it's possible we raced against a DELETE operation. The thing we
# really don't want is active access_tokens without a record of the
# device, so we double-check it here.
if device_id is not None:
yield self.device_handler.check_device_registered(
user_id, device_id, initial_display_name
)
if not (yield self._check_password(user_id, password)):
logger.warn("Failed password login for user %s", user_id)
raise LoginError(403, "", errcode=Codes.FORBIDDEN)
defer.returnValue(access_token)
logger.info("Logging in user %s", user_id)
access_token = yield self.issue_access_token(user_id)
refresh_token = yield self.issue_refresh_token(user_id)
defer.returnValue((user_id, access_token, refresh_token))
@defer.inlineCallbacks
def check_user_exists(self, user_id):
def get_login_tuple_for_user_id(self, user_id):
"""
Checks to see if a user with the given id exists. Will check case
insensitively, but return None if there are multiple inexact matches.
Gets login tuple for the user with the given user ID.
The user is assumed to have been authenticated by some other
machanism (e.g. CAS)
Args:
(str) user_id: complete @user:id
user_id (str): User ID
Returns:
defer.Deferred: (str) canonical_user_id, or None if zero or
multiple matches
A tuple of:
The user's ID.
The access token for the user's session.
The refresh token for the user's session.
Raises:
StoreError if there was a problem storing the token.
LoginError if there was an authentication problem.
"""
res = yield self._find_user_id_and_pwd_hash(user_id)
if res is not None:
defer.returnValue(res[0])
defer.returnValue(None)
user_id, ignored = yield self._find_user_id_and_pwd_hash(user_id)
logger.info("Logging in user %s", user_id)
access_token = yield self.issue_access_token(user_id)
refresh_token = yield self.issue_refresh_token(user_id)
defer.returnValue((user_id, access_token, refresh_token))
@defer.inlineCallbacks
def does_user_exist(self, user_id):
try:
yield self._find_user_id_and_pwd_hash(user_id)
defer.returnValue(True)
except LoginError:
defer.returnValue(False)
@defer.inlineCallbacks
def _find_user_id_and_pwd_hash(self, user_id):
"""Checks to see if a user with the given id exists. Will check case
insensitively, but will return None if there are multiple inexact
matches.
insensitively, but will throw if there are multiple inexact matches.
Returns:
tuple: A 2-tuple of `(canonical_user_id, password_hash)`
None: if there is not exactly one match
"""
user_infos = yield self.store.get_users_by_id_case_insensitive(user_id)
result = None
if not user_infos:
logger.warn("Attempted to login as %s but they do not exist", user_id)
elif len(user_infos) == 1:
# a single match (possibly not exact)
result = user_infos.popitem()
elif user_id in user_infos:
# multiple matches, but one is exact
result = (user_id, user_infos[user_id])
raise LoginError(403, "", errcode=Codes.FORBIDDEN)
if len(user_infos) > 1:
if user_id not in user_infos:
logger.warn(
"Attempted to login as %s but it matches more than one user "
"inexactly: %r",
user_id, user_infos.keys()
)
raise LoginError(403, "", errcode=Codes.FORBIDDEN)
defer.returnValue((user_id, user_infos[user_id]))
else:
# multiple matches, none of them exact
logger.warn(
"Attempted to login as %s but it matches more than one user "
"inexactly: %r",
user_id, user_infos.keys()
)
defer.returnValue(result)
defer.returnValue(user_infos.popitem())
@defer.inlineCallbacks
def _check_password(self, user_id, password):
"""Authenticate a user against the LDAP and local databases.
user_id is checked case insensitively against the local database, but
will throw if there are multiple inexact matches.
Args:
user_id (str): complete @user:id
Returns:
(str) the canonical_user_id
Raises:
LoginError if login fails
"""
for provider in self.password_providers:
is_valid = yield provider.check_password(user_id, password)
if is_valid:
defer.returnValue(user_id)
canonical_user_id = yield self._check_local_password(user_id, password)
if canonical_user_id:
defer.returnValue(canonical_user_id)
# unknown username or invalid password. We raise a 403 here, but note
# that if we're doing user-interactive login, it turns all LoginErrors
# into a 401 anyway.
raise LoginError(
403, "Invalid password",
errcode=Codes.FORBIDDEN
)
defer.returnValue(
not (
(yield self._check_ldap_password(user_id, password))
or
(yield self._check_local_password(user_id, password))
))
@defer.inlineCallbacks
def _check_local_password(self, user_id, password):
"""Authenticate a user against the local password database.
user_id is checked case insensitively, but will return None if there are
multiple inexact matches.
Args:
user_id (str): complete @user:id
Returns:
(str) the canonical_user_id, or None if unknown user / bad password
"""
lookupres = yield self._find_user_id_and_pwd_hash(user_id)
if not lookupres:
defer.returnValue(None)
(user_id, password_hash) = lookupres
result = self.validate_hash(password, password_hash)
if not result:
logger.warn("Failed password login for user %s", user_id)
defer.returnValue(None)
defer.returnValue(user_id)
try:
user_id, password_hash = yield self._find_user_id_and_pwd_hash(user_id)
defer.returnValue(not self.validate_hash(password, password_hash))
except LoginError:
defer.returnValue(False)
@defer.inlineCallbacks
def issue_access_token(self, user_id, device_id=None):
def _check_ldap_password(self, user_id, password):
if self.ldap_enabled is not True:
logger.debug("LDAP not configured")
defer.returnValue(False)
import ldap
logger.info("Authenticating %s with LDAP" % user_id)
try:
ldap_url = "%s:%s" % (self.ldap_server, self.ldap_port)
logger.debug("Connecting LDAP server at %s" % ldap_url)
l = ldap.initialize(ldap_url)
if self.ldap_tls:
logger.debug("Initiating TLS")
self._connection.start_tls_s()
local_name = UserID.from_string(user_id).localpart
dn = "%s=%s, %s" % (
self.ldap_search_property,
local_name,
self.ldap_search_base)
logger.debug("DN for LDAP authentication: %s" % dn)
l.simple_bind_s(dn.encode('utf-8'), password.encode('utf-8'))
if not (yield self.does_user_exist(user_id)):
handler = self.hs.get_handlers().registration_handler
user_id, access_token = (
yield handler.register(localpart=local_name)
)
defer.returnValue(True)
except ldap.LDAPError, e:
logger.warn("LDAP error: %s", e)
defer.returnValue(False)
@defer.inlineCallbacks
def issue_access_token(self, user_id):
access_token = self.generate_access_token(user_id)
yield self.store.add_access_token_to_user(user_id, access_token,
device_id)
yield self.store.add_access_token_to_user(user_id, access_token)
defer.returnValue(access_token)
@defer.inlineCallbacks
def issue_refresh_token(self, user_id):
refresh_token = self.generate_refresh_token(user_id)
yield self.store.add_refresh_token_to_user(user_id, refresh_token)
defer.returnValue(refresh_token)
def generate_access_token(self, user_id, extra_caveats=None):
extra_caveats = extra_caveats or []
macaroon = self._generate_base_macaroon(user_id)
macaroon.add_first_party_caveat("type = access")
# Include a nonce, to make sure that each login gets a different
# access token.
macaroon.add_first_party_caveat("nonce = %s" % (
stringutils.random_string_with_symbols(16),
))
now = self.hs.get_clock().time_msec()
expiry = now + (60 * 60 * 1000)
macaroon.add_first_party_caveat("time < %d" % (expiry,))
for caveat in extra_caveats:
macaroon.add_first_party_caveat(caveat)
return macaroon.serialize()
def generate_short_term_login_token(self, user_id, duration_in_ms=(2 * 60 * 1000)):
def generate_refresh_token(self, user_id):
m = self._generate_base_macaroon(user_id)
m.add_first_party_caveat("type = refresh")
# Important to add a nonce, because otherwise every refresh token for a
# user will be the same.
m.add_first_party_caveat("nonce = %s" % (
stringutils.random_string_with_symbols(16),
))
return m.serialize()
def generate_short_term_login_token(self, user_id):
macaroon = self._generate_base_macaroon(user_id)
macaroon.add_first_party_caveat("type = login")
now = self.hs.get_clock().time_msec()
expiry = now + duration_in_ms
expiry = now + (2 * 60 * 1000)
macaroon.add_first_party_caveat("time < %d" % (expiry,))
return macaroon.serialize()
def generate_delete_pusher_token(self, user_id):
macaroon = self._generate_base_macaroon(user_id)
macaroon.add_first_party_caveat("type = delete_pusher")
return macaroon.serialize()
def validate_short_term_login_token_and_get_user_id(self, login_token):
auth_api = self.hs.get_auth()
try:
macaroon = pymacaroons.Macaroon.deserialize(login_token)
user_id = auth_api.get_user_id_from_macaroon(macaroon)
auth_api.validate_macaroon(macaroon, "login", True, user_id)
return user_id
except Exception:
raise AuthError(403, "Invalid token", errcode=Codes.FORBIDDEN)
auth_api = self.hs.get_auth()
auth_api.validate_macaroon(macaroon, "login", True)
return self.get_user_from_macaroon(macaroon)
except (pymacaroons.exceptions.MacaroonException, TypeError, ValueError):
raise AuthError(401, "Invalid token", errcode=Codes.UNKNOWN_TOKEN)
def _generate_base_macaroon(self, user_id):
macaroon = pymacaroons.Macaroon(
@@ -579,39 +540,32 @@ class AuthHandler(BaseHandler):
macaroon.add_first_party_caveat("user_id = %s" % (user_id,))
return macaroon
def get_user_from_macaroon(self, macaroon):
user_prefix = "user_id = "
for caveat in macaroon.caveats:
if caveat.caveat_id.startswith(user_prefix):
return caveat.caveat_id[len(user_prefix):]
raise AuthError(
self.INVALID_TOKEN_HTTP_STATUS, "No user_id found in token",
errcode=Codes.UNKNOWN_TOKEN
)
@defer.inlineCallbacks
def set_password(self, user_id, newpassword, requester=None):
password_hash = self.hash(newpassword)
except_access_token_id = requester.access_token_id if requester else None
except_access_token_ids = [requester.access_token_id] if requester else []
try:
yield self.store.user_set_password_hash(user_id, password_hash)
except StoreError as e:
if e.code == 404:
raise SynapseError(404, "Unknown user", Codes.NOT_FOUND)
raise e
yield self.store.user_set_password_hash(user_id, password_hash)
yield self.store.user_delete_access_tokens(
user_id, except_access_token_id
user_id, except_access_token_ids
)
yield self.hs.get_pusherpool().remove_pushers_by_user(
user_id, except_access_token_id
user_id, except_access_token_ids
)
@defer.inlineCallbacks
def add_threepid(self, user_id, medium, address, validated_at):
# 'Canonicalise' email addresses down to lower case.
# We've now moving towards the Home Server being the entity that
# is responsible for validating threepids used for resetting passwords
# on accounts, so in future Synapse will gain knowledge of specific
# types (mediums) of threepid. For now, we still use the existing
# infrastructure, but this is the start of synapse gaining knowledge
# of specific types of threepid (and fixes the fact that checking
# for the presenc eof an email address during password reset was
# case sensitive).
if medium == 'email':
address = address.lower()
yield self.store.user_add_threepid(
user_id, medium, address, validated_at,
self.hs.get_clock().time_msec()
@@ -642,8 +596,7 @@ class AuthHandler(BaseHandler):
Returns:
Hashed password (str).
"""
return bcrypt.hashpw(password.encode('utf8') + self.hs.config.password_pepper,
bcrypt.gensalt(self.bcrypt_rounds))
return bcrypt.hashpw(password, bcrypt.gensalt(self.bcrypt_rounds))
def validate_hash(self, password, stored_hash):
"""Validates that self.hash(password) == stored_hash.
@@ -655,35 +608,4 @@ class AuthHandler(BaseHandler):
Returns:
Whether self.hash(password) == stored_hash (bool).
"""
if stored_hash:
return bcrypt.hashpw(password + self.hs.config.password_pepper,
stored_hash.encode('utf-8')) == stored_hash
else:
return False
class _AccountHandler(object):
"""A proxy object that gets passed to password auth providers so they
can register new users etc if necessary.
"""
def __init__(self, hs, check_user_exists):
self.hs = hs
self._check_user_exists = check_user_exists
def check_user_exists(self, user_id):
"""Check if user exissts.
Returns:
Deferred(bool)
"""
return self._check_user_exists(user_id)
def register(self, localpart):
"""Registers a new user with given localpart
Returns:
Deferred: a 2-tuple of (user_id, access_token)
"""
reg = self.hs.get_handlers().registration_handler
return reg.register(localpart=localpart)
return bcrypt.hashpw(password, stored_hash) == stored_hash

View File

@@ -1,181 +0,0 @@
# -*- coding: utf-8 -*-
# Copyright 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from synapse.api import errors
from synapse.util import stringutils
from twisted.internet import defer
from ._base import BaseHandler
import logging
logger = logging.getLogger(__name__)
class DeviceHandler(BaseHandler):
def __init__(self, hs):
super(DeviceHandler, self).__init__(hs)
@defer.inlineCallbacks
def check_device_registered(self, user_id, device_id,
initial_device_display_name=None):
"""
If the given device has not been registered, register it with the
supplied display name.
If no device_id is supplied, we make one up.
Args:
user_id (str): @user:id
device_id (str | None): device id supplied by client
initial_device_display_name (str | None): device display name from
client
Returns:
str: device id (generated if none was supplied)
"""
if device_id is not None:
yield self.store.store_device(
user_id=user_id,
device_id=device_id,
initial_device_display_name=initial_device_display_name,
ignore_if_known=True,
)
defer.returnValue(device_id)
# if the device id is not specified, we'll autogen one, but loop a few
# times in case of a clash.
attempts = 0
while attempts < 5:
try:
device_id = stringutils.random_string(10).upper()
yield self.store.store_device(
user_id=user_id,
device_id=device_id,
initial_device_display_name=initial_device_display_name,
ignore_if_known=False,
)
defer.returnValue(device_id)
except errors.StoreError:
attempts += 1
raise errors.StoreError(500, "Couldn't generate a device ID.")
@defer.inlineCallbacks
def get_devices_by_user(self, user_id):
"""
Retrieve the given user's devices
Args:
user_id (str):
Returns:
defer.Deferred: list[dict[str, X]]: info on each device
"""
device_map = yield self.store.get_devices_by_user(user_id)
ips = yield self.store.get_last_client_ip_by_device(
devices=((user_id, device_id) for device_id in device_map.keys())
)
devices = device_map.values()
for device in devices:
_update_device_from_client_ips(device, ips)
defer.returnValue(devices)
@defer.inlineCallbacks
def get_device(self, user_id, device_id):
""" Retrieve the given device
Args:
user_id (str):
device_id (str):
Returns:
defer.Deferred: dict[str, X]: info on the device
Raises:
errors.NotFoundError: if the device was not found
"""
try:
device = yield self.store.get_device(user_id, device_id)
except errors.StoreError:
raise errors.NotFoundError
ips = yield self.store.get_last_client_ip_by_device(
devices=((user_id, device_id),)
)
_update_device_from_client_ips(device, ips)
defer.returnValue(device)
@defer.inlineCallbacks
def delete_device(self, user_id, device_id):
""" Delete the given device
Args:
user_id (str):
device_id (str):
Returns:
defer.Deferred:
"""
try:
yield self.store.delete_device(user_id, device_id)
except errors.StoreError, e:
if e.code == 404:
# no match
pass
else:
raise
yield self.store.user_delete_access_tokens(
user_id, device_id=device_id,
delete_refresh_tokens=True,
)
yield self.store.delete_e2e_keys_by_device(
user_id=user_id, device_id=device_id
)
@defer.inlineCallbacks
def update_device(self, user_id, device_id, content):
""" Update the given device
Args:
user_id (str):
device_id (str):
content (dict): body of update request
Returns:
defer.Deferred:
"""
try:
yield self.store.update_device(
user_id,
device_id,
new_display_name=content.get("display_name")
)
except errors.StoreError, e:
if e.code == 404:
raise errors.NotFoundError()
else:
raise
def _update_device_from_client_ips(device, client_ips):
ip = client_ips.get((device["user_id"], device["device_id"]), {})
device.update({
"last_seen_ts": ip.get("last_seen"),
"last_seen_ip": ip.get("ip"),
})

View File

@@ -1,117 +0,0 @@
# -*- coding: utf-8 -*-
# Copyright 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import logging
from twisted.internet import defer
from synapse.types import get_domain_from_id
from synapse.util.stringutils import random_string
logger = logging.getLogger(__name__)
class DeviceMessageHandler(object):
def __init__(self, hs):
"""
Args:
hs (synapse.server.HomeServer): server
"""
self.store = hs.get_datastore()
self.notifier = hs.get_notifier()
self.is_mine_id = hs.is_mine_id
self.federation = hs.get_federation_sender()
hs.get_replication_layer().register_edu_handler(
"m.direct_to_device", self.on_direct_to_device_edu
)
@defer.inlineCallbacks
def on_direct_to_device_edu(self, origin, content):
local_messages = {}
sender_user_id = content["sender"]
if origin != get_domain_from_id(sender_user_id):
logger.warn(
"Dropping device message from %r with spoofed sender %r",
origin, sender_user_id
)
message_type = content["type"]
message_id = content["message_id"]
for user_id, by_device in content["messages"].items():
messages_by_device = {
device_id: {
"content": message_content,
"type": message_type,
"sender": sender_user_id,
}
for device_id, message_content in by_device.items()
}
if messages_by_device:
local_messages[user_id] = messages_by_device
stream_id = yield self.store.add_messages_from_remote_to_device_inbox(
origin, message_id, local_messages
)
self.notifier.on_new_event(
"to_device_key", stream_id, users=local_messages.keys()
)
@defer.inlineCallbacks
def send_device_message(self, sender_user_id, message_type, messages):
local_messages = {}
remote_messages = {}
for user_id, by_device in messages.items():
if self.is_mine_id(user_id):
messages_by_device = {
device_id: {
"content": message_content,
"type": message_type,
"sender": sender_user_id,
}
for device_id, message_content in by_device.items()
}
if messages_by_device:
local_messages[user_id] = messages_by_device
else:
destination = get_domain_from_id(user_id)
remote_messages.setdefault(destination, {})[user_id] = by_device
message_id = random_string(16)
remote_edu_contents = {}
for destination, messages in remote_messages.items():
remote_edu_contents[destination] = {
"messages": messages,
"sender": sender_user_id,
"type": message_type,
"message_id": message_id,
}
stream_id = yield self.store.add_messages_to_device_inbox(
local_messages, remote_edu_contents
)
self.notifier.on_new_event(
"to_device_key", stream_id, users=local_messages.keys()
)
for destination in remote_messages.keys():
# Enqueue a new federation transaction to send the new
# device messages to each remote destination.
self.federation.send_device_messages(destination)

View File

@@ -19,7 +19,7 @@ from ._base import BaseHandler
from synapse.api.errors import SynapseError, Codes, CodeMessageException, AuthError
from synapse.api.constants import EventTypes
from synapse.types import RoomAlias, UserID, get_domain_from_id
from synapse.types import RoomAlias, UserID
import logging
import string
@@ -33,7 +33,6 @@ class DirectoryHandler(BaseHandler):
super(DirectoryHandler, self).__init__(hs)
self.state = hs.get_state_handler()
self.appservice_handler = hs.get_application_service_handler()
self.federation = hs.get_replication_layer()
self.federation.register_query_handler(
@@ -55,8 +54,7 @@ class DirectoryHandler(BaseHandler):
# TODO(erikj): Add transactions.
# TODO(erikj): Check if there is a current association.
if not servers:
users = yield self.state.get_current_user_in_room(room_id)
servers = set(get_domain_from_id(u) for u in users)
servers = yield self.store.get_joined_hosts_for_room(room_id)
if not servers:
raise SynapseError(400, "Failed to get server list")
@@ -194,8 +192,7 @@ class DirectoryHandler(BaseHandler):
Codes.NOT_FOUND
)
users = yield self.state.get_current_user_in_room(room_id)
extra_servers = set(get_domain_from_id(u) for u in users)
extra_servers = yield self.store.get_joined_hosts_for_room(room_id)
servers = set(extra_servers) | set(servers)
# If this server is in the list of servers, return it first.
@@ -284,16 +281,17 @@ class DirectoryHandler(BaseHandler):
)
if not result:
# Query AS to see if it exists
as_handler = self.appservice_handler
as_handler = self.hs.get_handlers().appservice_handler
result = yield as_handler.query_room_alias_exists(room_alias)
defer.returnValue(result)
@defer.inlineCallbacks
def can_modify_alias(self, alias, user_id=None):
# Any application service "interested" in an alias they are regexing on
# can modify the alias.
# Users can only modify the alias if ALL the interested services have
# non-exclusive locks on the alias (or there are no interested services)
services = self.store.get_app_services()
services = yield self.store.get_app_services()
interested_services = [
s for s in services if s.is_interested_in_alias(alias.to_string())
]
@@ -301,12 +299,14 @@ class DirectoryHandler(BaseHandler):
for service in interested_services:
if user_id == service.sender:
# this user IS the app service so they can do whatever they like
return defer.succeed(True)
defer.returnValue(True)
return
elif service.is_exclusive_alias(alias.to_string()):
# another service has an exclusive lock on this alias.
return defer.succeed(False)
defer.returnValue(False)
return
# either no interested services, or no service with an exclusive lock
return defer.succeed(True)
defer.returnValue(True)
@defer.inlineCallbacks
def _user_can_delete_alias(self, alias, user_id):
@@ -339,22 +339,3 @@ class DirectoryHandler(BaseHandler):
yield self.auth.check_can_change_room_list(room_id, requester.user)
yield self.store.set_room_is_public(room_id, visibility == "public")
@defer.inlineCallbacks
def edit_published_appservice_room_list(self, appservice_id, network_id,
room_id, visibility):
"""Add or remove a room from the appservice/network specific public
room list.
Args:
appservice_id (str): ID of the appservice that owns the list
network_id (str): The ID of the network the list is associated with
room_id (str)
visibility (str): either "public" or "private"
"""
if visibility not in ["public", "private"]:
raise SynapseError(400, "Invalid visibility setting")
yield self.store.set_room_is_public_appservice(
room_id, appservice_id, network_id, visibility == "public"
)

View File

@@ -1,289 +0,0 @@
# -*- coding: utf-8 -*-
# Copyright 2016 OpenMarket Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import ujson as json
import logging
from canonicaljson import encode_canonical_json
from twisted.internet import defer
from synapse.api.errors import SynapseError, CodeMessageException
from synapse.types import get_domain_from_id
from synapse.util.logcontext import preserve_fn, preserve_context_over_deferred
from synapse.util.retryutils import get_retry_limiter, NotRetryingDestination
logger = logging.getLogger(__name__)
class E2eKeysHandler(object):
def __init__(self, hs):
self.store = hs.get_datastore()
self.federation = hs.get_replication_layer()
self.device_handler = hs.get_device_handler()
self.is_mine_id = hs.is_mine_id
self.clock = hs.get_clock()
# doesn't really work as part of the generic query API, because the
# query request requires an object POST, but we abuse the
# "query handler" interface.
self.federation.register_query_handler(
"client_keys", self.on_federation_query_client_keys
)
@defer.inlineCallbacks
def query_devices(self, query_body, timeout):
""" Handle a device key query from a client
{
"device_keys": {
"<user_id>": ["<device_id>"]
}
}
->
{
"device_keys": {
"<user_id>": {
"<device_id>": {
...
}
}
}
}
"""
device_keys_query = query_body.get("device_keys", {})
# separate users by domain.
# make a map from domain to user_id to device_ids
local_query = {}
remote_queries = {}
for user_id, device_ids in device_keys_query.items():
if self.is_mine_id(user_id):
local_query[user_id] = device_ids
else:
domain = get_domain_from_id(user_id)
remote_queries.setdefault(domain, {})[user_id] = device_ids
# do the queries
failures = {}
results = {}
if local_query:
local_result = yield self.query_local_devices(local_query)
for user_id, keys in local_result.items():
if user_id in local_query:
results[user_id] = keys
@defer.inlineCallbacks
def do_remote_query(destination):
destination_query = remote_queries[destination]
try:
limiter = yield get_retry_limiter(
destination, self.clock, self.store
)
with limiter:
remote_result = yield self.federation.query_client_keys(
destination,
{"device_keys": destination_query},
timeout=timeout
)
for user_id, keys in remote_result["device_keys"].items():
if user_id in destination_query:
results[user_id] = keys
except CodeMessageException as e:
failures[destination] = {
"status": e.code, "message": e.message
}
except NotRetryingDestination as e:
failures[destination] = {
"status": 503, "message": "Not ready for retry",
}
except Exception as e:
# include ConnectionRefused and other errors
failures[destination] = {
"status": 503, "message": e.message
}
yield preserve_context_over_deferred(defer.gatherResults([
preserve_fn(do_remote_query)(destination)
for destination in remote_queries
]))
defer.returnValue({
"device_keys": results, "failures": failures,
})
@defer.inlineCallbacks
def query_local_devices(self, query):
"""Get E2E device keys for local users
Args:
query (dict[string, list[string]|None): map from user_id to a list
of devices to query (None for all devices)
Returns:
defer.Deferred: (resolves to dict[string, dict[string, dict]]):
map from user_id -> device_id -> device details
"""
local_query = []
result_dict = {}
for user_id, device_ids in query.items():
if not self.is_mine_id(user_id):
logger.warning("Request for keys for non-local user %s",
user_id)
raise SynapseError(400, "Not a user here")
if not device_ids:
local_query.append((user_id, None))
else:
for device_id in device_ids:
local_query.append((user_id, device_id))
# make sure that each queried user appears in the result dict
result_dict[user_id] = {}
results = yield self.store.get_e2e_device_keys(local_query)
# Build the result structure, un-jsonify the results, and add the
# "unsigned" section
for user_id, device_keys in results.items():
for device_id, device_info in device_keys.items():
r = json.loads(device_info["key_json"])
r["unsigned"] = {}
display_name = device_info["device_display_name"]
if display_name is not None:
r["unsigned"]["device_display_name"] = display_name
result_dict[user_id][device_id] = r
defer.returnValue(result_dict)
@defer.inlineCallbacks
def on_federation_query_client_keys(self, query_body):
""" Handle a device key query from a federated server
"""
device_keys_query = query_body.get("device_keys", {})
res = yield self.query_local_devices(device_keys_query)
defer.returnValue({"device_keys": res})
@defer.inlineCallbacks
def claim_one_time_keys(self, query, timeout):
local_query = []
remote_queries = {}
for user_id, device_keys in query.get("one_time_keys", {}).items():
if self.is_mine_id(user_id):
for device_id, algorithm in device_keys.items():
local_query.append((user_id, device_id, algorithm))
else:
domain = get_domain_from_id(user_id)
remote_queries.setdefault(domain, {})[user_id] = device_keys
results = yield self.store.claim_e2e_one_time_keys(local_query)
json_result = {}
failures = {}
for user_id, device_keys in results.items():
for device_id, keys in device_keys.items():
for key_id, json_bytes in keys.items():
json_result.setdefault(user_id, {})[device_id] = {
key_id: json.loads(json_bytes)
}
@defer.inlineCallbacks
def claim_client_keys(destination):
device_keys = remote_queries[destination]
try:
limiter = yield get_retry_limiter(
destination, self.clock, self.store
)
with limiter:
remote_result = yield self.federation.claim_client_keys(
destination,
{"one_time_keys": device_keys},
timeout=timeout
)
for user_id, keys in remote_result["one_time_keys"].items():
if user_id in device_keys:
json_result[user_id] = keys
except CodeMessageException as e:
failures[destination] = {
"status": e.code, "message": e.message
}
except NotRetryingDestination as e:
failures[destination] = {
"status": 503, "message": "Not ready for retry",
}
except Exception as e:
# include ConnectionRefused and other errors
failures[destination] = {
"status": 503, "message": e.message
}
yield preserve_context_over_deferred(defer.gatherResults([
preserve_fn(claim_client_keys)(destination)
for destination in remote_queries
]))
defer.returnValue({
"one_time_keys": json_result,
"failures": failures
})
@defer.inlineCallbacks
def upload_keys_for_user(self, user_id, device_id, keys):
time_now = self.clock.time_msec()
# TODO: Validate the JSON to make sure it has the right keys.
device_keys = keys.get("device_keys", None)
if device_keys:
logger.info(
"Updating device_keys for device %r for user %s at %d",
device_id, user_id, time_now
)
# TODO: Sign the JSON with the server key
yield self.store.set_e2e_device_keys(
user_id, device_id, time_now,
encode_canonical_json(device_keys)
)
one_time_keys = keys.get("one_time_keys", None)
if one_time_keys:
logger.info(
"Adding %d one_time_keys for device %r for user %r at %d",
len(one_time_keys), device_id, user_id, time_now
)
key_list = []
for key_id, key_json in one_time_keys.items():
algorithm, key_id = key_id.split(":")
key_list.append((
algorithm, key_id, encode_canonical_json(key_json)
))
yield self.store.add_e2e_one_time_keys(
user_id, device_id, time_now, key_list
)
# the device should have been registered already, but it may have been
# deleted due to a race with a DELETE request. Or we may be using an
# old access_token without an associated device_id. Either way, we
# need to double-check the device is registered to avoid ending up with
# keys without a corresponding device.
self.device_handler.check_device_registered(user_id, device_id)
result = yield self.store.count_e2e_one_time_keys(user_id, device_id)
defer.returnValue({"one_time_key_counts": result})

View File

@@ -47,7 +47,6 @@ class EventStreamHandler(BaseHandler):
self.clock = hs.get_clock()
self.notifier = hs.get_notifier()
self.state = hs.get_state_handler()
@defer.inlineCallbacks
@log_function
@@ -59,7 +58,7 @@ class EventStreamHandler(BaseHandler):
If `only_keys` is not None, events from keys will be sent down.
"""
auth_user = UserID.from_string(auth_user_id)
presence_handler = self.hs.get_presence_handler()
presence_handler = self.hs.get_handlers().presence_handler
context = yield presence_handler.user_syncing(
auth_user_id, affect_presence=affect_presence,
@@ -91,7 +90,7 @@ class EventStreamHandler(BaseHandler):
# Send down presence.
if event.state_key == auth_user_id:
# Send down presence for everyone in the room.
users = yield self.state.get_current_user_in_room(event.room_id)
users = yield self.store.get_users_in_room(event.room_id)
states = yield presence_handler.get_states(
users,
as_event=True,

View File

@@ -26,17 +26,14 @@ from synapse.api.errors import (
from synapse.api.constants import EventTypes, Membership, RejectedReason
from synapse.events.validator import EventValidator
from synapse.util import unwrapFirstError
from synapse.util.logcontext import (
PreserveLoggingContext, preserve_fn, preserve_context_over_deferred
)
from synapse.util.metrics import measure_func
from synapse.util.logcontext import PreserveLoggingContext
from synapse.util.logutils import log_function
from synapse.util.async import run_on_reactor
from synapse.util.frozenutils import unfreeze
from synapse.crypto.event_signing import (
compute_event_signature, add_hashes_and_signatures,
)
from synapse.types import UserID, get_domain_from_id
from synapse.types import UserID
from synapse.events.utils import prune_event
@@ -69,6 +66,10 @@ class FederationHandler(BaseHandler):
self.hs = hs
self.distributor.observe("user_joined_room", self.user_joined_room)
self.waiting_for_join_list = {}
self.store = hs.get_datastore()
self.replication_layer = hs.get_replication_layer()
self.state_handler = hs.get_state_handler()
@@ -80,14 +81,27 @@ class FederationHandler(BaseHandler):
# When joining a room we need to queue any events for that room up
self.room_queues = {}
def handle_new_event(self, event, destinations):
""" Takes in an event from the client to server side, that has already
been authed and handled by the state module, and sends it to any
remote home servers that may be interested.
Args:
event: The event to send
destinations: A list of destinations to send it to
Returns:
Deferred: Resolved when it has successfully been queued for
processing.
"""
return self.replication_layer.send_pdu(event, destinations)
@log_function
@defer.inlineCallbacks
def on_receive_pdu(self, origin, pdu, state=None, auth_chain=None):
""" Called by the ReplicationLayer when we have a new pdu. We need to
do auth checks and put it through the StateHandler.
auth_chain and state are None if we already have the necessary state
and prev_events in the db
"""
event = pdu
@@ -105,25 +119,16 @@ class FederationHandler(BaseHandler):
# FIXME (erikj): Awful hack to make the case where we are not currently
# in the room work
# If state and auth_chain are None, then we don't need to do this check
# as we already know we have enough state in the DB to handle this
# event.
if state and auth_chain and not event.internal_metadata.is_outlier():
is_in_room = yield self.auth.check_host_in_room(
event.room_id,
self.server_name
)
else:
is_in_room = True
if not is_in_room:
logger.info(
"Got event for room we're not in: %r %r",
event.room_id, event.event_id
)
is_in_room = yield self.auth.check_host_in_room(
event.room_id,
self.server_name
)
if not is_in_room and not event.internal_metadata.is_outlier():
logger.debug("Got event for room we're not in.")
try:
event_stream_id, max_stream_id = yield self._persist_auth_tree(
origin, auth_chain, state, event
auth_chain, state, event
)
except AuthError as e:
raise FederationError(
@@ -214,28 +219,17 @@ class FederationHandler(BaseHandler):
if event.type == EventTypes.Member:
if event.membership == Membership.JOIN:
# Only fire user_joined_room if the user has acutally
# joined the room. Don't bother if the user is just
# changing their profile info.
newly_joined = True
prev_state_id = context.prev_state_ids.get(
(event.type, event.state_key)
)
if prev_state_id:
prev_state = yield self.store.get_event(
prev_state_id, allow_none=True,
)
if prev_state and prev_state.membership == Membership.JOIN:
newly_joined = False
if newly_joined:
prev_state = context.current_state.get((event.type, event.state_key))
if not prev_state or prev_state.membership != Membership.JOIN:
# Only fire user_joined_room if the user has acutally
# joined the room. Don't bother if the user is just
# changing their profile info.
user = UserID.from_string(event.state_key)
yield user_joined_room(self.distributor, user, event.room_id)
@measure_func("_filter_events_for_server")
@defer.inlineCallbacks
def _filter_events_for_server(self, server_name, room_id, events):
event_to_state_ids = yield self.store.get_state_ids_for_events(
event_to_state = yield self.store.get_state_for_events(
frozenset(e.event_id for e in events),
types=(
(EventTypes.RoomHistoryVisibility, ""),
@@ -243,33 +237,6 @@ class FederationHandler(BaseHandler):
)
)
# We only want to pull out member events that correspond to the
# server's domain.
def check_match(id):
try:
return server_name == get_domain_from_id(id)
except:
return False
# Parses mapping `event_id -> (type, state_key) -> state event_id`
# to get all state ids that we're interested in.
event_map = yield self.store.get_events([
e_id
for key_to_eid in event_to_state_ids.values()
for key, e_id in key_to_eid.items()
if key[0] != EventTypes.Member or check_match(key[1])
])
event_to_state = {
e_id: {
key: event_map[inner_e_id]
for key, inner_e_id in key_to_eid.items()
if inner_e_id in event_map
}
for e_id, key_to_eid in event_to_state_ids.items()
}
def redact_disallowed(event, state):
if not state:
return event
@@ -286,7 +253,7 @@ class FederationHandler(BaseHandler):
if ev.type != EventTypes.Member:
continue
try:
domain = get_domain_from_id(ev.state_key)
domain = UserID.from_string(ev.state_key).domain
except:
continue
@@ -311,16 +278,15 @@ class FederationHandler(BaseHandler):
@log_function
@defer.inlineCallbacks
def backfill(self, dest, room_id, limit, extremities):
def backfill(self, dest, room_id, limit, extremities=[]):
""" Trigger a backfill request to `dest` for the given `room_id`
This will attempt to get more events from the remote. This may return
be successfull and still return no events if the other side has no new
events to offer.
"""
if dest == self.server_name:
raise SynapseError(400, "Can't backfill from self.")
if not extremities:
extremities = yield self.store.get_oldest_events_in_room(room_id)
events = yield self.replication_layer.backfill(
dest,
room_id,
@@ -328,16 +294,6 @@ class FederationHandler(BaseHandler):
extremities=extremities,
)
# Don't bother processing events we already have.
seen_events = yield self.store.have_events_in_timeline(
set(e.event_id for e in events)
)
events = [e for e in events if e.event_id not in seen_events]
if not events:
defer.returnValue([])
event_map = {e.event_id: e for e in events}
event_ids = set(e.event_id for e in events)
@@ -369,73 +325,40 @@ class FederationHandler(BaseHandler):
state_events.update({s.event_id: s for s in state})
events_to_state[e_id] = state
required_auth = set(
a_id
for event in events + state_events.values() + auth_events.values()
for a_id, _ in event.auth_events
)
auth_events.update({
e_id: event_map[e_id] for e_id in required_auth if e_id in event_map
})
missing_auth = required_auth - set(auth_events)
failed_to_fetch = set()
# Try and fetch any missing auth events from both DB and remote servers.
# We repeatedly do this until we stop finding new auth events.
while missing_auth - failed_to_fetch:
logger.info("Missing auth for backfill: %r", missing_auth)
ret_events = yield self.store.get_events(missing_auth - failed_to_fetch)
auth_events.update(ret_events)
required_auth.update(
a_id for event in ret_events.values() for a_id, _ in event.auth_events
)
missing_auth = required_auth - set(auth_events)
if missing_auth - failed_to_fetch:
logger.info(
"Fetching missing auth for backfill: %r",
missing_auth - failed_to_fetch
)
results = yield preserve_context_over_deferred(defer.gatherResults(
[
preserve_fn(self.replication_layer.get_pdu)(
[dest],
event_id,
outlier=True,
timeout=10000,
)
for event_id in missing_auth - failed_to_fetch
],
consumeErrors=True
)).addErrback(unwrapFirstError)
auth_events.update({a.event_id: a for a in results if a})
required_auth.update(
a_id
for event in results if event
for a_id, _ in event.auth_events
)
missing_auth = required_auth - set(auth_events)
failed_to_fetch = missing_auth - set(auth_events)
seen_events = yield self.store.have_events(
set(auth_events.keys()) | set(state_events.keys())
)
all_events = events + state_events.values() + auth_events.values()
required_auth = set(
a_id for event in all_events for a_id, _ in event.auth_events
)
missing_auth = required_auth - set(auth_events)
results = yield defer.gatherResults(
[
self.replication_layer.get_pdu(
[dest],
event_id,
outlier=True,
timeout=10000,
)
for event_id in missing_auth
],
consumeErrors=True
).addErrback(unwrapFirstError)
auth_events.update({a.event_id: a for a in results})
ev_infos = []
for a in auth_events.values():
if a.event_id in seen_events:
continue
a.internal_metadata.outlier = True
ev_infos.append({
"event": a,
"auth_events": {
(auth_events[a_id].type, auth_events[a_id].state_key):
auth_events[a_id]
for a_id, _ in a.auth_events
if a_id in auth_events
}
})
@@ -447,27 +370,23 @@ class FederationHandler(BaseHandler):
(auth_events[a_id].type, auth_events[a_id].state_key):
auth_events[a_id]
for a_id, _ in event_map[e_id].auth_events
if a_id in auth_events
}
})
yield self._handle_new_events(
dest, ev_infos,
backfilled=True,
)
events.sort(key=lambda e: e.depth)
for event in events:
if event in events_to_state:
continue
# We store these one at a time since each event depends on the
# previous to work out the state.
# TODO: We can probably do something more clever here.
yield self._handle_new_event(
dest, event, backfilled=True,
)
ev_infos.append({
"event": event,
})
yield self._handle_new_events(
dest, ev_infos,
backfilled=True,
)
defer.returnValue(events)
@@ -491,10 +410,6 @@ class FederationHandler(BaseHandler):
)
max_depth = sorted_extremeties_tuple[0][1]
# We don't want to specify too many extremities as it causes the backfill
# request URI to be too long.
extremities = dict(sorted_extremeties_tuple[:5])
if current_depth > max_depth:
logger.debug(
"Not backfilling as we don't need to. %d < %d",
@@ -520,7 +435,7 @@ class FederationHandler(BaseHandler):
joined_domains = {}
for u, d in joined_users:
try:
dom = get_domain_from_id(u)
dom = UserID.from_string(u).domain
old_d = joined_domains.get(dom)
if old_d:
joined_domains[dom] = min(d, old_d)
@@ -543,15 +458,11 @@ class FederationHandler(BaseHandler):
# TODO: Should we try multiple of these at a time?
for dom in domains:
try:
yield self.backfill(
events = yield self.backfill(
dom, room_id,
limit=100,
extremities=[e for e in extremities.keys()]
)
# If this succeeded then we probably already have the
# appropriate stuff.
# TODO: We can probably do something more intelligent here.
defer.returnValue(True)
except SynapseError as e:
logger.info(
"Failed to backfill from %s because %s",
@@ -577,6 +488,8 @@ class FederationHandler(BaseHandler):
)
continue
if events:
defer.returnValue(True)
defer.returnValue(False)
success = yield try_backfill(likely_domains)
@@ -591,24 +504,12 @@ class FederationHandler(BaseHandler):
event_ids = list(extremities.keys())
states = yield preserve_context_over_deferred(defer.gatherResults([
preserve_fn(self.state_handler.resolve_state_groups)(room_id, [e])
states = yield defer.gatherResults([
self.state_handler.resolve_state_groups(room_id, [e])
for e in event_ids
]))
])
states = dict(zip(event_ids, [s[1] for s in states]))
state_map = yield self.store.get_events(
[e_id for ids in states.values() for e_id in ids],
get_prev_content=False
)
states = {
key: {
k: state_map[e_id]
for k, e_id in state_dict.items()
if e_id in state_map
} for key, state_dict in states.items()
}
for e_id, _ in sorted_extremeties_tuple:
likely_domains = get_domains_from_state(states[e_id])
@@ -718,7 +619,7 @@ class FederationHandler(BaseHandler):
pass
event_stream_id, max_stream_id = yield self._persist_auth_tree(
origin, auth_chain, state, event
auth_chain, state, event
)
with PreserveLoggingContext():
@@ -760,18 +661,11 @@ class FederationHandler(BaseHandler):
"state_key": user_id,
})
try:
message_handler = self.hs.get_handlers().message_handler
event, context = yield message_handler._create_new_client_event(
builder=builder,
)
except AuthError as e:
logger.warn("Failed to create join %r because %s", event, e)
raise e
event, context = yield self._create_new_client_event(
builder=builder,
)
# The remote hasn't signed it yet, obviously. We'll do the full checks
# when we get the event back in `on_send_join_request`
yield self.auth.check_from_context(event, context, do_sig_check=False)
self.auth.check(event, auth_events=context.current_state)
defer.returnValue(event)
@@ -817,15 +711,39 @@ class FederationHandler(BaseHandler):
user = UserID.from_string(event.state_key)
yield user_joined_room(self.distributor, user, event.room_id)
state_ids = context.prev_state_ids.values()
new_pdu = event
destinations = set()
for k, s in context.current_state.items():
try:
if k[0] == EventTypes.Member:
if s.content["membership"] == Membership.JOIN:
destinations.add(
UserID.from_string(s.state_key).domain
)
except:
logger.warn(
"Failed to get destination from event %s", s.event_id
)
destinations.discard(origin)
logger.debug(
"on_send_join_request: Sending event: %s, signatures: %s",
event.event_id,
event.signatures,
)
self.replication_layer.send_pdu(new_pdu, destinations)
state_ids = [e.event_id for e in context.current_state.values()]
auth_chain = yield self.store.get_auth_chain(set(
[event.event_id] + state_ids
))
state = yield self.store.get_events(context.prev_state_ids.values())
defer.returnValue({
"state": state.values(),
"state": context.current_state.values(),
"auth_chain": auth_chain,
})
@@ -973,18 +891,11 @@ class FederationHandler(BaseHandler):
"state_key": user_id,
})
message_handler = self.hs.get_handlers().message_handler
event, context = yield message_handler._create_new_client_event(
event, context = yield self._create_new_client_event(
builder=builder,
)
try:
# The remote hasn't signed it yet, obviously. We'll do the full checks
# when we get the event back in `on_send_leave_request`
yield self.auth.check_from_context(event, context, do_sig_check=False)
except AuthError as e:
logger.warn("Failed to create new leave %r because %s", event, e)
raise e
self.auth.check(event, auth_events=context.current_state)
defer.returnValue(event)
@@ -1023,14 +934,43 @@ class FederationHandler(BaseHandler):
event, event_stream_id, max_stream_id, extra_users=extra_users
)
new_pdu = event
destinations = set()
for k, s in context.current_state.items():
try:
if k[0] == EventTypes.Member:
if s.content["membership"] == Membership.LEAVE:
destinations.add(
UserID.from_string(s.state_key).domain
)
except:
logger.warn(
"Failed to get destination from event %s", s.event_id
)
destinations.discard(origin)
logger.debug(
"on_send_leave_request: Sending event: %s, signatures: %s",
event.event_id,
event.signatures,
)
self.replication_layer.send_pdu(new_pdu, destinations)
defer.returnValue(None)
@defer.inlineCallbacks
def get_state_for_pdu(self, room_id, event_id):
"""Returns the state at the event. i.e. not including said event.
"""
def get_state_for_pdu(self, origin, room_id, event_id, do_auth=True):
yield run_on_reactor()
if do_auth:
in_room = yield self.auth.check_host_in_room(room_id, origin)
if not in_room:
raise AuthError(403, "Host not in room.")
state_groups = yield self.store.get_state_groups(
room_id, [event_id]
)
@@ -1054,49 +994,18 @@ class FederationHandler(BaseHandler):
res = results.values()
for event in res:
# We sign these again because there was a bug where we
# incorrectly signed things the first time round
if self.hs.is_mine_id(event.event_id):
event.signatures.update(
compute_event_signature(
event,
self.hs.hostname,
self.hs.config.signing_key[0]
)
event.signatures.update(
compute_event_signature(
event,
self.hs.hostname,
self.hs.config.signing_key[0]
)
)
defer.returnValue(res)
else:
defer.returnValue([])
@defer.inlineCallbacks
def get_state_ids_for_pdu(self, room_id, event_id):
"""Returns the state at the event. i.e. not including said event.
"""
yield run_on_reactor()
state_groups = yield self.store.get_state_groups_ids(
room_id, [event_id]
)
if state_groups:
_, state = state_groups.items().pop()
results = state
event = yield self.store.get_event(event_id)
if event and event.is_state():
# Get previous state
if "replaces_state" in event.unsigned:
prev_id = event.unsigned["replaces_state"]
if prev_id != event.event_id:
results[(event.type, event.state_key)] = prev_id
else:
del results[(event.type, event.state_key)]
defer.returnValue(results.values())
else:
defer.returnValue([])
@defer.inlineCallbacks
@log_function
def on_backfill_request(self, origin, room_id, pdu_list, limit):
@@ -1129,17 +1038,16 @@ class FederationHandler(BaseHandler):
)
if event:
if self.hs.is_mine_id(event.event_id):
# FIXME: This is a temporary work around where we occasionally
# return events slightly differently than when they were
# originally signed
event.signatures.update(
compute_event_signature(
event,
self.hs.hostname,
self.hs.config.signing_key[0]
)
# FIXME: This is a temporary work around where we occasionally
# return events slightly differently than when they were
# originally signed
event.signatures.update(
compute_event_signature(
event,
self.hs.hostname,
self.hs.config.signing_key[0]
)
)
if do_auth:
in_room = yield self.auth.check_host_in_room(
@@ -1149,12 +1057,6 @@ class FederationHandler(BaseHandler):
if not in_room:
raise AuthError(403, "Host not in room.")
events = yield self._filter_events_for_server(
origin, event.room_id, [event]
)
event = events[0]
defer.returnValue(event)
else:
defer.returnValue(None)
@@ -1163,10 +1065,18 @@ class FederationHandler(BaseHandler):
def get_min_depth_for_context(self, context):
return self.store.get_min_depth(context)
@log_function
def user_joined_room(self, user, room_id):
waiters = self.waiting_for_join_list.get(
(user.to_string(), room_id),
[]
)
while waiters:
waiters.pop().callback(None)
@defer.inlineCallbacks
@log_function
def _handle_new_event(self, origin, event, state=None, auth_events=None,
backfilled=False):
def _handle_new_event(self, origin, event, state=None, auth_events=None):
context = yield self._prep_event(
origin, event,
state=state,
@@ -1176,34 +1086,21 @@ class FederationHandler(BaseHandler):
if not event.internal_metadata.is_outlier():
action_generator = ActionGenerator(self.hs)
yield action_generator.handle_push_actions_for_event(
event, context
event, context, self
)
event_stream_id, max_stream_id = yield self.store.persist_event(
event,
context=context,
backfilled=backfilled,
)
if not backfilled:
# this intentionally does not yield: we don't care about the result
# and don't need to wait for it.
preserve_fn(self.hs.get_pusherpool().on_new_notifications)(
event_stream_id, max_stream_id
)
defer.returnValue((context, event_stream_id, max_stream_id))
@defer.inlineCallbacks
def _handle_new_events(self, origin, event_infos, backfilled=False):
"""Creates the appropriate contexts and persists events. The events
should not depend on one another, e.g. this should be used to persist
a bunch of outliers, but not a chunk of individual events that depend
on each other for state calculations.
"""
contexts = yield preserve_context_over_deferred(defer.gatherResults(
contexts = yield defer.gatherResults(
[
preserve_fn(self._prep_event)(
self._prep_event(
origin,
ev_info["event"],
state=ev_info.get("state"),
@@ -1211,7 +1108,7 @@ class FederationHandler(BaseHandler):
)
for ev_info in event_infos
]
))
)
yield self.store.persist_events(
[
@@ -1222,19 +1119,11 @@ class FederationHandler(BaseHandler):
)
@defer.inlineCallbacks
def _persist_auth_tree(self, origin, auth_events, state, event):
def _persist_auth_tree(self, auth_events, state, event):
"""Checks the auth chain is valid (and passes auth checks) for the
state and event. Then persists the auth chain and state atomically.
Persists the event seperately.
Will attempt to fetch missing auth events.
Args:
origin (str): Where the events came from
auth_events (list)
state (list)
event (Event)
Returns:
2-tuple of (event_stream_id, max_stream_id) from the persist_event
call for `event`
@@ -1247,7 +1136,7 @@ class FederationHandler(BaseHandler):
event_map = {
e.event_id: e
for e in itertools.chain(auth_events, state, [event])
for e in auth_events
}
create_event = None
@@ -1256,29 +1145,10 @@ class FederationHandler(BaseHandler):
create_event = e
break
missing_auth_events = set()
for e in itertools.chain(auth_events, state, [event]):
for e_id, _ in e.auth_events:
if e_id not in event_map:
missing_auth_events.add(e_id)
for e_id in missing_auth_events:
m_ev = yield self.replication_layer.get_pdu(
[origin],
e_id,
outlier=True,
timeout=10000,
)
if m_ev and m_ev.event_id == e_id:
event_map[e_id] = m_ev
else:
logger.info("Failed to find auth event %r", e_id)
for e in itertools.chain(auth_events, state, [event]):
auth_for_e = {
(event_map[e_id].type, event_map[e_id].state_key): event_map[e_id]
for e_id, _ in e.auth_events
if e_id in event_map
}
if create_event:
auth_for_e[(EventTypes.Create, "")] = create_event
@@ -1327,13 +1197,7 @@ class FederationHandler(BaseHandler):
)
if not auth_events:
auth_events_ids = yield self.auth.compute_auth_events(
event, context.prev_state_ids, for_verification=True,
)
auth_events = yield self.store.get_events(auth_events_ids)
auth_events = {
(e.type, e.state_key): e for e in auth_events.values()
}
auth_events = context.current_state
# This is a hack to fix some old rooms where the initial join event
# didn't reference the create event in its auth events.
@@ -1359,7 +1223,8 @@ class FederationHandler(BaseHandler):
context.rejected = RejectedReason.AUTH_ERROR
if event.type == EventTypes.GuestAccess:
yield self.maybe_kick_guest_users(event)
full_context = yield self.store.get_current_state(room_id=event.room_id)
yield self.maybe_kick_guest_users(event, full_context)
defer.returnValue(context)
@@ -1427,11 +1292,6 @@ class FederationHandler(BaseHandler):
current_state = set(e.event_id for e in auth_events.values())
event_auth_events = set(e_id for e_id, _ in event.auth_events)
if event.is_state():
event_key = (event.type, event.state_key)
else:
event_key = None
if event_auth_events - current_state:
have_events = yield self.store.have_events(
event_auth_events - current_state
@@ -1505,9 +1365,9 @@ class FederationHandler(BaseHandler):
# Do auth conflict res.
logger.info("Different auth: %s", different_auth)
different_events = yield preserve_context_over_deferred(defer.gatherResults(
different_events = yield defer.gatherResults(
[
preserve_fn(self.store.get_event)(
self.store.get_event(
d,
allow_none=True,
allow_rejected=False,
@@ -1516,13 +1376,13 @@ class FederationHandler(BaseHandler):
if d in have_events and not have_events[d]
],
consumeErrors=True
)).addErrback(unwrapFirstError)
).addErrback(unwrapFirstError)
if different_events:
local_view = dict(auth_events)
remote_view = dict(auth_events)
remote_view.update({
(d.type, d.state_key): d for d in different_events if d
(d.type, d.state_key): d for d in different_events
})
new_state, prev_state = self.state_handler.resolve_events(
@@ -1535,16 +1395,8 @@ class FederationHandler(BaseHandler):
current_state = set(e.event_id for e in auth_events.values())
different_auth = event_auth_events - current_state
context.current_state_ids = dict(context.current_state_ids)
context.current_state_ids.update({
k: a.event_id for k, a in auth_events.items()
if k != event_key
})
context.prev_state_ids = dict(context.prev_state_ids)
context.prev_state_ids.update({
k: a.event_id for k, a in auth_events.items()
})
context.state_group = self.store.get_next_state_group()
context.current_state.update(auth_events)
context.state_group = None
if different_auth and not event.internal_metadata.is_outlier():
logger.info("Different auth after resolution: %s", different_auth)
@@ -1565,8 +1417,8 @@ class FederationHandler(BaseHandler):
if do_resolution:
# 1. Get what we think is the auth chain.
auth_ids = yield self.auth.compute_auth_events(
event, context.prev_state_ids
auth_ids = self.auth.compute_auth_events(
event, context.current_state
)
local_auth_chain = yield self.store.get_auth_chain(auth_ids)
@@ -1622,22 +1474,13 @@ class FederationHandler(BaseHandler):
# 4. Look at rejects and their proofs.
# TODO.
context.current_state_ids = dict(context.current_state_ids)
context.current_state_ids.update({
k: a.event_id for k, a in auth_events.items()
if k != event_key
})
context.prev_state_ids = dict(context.prev_state_ids)
context.prev_state_ids.update({
k: a.event_id for k, a in auth_events.items()
})
context.state_group = self.store.get_next_state_group()
context.current_state.update(auth_events)
context.state_group = None
try:
self.auth.check(event, auth_events=auth_events)
except AuthError as e:
logger.warn("Failed auth resolution for %r because %s", event, e)
raise e
except AuthError:
raise
@defer.inlineCallbacks
def construct_auth_difference(self, local_auth, remote_auth):
@@ -1807,22 +1650,14 @@ class FederationHandler(BaseHandler):
if (yield self.auth.check_host_in_room(room_id, self.hs.hostname)):
builder = self.event_builder_factory.new(event_dict)
EventValidator().validate_new(builder)
message_handler = self.hs.get_handlers().message_handler
event, context = yield message_handler._create_new_client_event(
builder=builder
)
event, context = yield self._create_new_client_event(builder=builder)
event, context = yield self.add_display_name_to_third_party_invite(
event_dict, event, context
)
try:
yield self.auth.check_from_context(event, context)
except AuthError as e:
logger.warn("Denying new third party invite %r because %s", event, e)
raise e
yield self._check_signature(event, context)
self.auth.check(event, context.current_state)
yield self._check_signature(event, auth_events=context.current_state)
member_handler = self.hs.get_handlers().room_member_handler
yield member_handler.send_membership_event(None, event, context)
else:
@@ -1838,8 +1673,7 @@ class FederationHandler(BaseHandler):
def on_exchange_third_party_invite_request(self, origin, room_id, event_dict):
builder = self.event_builder_factory.new(event_dict)
message_handler = self.hs.get_handlers().message_handler
event, context = yield message_handler._create_new_client_event(
event, context = yield self._create_new_client_event(
builder=builder,
)
@@ -1847,12 +1681,8 @@ class FederationHandler(BaseHandler):
event_dict, event, context
)
try:
self.auth.check_from_context(event, context)
except AuthError as e:
logger.warn("Denying third party invite %r because %s", event, e)
raise e
yield self._check_signature(event, context)
self.auth.check(event, auth_events=context.current_state)
yield self._check_signature(event, auth_events=context.current_state)
returned_invite = yield self.send_invite(origin, event)
# TODO: Make sure the signatures actually are correct.
@@ -1866,38 +1696,29 @@ class FederationHandler(BaseHandler):
EventTypes.ThirdPartyInvite,
event.content["third_party_invite"]["signed"]["token"]
)
original_invite = None
original_invite_id = context.prev_state_ids.get(key)
if original_invite_id:
original_invite = yield self.store.get_event(
original_invite_id, allow_none=True
)
if original_invite:
display_name = original_invite.content["display_name"]
event_dict["content"]["third_party_invite"]["display_name"] = display_name
else:
original_invite = context.current_state.get(key)
if not original_invite:
logger.info(
"Could not find invite event for third_party_invite: %r",
event_dict
"Could not find invite event for third_party_invite - "
"discarding: %s" % (event_dict,)
)
# We don't discard here as this is not the appropriate place to do
# auth checks. If we need the invite and don't have it then the
# auth check code will explode appropriately.
return
display_name = original_invite.content["display_name"]
event_dict["content"]["third_party_invite"]["display_name"] = display_name
builder = self.event_builder_factory.new(event_dict)
EventValidator().validate_new(builder)
message_handler = self.hs.get_handlers().message_handler
event, context = yield message_handler._create_new_client_event(builder=builder)
event, context = yield self._create_new_client_event(builder=builder)
defer.returnValue((event, context))
@defer.inlineCallbacks
def _check_signature(self, event, context):
def _check_signature(self, event, auth_events):
"""
Checks that the signature in the event is consistent with its invite.
Args:
event (Event): The m.room.member event to check
context (EventContext):
auth_events (dict<(event type, state_key), event>):
Raises:
AuthError: if signature didn't match any keys, or key has been
@@ -1908,14 +1729,10 @@ class FederationHandler(BaseHandler):
signed = event.content["third_party_invite"]["signed"]
token = signed["token"]
invite_event_id = context.prev_state_ids.get(
invite_event = auth_events.get(
(EventTypes.ThirdPartyInvite, token,)
)
invite_event = None
if invite_event_id:
invite_event = yield self.store.get_event(invite_event_id, allow_none=True)
if not invite_event:
raise AuthError(403, "Could not find invite")

View File

@@ -21,7 +21,7 @@ from synapse.api.errors import (
)
from ._base import BaseHandler
from synapse.util.async import run_on_reactor
from synapse.api.errors import SynapseError, Codes
from synapse.api.errors import SynapseError
import json
import logging
@@ -41,20 +41,6 @@ class IdentityHandler(BaseHandler):
hs.config.use_insecure_ssl_client_just_for_testing_do_not_use
)
def _should_trust_id_server(self, id_server):
if id_server not in self.trusted_id_servers:
if self.trust_any_id_server_just_for_testing_do_not_use:
logger.warn(
"Trusting untrustworthy ID server %r even though it isn't"
" in the trusted id list for testing because"
" 'use_insecure_ssl_client_just_for_testing_do_not_use'"
" is set in the config",
id_server,
)
else:
return False
return True
@defer.inlineCallbacks
def threepid_from_creds(self, creds):
yield run_on_reactor()
@@ -73,12 +59,19 @@ class IdentityHandler(BaseHandler):
else:
raise SynapseError(400, "No client_secret in creds")
if not self._should_trust_id_server(id_server):
logger.warn(
'%s is not a trusted ID server: rejecting 3pid ' +
'credentials', id_server
)
defer.returnValue(None)
if id_server not in self.trusted_id_servers:
if self.trust_any_id_server_just_for_testing_do_not_use:
logger.warn(
"Trusting untrustworthy ID server %r even though it isn't"
" in the trusted id list for testing because"
" 'use_insecure_ssl_client_just_for_testing_do_not_use'"
" is set in the config",
id_server,
)
else:
logger.warn('%s is not a trusted ID server: rejecting 3pid ' +
'credentials', id_server)
defer.returnValue(None)
data = {}
try:
@@ -136,12 +129,6 @@ class IdentityHandler(BaseHandler):
def requestEmailToken(self, id_server, email, client_secret, send_attempt, **kwargs):
yield run_on_reactor()
if not self._should_trust_id_server(id_server):
raise SynapseError(
400, "Untrusted ID server '%s'" % id_server,
Codes.SERVER_NOT_TRUSTED
)
params = {
'email': email,
'client_secret': client_secret,

Some files were not shown because too many files have changed in this diff Show More