Fixup changelog

Fixes #6552

.buildkite/worker-blacklist

@@ -34,33 +34,8 @@ Device list doesn't change if remote server is down
 Remote servers cannot set power levels in rooms without existing powerlevels
 Remote servers should reject attempts by non-creators to set the power levels
 
-# new failures as of https://github.com/matrix-org/sytest/pull/753
-GET /rooms/:room_id/messages returns a message
-GET /rooms/:room_id/messages lazy loads members correctly
-Read receipts are sent as events
-Only original members of the room can see messages from erased users
-Device deletion propagates over federation
-If user leaves room, remote user changes device and rejoins we see update in /sync and /keys/changes
-Changing user-signing key notifies local users
-Newly updated tags appear in an incremental v2 /sync
+# https://buildkite.com/matrix-dot-org/synapse/builds/6134#6f67bf47-e234-474d-80e8-c6e1868b15c5
 Server correctly handles incoming m.device_list_update
-Local device key changes get to remote servers with correct prev_id
-AS-ghosted users can use rooms via AS
-Ghost user must register before joining room
-Test that a message is pushed
-Invites are pushed
-Rooms with aliases are correctly named in pushed
-Rooms with names are correctly named in pushed
-Rooms with canonical alias are correctly named in pushed
-Rooms with many users are correctly pushed
-Don't get pushed for rooms you've muted
-Rejected events are not pushed
-Test that rejected pushers are removed.
-Events come down the correct room
 
-# https://buildkite.com/matrix-dot-org/sytest/builds/326#cca62404-a88a-4fcb-ad41-175fd3377603
-Presence changes to UNAVAILABLE are reported to remote room members
-If remote user leaves room, changes device and rejoins we see update in sync
-uploading self-signing key notifies over federation
-Inbound federation can receive redacted events
-Outbound federation can request missing events
+# this fails reliably with a torture level of 100 due to https://github.com/matrix-org/synapse/issues/6536
+Outbound federation requests missing prev_events and then asks for /state_ids and resolves the state

AUTHORS.rst

@@ -46,3 +46,6 @@ Joseph Weston <joseph at weston.cloud>
 
 Benjamin Saunders <ben.e.saunders at gmail dot com>
  * Documentation improvements
+
+Werner Sembach <werner.sembach at fau dot de>
+ * Automatically remove a group/community when it is empty

CHANGES.md

@@ -1,3 +1,94 @@
+Synapse 1.8.0 (2020-01-09)
+==========================
+
+Bugfixes
+--------
+
+- Fix `GET` request on `/_synapse/admin/v2/users` endpoint. Contributed by Awesome Technologies Innovationslabor GmbH. ([\#6563](https://github.com/matrix-org/synapse/issues/6563))
+- Fix incorrect signing of responses from the key server implementation. ([\#6657](https://github.com/matrix-org/synapse/issues/6657))
+
+
+Synapse 1.8.0rc1 (2020-01-07)
+=============================
+
+Features
+--------
+
+- Add v2 APIs for the `send_join` and `send_leave` federation endpoints (as described in [MSC1802](https://github.com/matrix-org/matrix-doc/pull/1802)). ([\#6349](https://github.com/matrix-org/synapse/issues/6349))
+- Add a develop script to generate full SQL schemas. ([\#6394](https://github.com/matrix-org/synapse/issues/6394))
+- Add custom SAML username mapping functinality through an external provider plugin. ([\#6411](https://github.com/matrix-org/synapse/issues/6411))
+- Automatically delete empty groups/communities. ([\#6453](https://github.com/matrix-org/synapse/issues/6453))
+- Add option `limit_profile_requests_to_users_who_share_rooms` to prevent requirement of a local user sharing a room with another user to query their profile information. ([\#6523](https://github.com/matrix-org/synapse/issues/6523))
+- Add an `export_signing_key` script to extract the public part of signing keys when rotating them. ([\#6546](https://github.com/matrix-org/synapse/issues/6546))
+- Add experimental config option to specify multiple databases. ([\#6580](https://github.com/matrix-org/synapse/issues/6580))
+- Raise an error if someone tries to use the `log_file` config option. ([\#6626](https://github.com/matrix-org/synapse/issues/6626))
+
+
+Bugfixes
+--------
+
+- Prevent redacted events from being returned during message search. ([\#6377](https://github.com/matrix-org/synapse/issues/6377), [\#6522](https://github.com/matrix-org/synapse/issues/6522))
+- Prevent error on trying to search a upgraded room when the server is not in the predecessor room. ([\#6385](https://github.com/matrix-org/synapse/issues/6385))
+- Improve performance of looking up cross-signing keys. ([\#6486](https://github.com/matrix-org/synapse/issues/6486))
+- Fix race which occasionally caused deleted devices to reappear. ([\#6514](https://github.com/matrix-org/synapse/issues/6514))
+- Fix missing row in `device_max_stream_id` that could cause unable to decrypt errors after server restart. ([\#6555](https://github.com/matrix-org/synapse/issues/6555))
+- Fix a bug which meant that we did not send systemd notifications on startup if acme was enabled. ([\#6571](https://github.com/matrix-org/synapse/issues/6571))
+- Fix exception when fetching the `matrix.org:ed25519:auto` key. ([\#6625](https://github.com/matrix-org/synapse/issues/6625))
+- Fix bug where a moderator upgraded a room and became an admin in the new room. ([\#6633](https://github.com/matrix-org/synapse/issues/6633))
+- Fix an error which was thrown by the `PresenceHandler` `_on_shutdown` handler. ([\#6640](https://github.com/matrix-org/synapse/issues/6640))
+- Fix exceptions in the synchrotron worker log when events are rejected. ([\#6645](https://github.com/matrix-org/synapse/issues/6645))
+- Ensure that upgraded rooms are removed from the directory. ([\#6648](https://github.com/matrix-org/synapse/issues/6648))
+- Fix a bug causing Synapse not to fetch missing events when it believes it has every event in the room. ([\#6652](https://github.com/matrix-org/synapse/issues/6652))
+
+
+Improved Documentation
+----------------------
+
+- Document the Room Shutdown Admin API. ([\#6541](https://github.com/matrix-org/synapse/issues/6541))
+- Reword sections of [docs/federate.md](docs/federate.md) that explained delegation at time of Synapse 1.0 transition. ([\#6601](https://github.com/matrix-org/synapse/issues/6601))
+- Added the section 'Configuration' in [docs/turn-howto.md](docs/turn-howto.md). ([\#6614](https://github.com/matrix-org/synapse/issues/6614))
+
+
+Deprecations and Removals
+-------------------------
+
+- Remove redundant code from event authorisation implementation. ([\#6502](https://github.com/matrix-org/synapse/issues/6502))
+- Remove unused, undocumented `/_matrix/content` API. ([\#6628](https://github.com/matrix-org/synapse/issues/6628))
+
+
+Internal Changes
+----------------
+
+- Add *experimental* support for multiple physical databases and split out state storage to separate data store. ([\#6245](https://github.com/matrix-org/synapse/issues/6245), [\#6510](https://github.com/matrix-org/synapse/issues/6510), [\#6511](https://github.com/matrix-org/synapse/issues/6511), [\#6513](https://github.com/matrix-org/synapse/issues/6513), [\#6564](https://github.com/matrix-org/synapse/issues/6564), [\#6565](https://github.com/matrix-org/synapse/issues/6565))
+- Port sections of code base to async/await. ([\#6496](https://github.com/matrix-org/synapse/issues/6496), [\#6504](https://github.com/matrix-org/synapse/issues/6504), [\#6505](https://github.com/matrix-org/synapse/issues/6505), [\#6517](https://github.com/matrix-org/synapse/issues/6517), [\#6559](https://github.com/matrix-org/synapse/issues/6559), [\#6647](https://github.com/matrix-org/synapse/issues/6647), [\#6653](https://github.com/matrix-org/synapse/issues/6653))
+- Remove `SnapshotCache` in favour of `ResponseCache`. ([\#6506](https://github.com/matrix-org/synapse/issues/6506))
+- Silence mypy errors for files outside those specified. ([\#6512](https://github.com/matrix-org/synapse/issues/6512))
+- Clean up some logging when handling incoming events over federation. ([\#6515](https://github.com/matrix-org/synapse/issues/6515))
+- Test more folders against mypy. ([\#6534](https://github.com/matrix-org/synapse/issues/6534))
+- Update `mypy` to new version. ([\#6537](https://github.com/matrix-org/synapse/issues/6537))
+- Adjust the sytest blacklist for worker mode. ([\#6538](https://github.com/matrix-org/synapse/issues/6538))
+- Remove unused `get_pagination_rows` methods from `EventSource` classes. ([\#6557](https://github.com/matrix-org/synapse/issues/6557))
+- Clean up logs from the push notifier at startup. ([\#6558](https://github.com/matrix-org/synapse/issues/6558))
+- Improve diagnostics on database upgrade failure. ([\#6570](https://github.com/matrix-org/synapse/issues/6570))
+- Reduce the reconnect time when worker replication fails, to make it easier to catch up. ([\#6617](https://github.com/matrix-org/synapse/issues/6617))
+- Simplify http handling by removing redundant `SynapseRequestFactory`. ([\#6619](https://github.com/matrix-org/synapse/issues/6619))
+- Add a workaround for synapse raising exceptions when fetching the notary's own key from the notary. ([\#6620](https://github.com/matrix-org/synapse/issues/6620))
+- Automate generation of the sample log config. ([\#6627](https://github.com/matrix-org/synapse/issues/6627))
+- Simplify event creation code by removing redundant queries on the `event_reference_hashes` table. ([\#6629](https://github.com/matrix-org/synapse/issues/6629))
+- Fix errors when `frozen_dicts` are enabled. ([\#6642](https://github.com/matrix-org/synapse/issues/6642))
+
+
+Synapse 1.7.3 (2019-12-31)
+==========================
+
+This release fixes a long-standing bug in the state resolution algorithm.
+
+Bugfixes
+--------
+
+- Fix exceptions caused by state resolution choking on malformed events. ([\#6608](https://github.com/matrix-org/synapse/issues/6608))
+
+
 Synapse 1.7.2 (2019-12-20)
 ==========================
 

README.rst

@@ -393,4 +393,4 @@ something like the following in their logs::
     2019-09-11 19:32:04,271 - synapse.federation.transport.server - 288 - WARNING - GET-11752 - authenticate_request failed: 401: Invalid signature for server <server> with key ed25519:a_EqML: Unable to verify signature for <server>
 
 This is normally caused by a misconfiguration in your reverse-proxy. See
-`<docs/reverse_proxy.rst>`_ and double-check that your settings are correct.
+`<docs/reverse_proxy.md>`_ and double-check that your settings are correct.

debian/build_virtualenv

@@ -85,6 +85,9 @@ PYTHONPATH="$tmpdir" \
 
 ' > "${PACKAGE_BUILD_DIR}/etc/matrix-synapse/homeserver.yaml"
 
+# build the log config file
+"${TARGET_PYTHON}" -B "${VIRTUALENV_DIR}/bin/generate_log_config" \
+        --output-file="${PACKAGE_BUILD_DIR}/etc/matrix-synapse/log.yaml"
 
 # add a dependency on the right version of python to substvars.
 PYPKG=`basename $SNAKE`

debian/changelog

@@ -1,3 +1,19 @@
+matrix-synapse-py3 (1.8.0) stable; urgency=medium
+
+  [ Richard van der Hoff ]
+  * Automate generation of the default log configuration file.
+
+  [ Synapse Packaging team ]
+  * New synapse release 1.8.0.
+
+ -- Synapse Packaging team <packages@matrix.org>  Thu, 09 Jan 2020 11:39:27 +0000
+
+matrix-synapse-py3 (1.7.3) stable; urgency=medium
+
+  * New synapse release 1.7.3.
+
+ -- Synapse Packaging team <packages@matrix.org>  Tue, 31 Dec 2019 10:45:04 +0000
+
 matrix-synapse-py3 (1.7.2) stable; urgency=medium
 
   * New synapse release 1.7.2.

debian/install

@@ -1,2 +1 @@
-debian/log.yaml etc/matrix-synapse
 debian/manage_debconf.pl /opt/venvs/matrix-synapse/lib/

debian/log.yaml

@@ -1,36 +0,0 @@
-
-version: 1
-
-formatters:
-  precise:
-   format: '%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(request)s- %(message)s'
-
-filters:
-  context:
-    (): synapse.logging.context.LoggingContextFilter
-    request: ""
-
-handlers:
-  file:
-    class: logging.handlers.RotatingFileHandler
-    formatter: precise
-    filename: /var/log/matrix-synapse/homeserver.log
-    maxBytes: 104857600
-    backupCount: 10
-    filters: [context]
-    encoding: utf8
-  console:
-    class: logging.StreamHandler
-    formatter: precise
-    level: WARN
-
-loggers:
-    synapse:
-        level: INFO
-
-    synapse.storage.SQL:
-        level: INFO
-
-root:
-    level: INFO
-    handlers: [file, console]

docs/admin_api/shutdown_room.md

@@ -0,0 +1,72 @@
+# Shutdown room API
+
+Shuts down a room, preventing new joins and moves local users and room aliases automatically
+to a new room. The new room will be created with the user specified by the
+`new_room_user_id` parameter as room administrator and will contain a message
+explaining what happened. Users invited to the new room will have power level
+-10 by default, and thus be unable to speak. The old room's power levels will be changed to
+disallow any further invites or joins.
+
+The local server will only have the power to move local user and room aliases to
+the new room. Users on other servers will be unaffected.
+
+## API
+
+You will need to authenticate with an access token for an admin user.
+
+### URL
+
+`POST /_synapse/admin/v1/shutdown_room/{room_id}`
+
+### URL Parameters
+
+* `room_id` - The ID of the room (e.g `!someroom:example.com`)
+
+### JSON Body Parameters
+
+* `new_room_user_id` - Required. A string representing the user ID of the user that will admin
+                       the new room that all users in the old room will be moved to.
+* `room_name` - Optional. A string representing the name of the room that new users will be
+                invited to.
+* `message` - Optional. A string containing the first message that will be sent as
+              `new_room_user_id` in the new room. Ideally this will clearly convey why the
+               original room was shut down.
+              
+If not specified, the default value of `room_name` is "Content Violation
+Notification". The default value of `message` is "Sharing illegal content on
+othis server is not permitted and rooms in violation will be blocked."
+
+### Response Parameters
+
+* `kicked_users` - An integer number representing the number of users that
+                   were kicked.
+* `failed_to_kick_users` - An integer number representing the number of users
+                           that were not kicked.
+* `local_aliases` - An array of strings representing the local aliases that were migrated from
+                    the old room to the new.
+* `new_room_id` - A string representing the room ID of the new room.
+
+## Example
+
+Request:
+
+```
+POST /_synapse/admin/v1/shutdown_room/!somebadroom%3Aexample.com
+
+{
+    "new_room_user_id": "@someuser:example.com",
+    "room_name": "Content Violation Notification",
+    "message": "Bad Room has been shutdown due to content violations on this server. Please review our Terms of Service."
+}
+```
+
+Response:
+
+```
+{
+    "kicked_users": 5,
+    "failed_to_kick_users": 0,
+    "local_aliases": ["#badroom:example.com", "#evilsaloon:example.com],
+    "new_room_id": "!newroomid:example.com",
+},
+```

docs/code_style.md

@@ -137,6 +137,7 @@ Some guidelines follow:
     correctly handles the top-level option being set to `None` (as it
     will be if no sub-options are enabled).
 -   Lines should be wrapped at 80 characters.
+-   Use two-space indents.
 
 Example:
 
@@ -155,13 +156,13 @@ Example:
     # Settings for the frobber
     #
     frobber:
-       # frobbing speed. Defaults to 1.
-       #
-       #speed: 10
+      # frobbing speed. Defaults to 1.
+      #
+      #speed: 10
 
-       # frobbing distance. Defaults to 1000.
-       #
-       #distance: 100
+      # frobbing distance. Defaults to 1000.
+      #
+      #distance: 100
 
 Note that the sample configuration is generated from the synapse code
 and is maintained by a script, `scripts-dev/generate_sample_config`.

docs/federate.md

@@ -66,10 +66,6 @@ therefore cannot gain access to the necessary certificate. With .well-known,
 federation servers will check for a valid TLS certificate for the delegated
 hostname (in our example: ``synapse.example.com``).
 
-.well-known support first appeared in Synapse v0.99.0. To federate with older
-servers you may need to additionally configure SRV delegation. Alternatively,
-encourage the server admin in question to upgrade :).
-
 ### DNS SRV delegation
 
 To use this delegation method, you need to have write access to your
@@ -111,29 +107,15 @@ giving it a `server_name` of `example.com`, and once [ACME](acme.md) support is
 it would automatically generate a valid TLS certificate for you via Let's Encrypt
 and no SRV record or .well-known URI would be needed.
 
-This is the common case, although you can add an SRV record or
-`.well-known/matrix/server` URI for completeness if you wish.
-
 **However**, if your server does not listen on port 8448, or if your `server_name`
 does not point to the host that your homeserver runs on, you will need to let
 other servers know how to find it. The way to do this is via .well-known or an
 SRV record.
 
-#### I have created a .well-known URI. Do I still need an SRV record?
+#### I have created a .well-known URI. Do I also need an SRV record?
 
-As of Synapse 0.99, Synapse will first check for the existence of a .well-known
-URI and follow any delegation it suggests. It will only then check for the
-existence of an SRV record.
-
-That means that the SRV record will often be redundant. However, you should
-remember that there may still be older versions of Synapse in the federation
-which do not understand .well-known URIs, so if you removed your SRV record
-you would no longer be able to federate with them.
-
-It is therefore best to leave the SRV record in place for now. Synapse 0.34 and
-earlier will follow the SRV record (and not care about the invalid
-certificate). Synapse 0.99 and later will follow the .well-known URI, with the
-correct certificate chain.
+No. You can use either `.well-known` delegation or use an SRV record for delegation. You
+do not need to use both to delegate to the same location.
 
 #### Can I manage my own certificates rather than having Synapse renew certificates itself?
 

docs/saml_mapping_providers.md

@@ -0,0 +1,77 @@
+# SAML Mapping Providers
+
+A SAML mapping provider is a Python class (loaded via a Python module) that
+works out how to map attributes of a SAML response object to Matrix-specific
+user attributes. Details such as user ID localpart, displayname, and even avatar
+URLs are all things that can be mapped from talking to a SSO service.
+
+As an example, a SSO service may return the email address
+"john.smith@example.com" for a user, whereas Synapse will need to figure out how
+to turn that into a displayname when creating a Matrix user for this individual.
+It may choose `John Smith`, or `Smith, John [Example.com]` or any number of
+variations. As each Synapse configuration may want something different, this is
+where SAML mapping providers come into play.
+
+## Enabling Providers
+
+External mapping providers are provided to Synapse in the form of an external
+Python module. Retrieve this module from [PyPi](https://pypi.org) or elsewhere,
+then tell Synapse where to look for the handler class by editing the
+`saml2_config.user_mapping_provider.module` config option.
+
+`saml2_config.user_mapping_provider.config` allows you to provide custom
+configuration options to the module. Check with the module's documentation for
+what options it provides (if any). The options listed by default are for the
+user mapping provider built in to Synapse. If using a custom module, you should
+comment these options out and use those specified by the module instead.
+
+## Building a Custom Mapping Provider
+
+A custom mapping provider must specify the following methods:
+
+* `__init__(self, parsed_config)`
+   - Arguments:
+     - `parsed_config` - A configuration object that is the return value of the
+       `parse_config` method. You should set any configuration options needed by
+       the module here.
+* `saml_response_to_user_attributes(self, saml_response, failures)`
+    - Arguments:
+      - `saml_response` - A `saml2.response.AuthnResponse` object to extract user
+                          information from.
+      - `failures` - An `int` that represents the amount of times the returned
+                     mxid localpart mapping has failed.  This should be used
+                     to create a deduplicated mxid localpart which should be
+                     returned instead. For example, if this method returns
+                     `john.doe` as the value of `mxid_localpart` in the returned
+                     dict, and that is already taken on the homeserver, this
+                     method will be called again with the same parameters but
+                     with failures=1. The method should then return a different
+                     `mxid_localpart` value, such as `john.doe1`.
+    - This method must return a dictionary, which will then be used by Synapse
+      to build a new user. The following keys are allowed:
+       * `mxid_localpart` - Required. The mxid localpart of the new user.
+       * `displayname` - The displayname of the new user. If not provided, will default to
+                         the value of `mxid_localpart`.
+* `parse_config(config)`
+    - This method should have the `@staticmethod` decoration.
+    - Arguments:
+        - `config` - A `dict` representing the parsed content of the
+          `saml2_config.user_mapping_provider.config` homeserver config option.
+           Runs on homeserver startup. Providers should extract any option values
+           they need here.
+    - Whatever is returned will be passed back to the user mapping provider module's
+      `__init__` method during construction.
+* `get_saml_attributes(config)`
+    - This method should have the `@staticmethod` decoration.
+    - Arguments:
+        - `config` - A object resulting from a call to `parse_config`.
+    - Returns a tuple of two sets. The first set equates to the saml auth
+      response attributes that are required for the module to function, whereas
+      the second set consists of those attributes which can be used if available,
+      but are not necessary.
+
+## Synapse's Default Provider
+
+Synapse has a built-in SAML mapping provider if a custom provider isn't
+specified in the config. It is located at
+[`synapse.handlers.saml_handler.DefaultSamlMappingProvider`](../synapse/handlers/saml_handler.py).

docs/sample_config.yaml

@@ -54,6 +54,13 @@ pid_file: DATADIR/homeserver.pid
 #
 #require_auth_for_profile_requests: true
 
+# Uncomment to require a user to share a room with another user in order
+# to retrieve their profile information. Only checked on Client-Server
+# requests. Profile requests from other servers should be checked by the
+# requesting server. Defaults to 'false'.
+#
+#limit_profile_requests_to_users_who_share_rooms: true
+
 # If set to 'true', removes the need for authentication to access the server's
 # public rooms directory through the client API, meaning that anyone can
 # query the room directory. Defaults to 'false'.
@@ -685,10 +692,6 @@ media_store_path: "DATADIR/media_store"
 #    config:
 #       directory: /mnt/some/other/directory
 
-# Directory where in-progress uploads are stored.
-#
-uploads_path: "DATADIR/uploads"
-
 # The largest allowed upload size in bytes
 #
 #max_upload_size: 10M
@@ -1115,14 +1118,19 @@ metrics_flags:
 signing_key_path: "CONFDIR/SERVERNAME.signing.key"
 
 # The keys that the server used to sign messages with but won't use
-# to sign new messages. E.g. it has lost its private key
+# to sign new messages.
 #
-#old_signing_keys:
-#  "ed25519:auto":
-#    # Base64 encoded public key
-#    key: "The public part of your old signing key."
-#    # Millisecond POSIX timestamp when the key expired.
-#    expired_ts: 123456789123
+old_signing_keys:
+  # For each key, `key` should be the base64-encoded public key, and
+  # `expired_ts`should be the time (in milliseconds since the unix epoch) that
+  # it was last used.
+  #
+  # It is possible to build an entry from an old signing.key file using the
+  # `export_signing_key` script which is provided with synapse.
+  #
+  # For example:
+  #
+  #"ed25519:id": { key: "base64string", expired_ts: 123456789123 }
 
 # How long key response published by this server is valid for.
 # Used to set the valid_until_ts in /key/v2 APIs.
@@ -1250,33 +1258,58 @@ saml2_config:
   #
   #config_path: "CONFDIR/sp_conf.py"
 
-  # the lifetime of a SAML session. This defines how long a user has to
+  # The lifetime of a SAML session. This defines how long a user has to
   # complete the authentication process, if allow_unsolicited is unset.
   # The default is 5 minutes.
   #
   #saml_session_lifetime: 5m
 
-  # The SAML attribute (after mapping via the attribute maps) to use to derive
-  # the Matrix ID from. 'uid' by default.
+  # An external module can be provided here as a custom solution to
+  # mapping attributes returned from a saml provider onto a matrix user.
   #
-  #mxid_source_attribute: displayName
+  user_mapping_provider:
+    # The custom module's class. Uncomment to use a custom module.
+    #
+    #module: mapping_provider.SamlMappingProvider
 
-  # The mapping system to use for mapping the saml attribute onto a matrix ID.
-  # Options include:
-  #  * 'hexencode' (which maps unpermitted characters to '=xx')
-  #  * 'dotreplace' (which replaces unpermitted characters with '.').
-  # The default is 'hexencode'.
-  #
-  #mxid_mapping: dotreplace
+    # Custom configuration values for the module. Below options are
+    # intended for the built-in provider, they should be changed if
+    # using a custom module. This section will be passed as a Python
+    # dictionary to the module's `parse_config` method.
+    #
+    config:
+      # The SAML attribute (after mapping via the attribute maps) to use
+      # to derive the Matrix ID from. 'uid' by default.
+      #
+      # Note: This used to be configured by the
+      # saml2_config.mxid_source_attribute option. If that is still
+      # defined, its value will be used instead.
+      #
+      #mxid_source_attribute: displayName
 
-  # In previous versions of synapse, the mapping from SAML attribute to MXID was
-  # always calculated dynamically rather than stored in a table. For backwards-
-  # compatibility, we will look for user_ids matching such a pattern before
-  # creating a new account.
+      # The mapping system to use for mapping the saml attribute onto a
+      # matrix ID.
+      #
+      # Options include:
+      #  * 'hexencode' (which maps unpermitted characters to '=xx')
+      #  * 'dotreplace' (which replaces unpermitted characters with
+      #     '.').
+      # The default is 'hexencode'.
+      #
+      # Note: This used to be configured by the
+      # saml2_config.mxid_mapping option. If that is still defined, its
+      # value will be used instead.
+      #
+      #mxid_mapping: dotreplace
+
+  # In previous versions of synapse, the mapping from SAML attribute to
+  # MXID was always calculated dynamically rather than stored in a
+  # table. For backwards- compatibility, we will look for user_ids
+  # matching such a pattern before creating a new account.
   #
   # This setting controls the SAML attribute which will be used for this
-  # backwards-compatibility lookup. Typically it should be 'uid', but if the
-  # attribute maps are changed, it may be necessary to change it.
+  # backwards-compatibility lookup. Typically it should be 'uid', but if
+  # the attribute maps are changed, it may be necessary to change it.
   #
   # The default is 'uid'.
   #

docs/sample_log_config.yaml

@@ -0,0 +1,43 @@
+# Log configuration for Synapse.
+#
+# This is a YAML file containing a standard Python logging configuration
+# dictionary. See [1] for details on the valid settings.
+#
+# [1]: https://docs.python.org/3.7/library/logging.config.html#configuration-dictionary-schema
+
+version: 1
+
+formatters:
+    precise:
+        format: '%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(request)s - %(message)s'
+
+filters:
+    context:
+        (): synapse.logging.context.LoggingContextFilter
+        request: ""
+
+handlers:
+    file:
+        class: logging.handlers.RotatingFileHandler
+        formatter: precise
+        filename: /var/log/matrix-synapse/homeserver.log
+        maxBytes: 104857600
+        backupCount: 10
+        filters: [context]
+        encoding: utf8
+    console:
+        class: logging.StreamHandler
+        formatter: precise
+        filters: [context]
+
+loggers:
+    synapse.storage.SQL:
+        # beware: increasing this to DEBUG will make synapse log sensitive
+        # information such as access tokens.
+        level: INFO
+
+root:
+    level: INFO
+    handlers: [file, console]
+
+disable_existing_loggers: false

docs/turn-howto.md

@@ -39,6 +39,8 @@ The TURN daemon `coturn` is available from a variety of sources such as native p
         make
         make install
 
+### Configuration
+
 1.  Create or edit the config file in `/etc/turnserver.conf`. The relevant
     lines, with example values, are:
 

docs/workers.md

@@ -196,7 +196,7 @@ Handles the media repository. It can handle all endpoints starting with:
 
     /_matrix/media/
 
-And the following regular expressions matching media-specific administration APIs:
+... and the following regular expressions matching media-specific administration APIs:
 
     ^/_synapse/admin/v1/purge_media_cache$
     ^/_synapse/admin/v1/room/.*/media$
@@ -206,6 +206,18 @@ You should also set `enable_media_repo: False` in the shared configuration
 file to stop the main synapse running background jobs related to managing the
 media repository.
 
+In the `media_repository` worker configuration file, configure the http listener to
+expose the `media` resource. For example:
+
+```yaml
+    worker_listeners:
+     - type: http
+       port: 8085
+       resources:
+         - names:
+           - media
+```
+
 Note this worker cannot be load-balanced: only one instance should be active.
 
 ### `synapse.app.client_reader`

mypy.ini

@@ -1,7 +1,7 @@
 [mypy]
 namespace_packages = True
 plugins = mypy_zope:plugin
-follow_imports = normal
+follow_imports = silent
 check_untyped_defs = True
 show_error_codes = True
 show_traceback = True

scripts-dev/generate_sample_config

@@ -7,12 +7,22 @@ set -e
 cd `dirname $0`/..
 
 SAMPLE_CONFIG="docs/sample_config.yaml"
+SAMPLE_LOG_CONFIG="docs/sample_log_config.yaml"
+
+check() {
+    diff -u "$SAMPLE_LOG_CONFIG" <(./scripts/generate_log_config) >/dev/null || return 1
+}
 
 if [ "$1" == "--check" ]; then
     diff -u "$SAMPLE_CONFIG" <(./scripts/generate_config --header-file docs/.sample_config_header.yaml) >/dev/null || {
         echo -e "\e[1m\e[31m$SAMPLE_CONFIG is not up-to-date. Regenerate it with \`scripts-dev/generate_sample_config\`.\e[0m" >&2
         exit 1
     }
+    diff -u "$SAMPLE_LOG_CONFIG" <(./scripts/generate_log_config) >/dev/null || {
+        echo -e "\e[1m\e[31m$SAMPLE_LOG_CONFIG is not up-to-date. Regenerate it with \`scripts-dev/generate_sample_config\`.\e[0m" >&2
+        exit 1
+    }
 else
     ./scripts/generate_config --header-file docs/.sample_config_header.yaml -o "$SAMPLE_CONFIG"
+    ./scripts/generate_log_config -o "$SAMPLE_LOG_CONFIG"
 fi

scripts-dev/make_full_schema.sh

@@ -0,0 +1,184 @@
+#!/bin/bash
+#
+# This script generates SQL files for creating a brand new Synapse DB with the latest
+# schema, on both SQLite3 and Postgres.
+#
+# It does so by having Synapse generate an up-to-date SQLite DB, then running
+# synapse_port_db to convert it to Postgres. It then dumps the contents of both.
+
+POSTGRES_HOST="localhost"
+POSTGRES_DB_NAME="synapse_full_schema.$$"
+
+SQLITE_FULL_SCHEMA_OUTPUT_FILE="full.sql.sqlite"
+POSTGRES_FULL_SCHEMA_OUTPUT_FILE="full.sql.postgres"
+
+REQUIRED_DEPS=("matrix-synapse" "psycopg2")
+
+usage() {
+  echo
+  echo "Usage: $0 -p <postgres_username> -o <path> [-c] [-n] [-h]"
+  echo
+  echo "-p <postgres_username>"
+  echo "  Username to connect to local postgres instance. The password will be requested"
+  echo "  during script execution."
+  echo "-c"
+  echo "  CI mode. Enables coverage tracking and prints every command that the script runs."
+  echo "-o <path>"
+  echo "  Directory to output full schema files to."
+  echo "-h"
+  echo "  Display this help text."
+}
+
+while getopts "p:co:h" opt; do
+  case $opt in
+    p)
+      POSTGRES_USERNAME=$OPTARG
+      ;;
+    c)
+      # Print all commands that are being executed
+      set -x
+
+      # Modify required dependencies for coverage
+      REQUIRED_DEPS+=("coverage" "coverage-enable-subprocess")
+
+      COVERAGE=1
+      ;;
+    o)
+      command -v realpath > /dev/null || (echo "The -o flag requires the 'realpath' binary to be installed" && exit 1)
+      OUTPUT_DIR="$(realpath "$OPTARG")"
+      ;;
+    h)
+      usage
+      exit
+      ;;
+    \?)
+      echo "ERROR: Invalid option: -$OPTARG" >&2
+      usage
+      exit
+      ;;
+  esac
+done
+
+# Check that required dependencies are installed
+unsatisfied_requirements=()
+for dep in "${REQUIRED_DEPS[@]}"; do
+  pip show "$dep" --quiet || unsatisfied_requirements+=("$dep")
+done
+if [ ${#unsatisfied_requirements} -ne 0 ]; then
+  echo "Please install the following python packages: ${unsatisfied_requirements[*]}"
+  exit 1
+fi
+
+if [ -z "$POSTGRES_USERNAME" ]; then
+  echo "No postgres username supplied"
+  usage
+  exit 1
+fi
+
+if [ -z "$OUTPUT_DIR" ]; then
+  echo "No output directory supplied"
+  usage
+  exit 1
+fi
+
+# Create the output directory if it doesn't exist
+mkdir -p "$OUTPUT_DIR"
+
+read -rsp "Postgres password for '$POSTGRES_USERNAME': " POSTGRES_PASSWORD
+echo ""
+
+# Exit immediately if a command fails
+set -e
+
+# cd to root of the synapse directory
+cd "$(dirname "$0")/.."
+
+# Create temporary SQLite and Postgres homeserver db configs and key file
+TMPDIR=$(mktemp -d)
+KEY_FILE=$TMPDIR/test.signing.key # default Synapse signing key path
+SQLITE_CONFIG=$TMPDIR/sqlite.conf
+SQLITE_DB=$TMPDIR/homeserver.db
+POSTGRES_CONFIG=$TMPDIR/postgres.conf
+
+# Ensure these files are delete on script exit
+trap 'rm -rf $TMPDIR' EXIT
+
+cat > "$SQLITE_CONFIG" <<EOF
+server_name: "test"
+
+signing_key_path: "$KEY_FILE"
+macaroon_secret_key: "abcde"
+
+report_stats: false
+
+database:
+  name: "sqlite3"
+  args:
+    database: "$SQLITE_DB"
+
+# Suppress the key server warning.
+trusted_key_servers: []
+EOF
+
+cat > "$POSTGRES_CONFIG" <<EOF
+server_name: "test"
+
+signing_key_path: "$KEY_FILE"
+macaroon_secret_key: "abcde"
+
+report_stats: false
+
+database:
+  name: "psycopg2"
+  args:
+    user: "$POSTGRES_USERNAME"
+    host: "$POSTGRES_HOST"
+    password: "$POSTGRES_PASSWORD"
+    database: "$POSTGRES_DB_NAME"
+
+# Suppress the key server warning.
+trusted_key_servers: []
+EOF
+
+# Generate the server's signing key.
+echo "Generating SQLite3 db schema..."
+python -m synapse.app.homeserver --generate-keys -c "$SQLITE_CONFIG"
+
+# Make sure the SQLite3 database is using the latest schema and has no pending background update.
+echo "Running db background jobs..."
+scripts-dev/update_database --database-config "$SQLITE_CONFIG"
+
+# Create the PostgreSQL database.
+echo "Creating postgres database..."
+createdb $POSTGRES_DB_NAME
+
+echo "Copying data from SQLite3 to Postgres with synapse_port_db..."
+if [ -z "$COVERAGE" ]; then
+  # No coverage needed
+  scripts/synapse_port_db --sqlite-database "$SQLITE_DB" --postgres-config "$POSTGRES_CONFIG"
+else
+  # Coverage desired
+  coverage run scripts/synapse_port_db --sqlite-database "$SQLITE_DB" --postgres-config "$POSTGRES_CONFIG"
+fi
+
+# Delete schema_version, applied_schema_deltas and applied_module_schemas tables
+# This needs to be done after synapse_port_db is run
+echo "Dropping unwanted db tables..."
+SQL="
+DROP TABLE schema_version;
+DROP TABLE applied_schema_deltas;
+DROP TABLE applied_module_schemas;
+"
+sqlite3 "$SQLITE_DB" <<< "$SQL"
+psql $POSTGRES_DB_NAME -U "$POSTGRES_USERNAME" -w <<< "$SQL"
+
+echo "Dumping SQLite3 schema to '$OUTPUT_DIR/$SQLITE_FULL_SCHEMA_OUTPUT_FILE'..."
+sqlite3 "$SQLITE_DB" ".dump" > "$OUTPUT_DIR/$SQLITE_FULL_SCHEMA_OUTPUT_FILE"
+
+echo "Dumping Postgres schema to '$OUTPUT_DIR/$POSTGRES_FULL_SCHEMA_OUTPUT_FILE'..."
+pg_dump --format=plain --no-tablespaces --no-acl --no-owner $POSTGRES_DB_NAME | sed -e '/^--/d' -e 's/public\.//g' -e '/^SET /d' -e '/^SELECT /d' > "$OUTPUT_DIR/$POSTGRES_FULL_SCHEMA_OUTPUT_FILE"
+
+echo "Cleaning up temporary Postgres database..."
+dropdb $POSTGRES_DB_NAME
+
+echo "Done! Files dumped to: $OUTPUT_DIR"

scripts-dev/update_database

@@ -26,8 +26,6 @@ from synapse.config.homeserver import HomeServerConfig
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.server import HomeServer
 from synapse.storage import DataStore
-from synapse.storage.engines import create_engine
-from synapse.storage.prepare_database import prepare_database
 
 logger = logging.getLogger("update_database")
 
@@ -35,21 +33,11 @@ logger = logging.getLogger("update_database")
 class MockHomeserver(HomeServer):
     DATASTORE_CLASS = DataStore
 
-    def __init__(self, config, database_engine, db_conn, **kwargs):
+    def __init__(self, config, **kwargs):
         super(MockHomeserver, self).__init__(
-            config.server_name,
-            reactor=reactor,
-            config=config,
-            database_engine=database_engine,
-            **kwargs
+            config.server_name, reactor=reactor, config=config, **kwargs
         )
 
-        self.database_engine = database_engine
-        self.db_conn = db_conn
-
-    def get_db_conn(self):
-        return self.db_conn
-
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(
@@ -85,25 +73,11 @@ if __name__ == "__main__":
     config = HomeServerConfig()
     config.parse_config_dict(hs_config, "", "")
 
-    # Create the database engine and a connection to it.
-    database_engine = create_engine(config.database_config)
-    db_conn = database_engine.module.connect(
-        **{
-            k: v
-            for k, v in config.database_config.get("args", {}).items()
-            if not k.startswith("cp_")
-        }
-    )
-
-    # Update the database to the latest schema.
-    prepare_database(db_conn, database_engine, config=config)
-    db_conn.commit()
-
     # Instantiate and initialise the homeserver object.
-    hs = MockHomeserver(
-        config, database_engine, db_conn, db_config=config.database_config,
-    )
-    # setup instantiates the store within the homeserver object.
+    hs = MockHomeserver(config)
+
+    # Setup instantiates the store within the homeserver object and updates the
+    # DB.
     hs.setup()
     store = hs.get_datastore()
 

scripts/export_signing_key

@@ -0,0 +1,94 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# Copyright 2019 The Matrix.org Foundation C.I.C.
+#
+# 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 argparse
+import sys
+import time
+from typing import Optional
+
+import nacl.signing
+from signedjson.key import encode_verify_key_base64, get_verify_key, read_signing_keys
+
+
+def exit(status: int = 0, message: Optional[str] = None):
+    if message:
+        print(message, file=sys.stderr)
+    sys.exit(status)
+
+
+def format_plain(public_key: nacl.signing.VerifyKey):
+    print(
+        "%s:%s %s"
+        % (public_key.alg, public_key.version, encode_verify_key_base64(public_key),)
+    )
+
+
+def format_for_config(public_key: nacl.signing.VerifyKey, expiry_ts: int):
+    print(
+        '  "%s:%s": { key: "%s", expired_ts: %i }'
+        % (
+            public_key.alg,
+            public_key.version,
+            encode_verify_key_base64(public_key),
+            expiry_ts,
+        )
+    )
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+
+    parser.add_argument(
+        "key_file", nargs="+", type=argparse.FileType("r"), help="The key file to read",
+    )
+
+    parser.add_argument(
+        "-x",
+        action="store_true",
+        dest="for_config",
+        help="format the output for inclusion in the old_signing_keys config setting",
+    )
+
+    parser.add_argument(
+        "--expiry-ts",
+        type=int,
+        default=int(time.time() * 1000) + 6*3600000,
+        help=(
+            "The expiry time to use for -x, in milliseconds since 1970. The default "
+            "is (now+6h)."
+        ),
+    )
+
+    args = parser.parse_args()
+
+    formatter = (
+        (lambda k: format_for_config(k, args.expiry_ts))
+        if args.for_config
+        else format_plain
+    )
+
+    keys = []
+    for file in args.key_file:
+        try:
+            res = read_signing_keys(file)
+        except Exception as e:
+            exit(
+                status=1,
+                message="Error reading key from file %s: %s %s"
+                % (file.name, type(e), e),
+            )
+            res = []
+        for key in res:
+            formatter(get_verify_key(key))

scripts/generate_log_config

@@ -0,0 +1,43 @@
+#!/usr/bin/env python3
+
+# -*- coding: utf-8 -*-
+# Copyright 2020 The Matrix.org Foundation C.I.C.
+#
+# 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 argparse
+import sys
+
+from synapse.config.logger import DEFAULT_LOG_CONFIG
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+
+    parser.add_argument(
+        "-o",
+        "--output-file",
+        type=argparse.FileType("w"),
+        default=sys.stdout,
+        help="File to write the configuration to. Default: stdout",
+    )
+
+    parser.add_argument(
+        "-f",
+        "--log-file",
+        type=str,
+        default="/var/log/matrix-synapse/homeserver.log",
+        help="name of the log file",
+    )
+
+    args = parser.parse_args()
+    args.output_file.write(DEFAULT_LOG_CONFIG.substitute(log_file=args.log_file))

scripts/synapse_port_db

@@ -30,6 +30,7 @@ import yaml
 from twisted.enterprise import adbapi
 from twisted.internet import defer, reactor
 
+from synapse.config.database import DatabaseConnectionConfig
 from synapse.config.homeserver import HomeServerConfig
 from synapse.logging.context import PreserveLoggingContext
 from synapse.storage._base import LoggingTransaction
@@ -50,12 +51,13 @@ from synapse.storage.data_stores.main.registration import (
 from synapse.storage.data_stores.main.room import RoomBackgroundUpdateStore
 from synapse.storage.data_stores.main.roommember import RoomMemberBackgroundUpdateStore
 from synapse.storage.data_stores.main.search import SearchBackgroundUpdateStore
-from synapse.storage.data_stores.main.state import StateBackgroundUpdateStore
+from synapse.storage.data_stores.main.state import MainStateBackgroundUpdateStore
 from synapse.storage.data_stores.main.stats import StatsStore
 from synapse.storage.data_stores.main.user_directory import (
     UserDirectoryBackgroundUpdateStore,
 )
-from synapse.storage.database import Database
+from synapse.storage.data_stores.state.bg_updates import StateBackgroundUpdateStore
+from synapse.storage.database import Database, make_conn
 from synapse.storage.engines import create_engine
 from synapse.storage.prepare_database import prepare_database
 from synapse.util import Clock
@@ -137,6 +139,7 @@ class Store(
     RoomMemberBackgroundUpdateStore,
     SearchBackgroundUpdateStore,
     StateBackgroundUpdateStore,
+    MainStateBackgroundUpdateStore,
     UserDirectoryBackgroundUpdateStore,
     StatsStore,
 ):
@@ -163,25 +166,24 @@ class Store(
             logger.exception("Failed to insert: %s", table)
             raise
 
+    def set_room_is_public(self, room_id, is_public):
+        raise Exception(
+            "Attempt to set room_is_public during port_db: database not empty?"
+        )
+
 
 class MockHomeserver:
-    def __init__(self, config, database_engine, db_conn, db_pool):
-        self.database_engine = database_engine
-        self.db_conn = db_conn
-        self.db_pool = db_pool
+    def __init__(self, config):
         self.clock = Clock(reactor)
         self.config = config
         self.hostname = config.server_name
 
-    def get_db_conn(self):
-        return self.db_conn
-
-    def get_db_pool(self):
-        return self.db_pool
-
     def get_clock(self):
         return self.clock
 
+    def get_reactor(self):
+        return reactor
+
 
 class Porter(object):
     def __init__(self, **kwargs):
@@ -445,45 +447,36 @@ class Porter(object):
             else:
                 return
 
-    def setup_db(self, db_config, database_engine):
-        db_conn = database_engine.module.connect(
-            **{
-                k: v
-                for k, v in db_config.get("args", {}).items()
-                if not k.startswith("cp_")
-            }
-        )
-
-        prepare_database(db_conn, database_engine, config=None)
+    def setup_db(self, db_config: DatabaseConnectionConfig, engine):
+        db_conn = make_conn(db_config, engine)
+        prepare_database(db_conn, engine, config=None)
 
         db_conn.commit()
 
         return db_conn
 
     @defer.inlineCallbacks
-    def build_db_store(self, config):
+    def build_db_store(self, db_config: DatabaseConnectionConfig):
         """Builds and returns a database store using the provided configuration.
 
         Args:
-            config: The database configuration, i.e. a dict following the structure of
-                the "database" section of Synapse's configuration file.
+            config: The database configuration
 
         Returns:
             The built Store object.
         """
-        engine = create_engine(config)
+        self.progress.set_state("Preparing %s" % db_config.config["name"])
 
-        self.progress.set_state("Preparing %s" % config["name"])
-        conn = self.setup_db(config, engine)
+        engine = create_engine(db_config.config)
+        conn = self.setup_db(db_config, engine)
 
-        db_pool = adbapi.ConnectionPool(config["name"], **config["args"])
+        hs = MockHomeserver(self.hs_config)
 
-        hs = MockHomeserver(self.hs_config, engine, conn, db_pool)
-
-        store = Store(Database(hs), conn, hs)
+        store = Store(Database(hs, db_config, engine), conn, hs)
 
         yield store.db.runInteraction(
-            "%s_engine.check_database" % config["name"], engine.check_database,
+            "%s_engine.check_database" % db_config.config["name"],
+            engine.check_database,
         )
 
         return store
@@ -509,7 +502,9 @@ class Porter(object):
     @defer.inlineCallbacks
     def run(self):
         try:
-            self.sqlite_store = yield self.build_db_store(self.sqlite_config)
+            self.sqlite_store = yield self.build_db_store(
+                DatabaseConnectionConfig("master-sqlite", self.sqlite_config)
+            )
 
             # Check if all background updates are done, abort if not.
             updates_complete = (
@@ -524,7 +519,7 @@ class Porter(object):
                 defer.returnValue(None)
 
             self.postgres_store = yield self.build_db_store(
-                self.hs_config.database_config
+                self.hs_config.get_single_database()
             )
 
             yield self.run_background_updates_on_postgres()

synapse/__init__.py

@@ -36,7 +36,7 @@ try:
 except ImportError:
     pass
 
-__version__ = "1.7.2"
+__version__ = "1.8.0"
 
 if bool(os.environ.get("SYNAPSE_TEST_PATCH_LOG_CONTEXTS", False)):
     # We import here so that we don't have to install a bunch of deps when

synapse/api/auth.py

@@ -79,7 +79,7 @@ class Auth(object):
 
     @defer.inlineCallbacks
     def check_from_context(self, room_version, event, context, do_sig_check=True):
-        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        prev_state_ids = yield context.get_prev_state_ids()
         auth_events_ids = yield self.compute_auth_events(
             event, prev_state_ids, for_verification=True
         )

synapse/api/urls.py

@@ -29,7 +29,6 @@ FEDERATION_V2_PREFIX = FEDERATION_PREFIX + "/v2"
 FEDERATION_UNSTABLE_PREFIX = FEDERATION_PREFIX + "/unstable"
 STATIC_PREFIX = "/_matrix/static"
 WEB_CLIENT_PREFIX = "/_matrix/client"
-CONTENT_REPO_PREFIX = "/_matrix/content"
 SERVER_KEY_V2_PREFIX = "/_matrix/key/v2"
 MEDIA_PREFIX = "/_matrix/media/r0"
 LEGACY_MEDIA_PREFIX = "/_matrix/media/v1"

synapse/app/_base.py

@@ -237,6 +237,12 @@ def start(hs, listeners=None):
     """
     Start a Synapse server or worker.
 
+    Should be called once the reactor is running and (if we're using ACME) the
+    TLS certificates are in place.
+
+    Will start the main HTTP listeners and do some other startup tasks, and then
+    notify systemd.
+
     Args:
         hs (synapse.server.HomeServer)
         listeners (list[dict]): Listener configuration ('listeners' in homeserver.yaml)
@@ -311,9 +317,7 @@ def setup_sdnotify(hs):
 
     # Tell systemd our state, if we're using it. This will silently fail if
     # we're not using systemd.
-    hs.get_reactor().addSystemEventTrigger(
-        "after", "startup", sdnotify, b"READY=1\nMAINPID=%i" % (os.getpid(),)
-    )
+    sdnotify(b"READY=1\nMAINPID=%i" % (os.getpid(),))
 
     hs.get_reactor().addSystemEventTrigger(
         "before", "shutdown", sdnotify, b"STOPPING=1"

synapse/app/admin_cmd.py

@@ -45,7 +45,6 @@ from synapse.replication.slave.storage.registration import SlavedRegistrationSto
 from synapse.replication.slave.storage.room import RoomStore
 from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.server import HomeServer
-from synapse.storage.engines import create_engine
 from synapse.util.logcontext import LoggingContext
 from synapse.util.versionstring import get_version_string
 
@@ -105,8 +104,10 @@ def export_data_command(hs, args):
     user_id = args.user_id
     directory = args.output_directory
 
-    res = yield hs.get_handlers().admin_handler.export_user_data(
-        user_id, FileExfiltrationWriter(user_id, directory=directory)
+    res = yield defer.ensureDeferred(
+        hs.get_handlers().admin_handler.export_user_data(
+            user_id, FileExfiltrationWriter(user_id, directory=directory)
+        )
     )
     print(res)
 
@@ -229,14 +230,10 @@ def start(config_options):
 
     synapse.events.USE_FROZEN_DICTS = config.use_frozen_dicts
 
-    database_engine = create_engine(config.database_config)
-
     ss = AdminCmdServer(
         config.server_name,
-        db_config=config.database_config,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
-        database_engine=database_engine,
     )
 
     setup_logging(ss, config, use_worker_options=True)

synapse/app/appservice.py

@@ -34,7 +34,6 @@ from synapse.replication.slave.storage.events import SlavedEventStore
 from synapse.replication.slave.storage.registration import SlavedRegistrationStore
 from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.server import HomeServer
-from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
@@ -143,8 +142,6 @@ def start(config_options):
 
     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"
@@ -159,10 +156,8 @@ def start(config_options):
 
     ps = AppserviceServer(
         config.server_name,
-        db_config=config.database_config,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
-        database_engine=database_engine,
     )
 
     setup_logging(ps, config, use_worker_options=True)

synapse/app/client_reader.py

@@ -62,7 +62,6 @@ from synapse.rest.client.v2_alpha.keys import KeyChangesServlet, KeyQueryServlet
 from synapse.rest.client.v2_alpha.register import RegisterRestServlet
 from synapse.rest.client.versions import VersionsRestServlet
 from synapse.server import HomeServer
-from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
@@ -181,14 +180,10 @@ def start(config_options):
 
     events.USE_FROZEN_DICTS = config.use_frozen_dicts
 
-    database_engine = create_engine(config.database_config)
-
     ss = ClientReaderServer(
         config.server_name,
-        db_config=config.database_config,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
-        database_engine=database_engine,
     )
 
     setup_logging(ss, config, use_worker_options=True)

synapse/app/event_creator.py

@@ -57,7 +57,6 @@ from synapse.rest.client.v1.room import (
 )
 from synapse.server import HomeServer
 from synapse.storage.data_stores.main.user_directory import UserDirectoryStore
-from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
@@ -180,14 +179,10 @@ def start(config_options):
 
     events.USE_FROZEN_DICTS = config.use_frozen_dicts
 
-    database_engine = create_engine(config.database_config)
-
     ss = EventCreatorServer(
         config.server_name,
-        db_config=config.database_config,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
-        database_engine=database_engine,
     )
 
     setup_logging(ss, config, use_worker_options=True)

synapse/app/federation_reader.py

@@ -46,7 +46,6 @@ from synapse.replication.slave.storage.transactions import SlavedTransactionStor
 from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.rest.key.v2 import KeyApiV2Resource
 from synapse.server import HomeServer
-from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
@@ -162,14 +161,10 @@ def start(config_options):
 
     events.USE_FROZEN_DICTS = config.use_frozen_dicts
 
-    database_engine = create_engine(config.database_config)
-
     ss = FederationReaderServer(
         config.server_name,
-        db_config=config.database_config,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
-        database_engine=database_engine,
     )
 
     setup_logging(ss, config, use_worker_options=True)

synapse/app/federation_sender.py

@@ -41,7 +41,6 @@ from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.replication.tcp.streams._base import ReceiptsStream
 from synapse.server import HomeServer
 from synapse.storage.database import Database
-from synapse.storage.engines import create_engine
 from synapse.types import ReadReceipt
 from synapse.util.async_helpers import Linearizer
 from synapse.util.httpresourcetree import create_resource_tree
@@ -174,8 +173,6 @@ def start(config_options):
 
     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"
@@ -190,10 +187,8 @@ def start(config_options):
 
     ss = FederationSenderServer(
         config.server_name,
-        db_config=config.database_config,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
-        database_engine=database_engine,
     )
 
     setup_logging(ss, config, use_worker_options=True)

synapse/app/frontend_proxy.py

@@ -39,7 +39,6 @@ from synapse.replication.slave.storage.registration import SlavedRegistrationSto
 from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.rest.client.v2_alpha._base import client_patterns
 from synapse.server import HomeServer
-from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
@@ -234,14 +233,10 @@ def start(config_options):
 
     events.USE_FROZEN_DICTS = config.use_frozen_dicts
 
-    database_engine = create_engine(config.database_config)
-
     ss = FrontendProxyServer(
         config.server_name,
-        db_config=config.database_config,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
-        database_engine=database_engine,
     )
 
     setup_logging(ss, config, use_worker_options=True)

synapse/app/homeserver.py

@@ -39,7 +39,6 @@ import synapse
 import synapse.config.logger
 from synapse import events
 from synapse.api.urls import (
-    CONTENT_REPO_PREFIX,
     FEDERATION_PREFIX,
     LEGACY_MEDIA_PREFIX,
     MEDIA_PREFIX,
@@ -65,11 +64,10 @@ from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory
 from synapse.rest import ClientRestResource
 from synapse.rest.admin import AdminRestResource
 from synapse.rest.key.v2 import KeyApiV2Resource
-from synapse.rest.media.v0.content_repository import ContentRepoResource
 from synapse.rest.well_known import WellKnownResource
 from synapse.server import HomeServer
 from synapse.storage import DataStore
-from synapse.storage.engines import IncorrectDatabaseSetup, create_engine
+from synapse.storage.engines import IncorrectDatabaseSetup
 from synapse.storage.prepare_database import UpgradeDatabaseException
 from synapse.util.caches import CACHE_SIZE_FACTOR
 from synapse.util.httpresourcetree import create_resource_tree
@@ -223,13 +221,7 @@ class SynapseHomeServer(HomeServer):
             if self.get_config().enable_media_repo:
                 media_repo = self.get_media_repository_resource()
                 resources.update(
-                    {
-                        MEDIA_PREFIX: media_repo,
-                        LEGACY_MEDIA_PREFIX: media_repo,
-                        CONTENT_REPO_PREFIX: ContentRepoResource(
-                            self, self.config.uploads_path
-                        ),
-                    }
+                    {MEDIA_PREFIX: media_repo, LEGACY_MEDIA_PREFIX: media_repo}
                 )
             elif name == "media":
                 raise ConfigError(
@@ -318,7 +310,7 @@ def setup(config_options):
             "Synapse Homeserver", config_options
         )
     except ConfigError as e:
-        sys.stderr.write("\n" + str(e) + "\n")
+        sys.stderr.write("\nERROR: %s\n" % (e,))
         sys.exit(1)
 
     if not config:
@@ -328,15 +320,10 @@ def setup(config_options):
 
     events.USE_FROZEN_DICTS = config.use_frozen_dicts
 
-    database_engine = create_engine(config.database_config)
-    config.database_config["args"]["cp_openfun"] = database_engine.on_new_connection
-
     hs = SynapseHomeServer(
         config.server_name,
-        db_config=config.database_config,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
-        database_engine=database_engine,
     )
 
     synapse.config.logger.setup_logging(hs, config, use_worker_options=False)
@@ -347,13 +334,8 @@ def setup(config_options):
         hs.setup()
     except IncorrectDatabaseSetup as e:
         quit_with_error(str(e))
-    except UpgradeDatabaseException:
-        sys.stderr.write(
-            "\nFailed to upgrade database.\n"
-            "Have you checked for version specific instructions in"
-            " UPGRADES.rst?\n"
-        )
-        sys.exit(1)
+    except UpgradeDatabaseException as e:
+        quit_with_error("Failed to upgrade database: %s" % (e,))
 
     hs.setup_master()
 
@@ -519,8 +501,10 @@ def phone_stats_home(hs, stats, stats_process=_stats_process):
     # Database version
     #
 
-    stats["database_engine"] = hs.database_engine.module.__name__
-    stats["database_server_version"] = hs.database_engine.server_version
+    # This only reports info about the *main* database.
+    stats["database_engine"] = hs.get_datastore().db.engine.module.__name__
+    stats["database_server_version"] = hs.get_datastore().db.engine.server_version
+
     logger.info("Reporting stats to %s: %s" % (hs.config.report_stats_endpoint, stats))
     try:
         yield hs.get_proxied_http_client().put_json(

synapse/app/media_repository.py

@@ -21,7 +21,7 @@ from twisted.web.resource import NoResource
 
 import synapse
 from synapse import events
-from synapse.api.urls import CONTENT_REPO_PREFIX, LEGACY_MEDIA_PREFIX, MEDIA_PREFIX
+from synapse.api.urls import LEGACY_MEDIA_PREFIX, MEDIA_PREFIX
 from synapse.app import _base
 from synapse.config._base import ConfigError
 from synapse.config.homeserver import HomeServerConfig
@@ -37,10 +37,8 @@ from synapse.replication.slave.storage.registration import SlavedRegistrationSto
 from synapse.replication.slave.storage.transactions import SlavedTransactionStore
 from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.rest.admin import register_servlets_for_media_repo
-from synapse.rest.media.v0.content_repository import ContentRepoResource
 from synapse.server import HomeServer
 from synapse.storage.data_stores.main.media_repository import MediaRepositoryStore
-from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
@@ -83,9 +81,6 @@ class MediaRepositoryServer(HomeServer):
                         {
                             MEDIA_PREFIX: media_repo,
                             LEGACY_MEDIA_PREFIX: media_repo,
-                            CONTENT_REPO_PREFIX: ContentRepoResource(
-                                self, self.config.uploads_path
-                            ),
                             "/_synapse/admin": admin_resource,
                         }
                     )
@@ -157,14 +152,10 @@ def start(config_options):
 
     events.USE_FROZEN_DICTS = config.use_frozen_dicts
 
-    database_engine = create_engine(config.database_config)
-
     ss = MediaRepositoryServer(
         config.server_name,
-        db_config=config.database_config,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
-        database_engine=database_engine,
     )
 
     setup_logging(ss, config, use_worker_options=True)

synapse/app/pusher.py

@@ -37,7 +37,6 @@ from synapse.replication.slave.storage.room import RoomStore
 from synapse.replication.tcp.client import ReplicationClientHandler
 from synapse.server import HomeServer
 from synapse.storage import DataStore
-from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
@@ -203,14 +202,10 @@ def start(config_options):
     # 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,
     )
 
     setup_logging(ps, config, use_worker_options=True)

synapse/app/synchrotron.py

@@ -48,14 +48,13 @@ from synapse.replication.slave.storage.receipts import SlavedReceiptsStore
 from synapse.replication.slave.storage.registration import SlavedRegistrationStore
 from synapse.replication.slave.storage.room import RoomStore
 from synapse.replication.tcp.client import ReplicationClientHandler
-from synapse.replication.tcp.streams.events import EventsStreamEventRow
+from synapse.replication.tcp.streams.events import EventsStreamEventRow, EventsStreamRow
 from synapse.rest.client.v1 import events
 from synapse.rest.client.v1.initial_sync import InitialSyncRestServlet
 from synapse.rest.client.v1.room import RoomInitialSyncRestServlet
 from synapse.rest.client.v2_alpha import sync
 from synapse.server import HomeServer
 from synapse.storage.data_stores.main.presence import UserPresenceState
-from synapse.storage.engines import create_engine
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.manhole import manhole
 from synapse.util.stringutils import random_string
@@ -372,8 +371,7 @@ class SyncReplicationHandler(ReplicationClientHandler):
     def get_currently_syncing_users(self):
         return self.presence_handler.get_currently_syncing_users()
 
-    @defer.inlineCallbacks
-    def process_and_notify(self, stream_name, token, rows):
+    async def process_and_notify(self, stream_name, token, rows):
         try:
             if stream_name == "events":
                 # We shouldn't get multiple rows per token for events stream, so
@@ -381,7 +379,14 @@ class SyncReplicationHandler(ReplicationClientHandler):
                 for row in rows:
                     if row.type != EventsStreamEventRow.TypeId:
                         continue
-                    event = yield self.store.get_event(row.data.event_id)
+                    assert isinstance(row, EventsStreamRow)
+
+                    event = await self.store.get_event(
+                        row.data.event_id, allow_rejected=True
+                    )
+                    if event.rejected_reason:
+                        continue
+
                     extra_users = ()
                     if event.type == EventTypes.Member:
                         extra_users = (event.state_key,)
@@ -413,11 +418,11 @@ class SyncReplicationHandler(ReplicationClientHandler):
             elif stream_name == "device_lists":
                 all_room_ids = set()
                 for row in rows:
-                    room_ids = yield self.store.get_rooms_for_user(row.user_id)
+                    room_ids = await self.store.get_rooms_for_user(row.user_id)
                     all_room_ids.update(room_ids)
                 self.notifier.on_new_event("device_list_key", token, rooms=all_room_ids)
             elif stream_name == "presence":
-                yield self.presence_handler.process_replication_rows(token, rows)
+                await self.presence_handler.process_replication_rows(token, rows)
             elif stream_name == "receipts":
                 self.notifier.on_new_event(
                     "groups_key", token, users=[row.user_id for row in rows]
@@ -437,14 +442,10 @@ def start(config_options):
 
     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(),
     )
 

synapse/app/user_dir.py

@@ -44,7 +44,6 @@ from synapse.rest.client.v2_alpha import user_directory
 from synapse.server import HomeServer
 from synapse.storage.data_stores.main.user_directory import UserDirectoryStore
 from synapse.storage.database import Database
-from synapse.storage.engines import create_engine
 from synapse.util.caches.stream_change_cache import StreamChangeCache
 from synapse.util.httpresourcetree import create_resource_tree
 from synapse.util.manhole import manhole
@@ -200,8 +199,6 @@ def start(config_options):
 
     events.USE_FROZEN_DICTS = config.use_frozen_dicts
 
-    database_engine = create_engine(config.database_config)
-
     if config.update_user_directory:
         sys.stderr.write(
             "\nThe update_user_directory must be disabled in the main synapse process"
@@ -216,10 +213,8 @@ def start(config_options):
 
     ss = UserDirectoryServer(
         config.server_name,
-        db_config=config.database_config,
         config=config,
         version_string="Synapse/" + get_version_string(synapse),
-        database_engine=database_engine,
     )
 
     setup_logging(ss, config, use_worker_options=True)

synapse/config/database.py

@@ -12,12 +12,45 @@
 # 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
 import os
 from textwrap import indent
 
 import yaml
 
-from ._base import Config
+from synapse.config._base import Config, ConfigError
+
+logger = logging.getLogger(__name__)
+
+
+class DatabaseConnectionConfig:
+    """Contains the connection config for a particular database.
+
+    Args:
+        name: A label for the database, used for logging.
+        db_config: The config for a particular database, as per `database`
+            section of main config. Has three fields: `name` for database
+            module name, `args` for the args to give to the database
+            connector, and optional `data_stores` that is a list of stores to
+            provision on this database (defaulting to all).
+    """
+
+    def __init__(self, name: str, db_config: dict):
+        if db_config["name"] not in ("sqlite3", "psycopg2"):
+            raise ConfigError("Unsupported database type %r" % (db_config["name"],))
+
+        if db_config["name"] == "sqlite3":
+            db_config.setdefault("args", {}).update(
+                {"cp_min": 1, "cp_max": 1, "check_same_thread": False}
+            )
+
+        data_stores = db_config.get("data_stores")
+        if data_stores is None:
+            data_stores = ["main", "state"]
+
+        self.name = name
+        self.config = db_config
+        self.data_stores = data_stores
 
 
 class DatabaseConfig(Config):
@@ -26,22 +59,43 @@ class DatabaseConfig(Config):
     def read_config(self, config, **kwargs):
         self.event_cache_size = self.parse_size(config.get("event_cache_size", "10K"))
 
-        self.database_config = config.get("database")
+        # We *experimentally* support specifying multiple databases via the
+        # `databases` key. This is a map from a label to database config in the
+        # same format as the `database` config option, plus an extra
+        # `data_stores` key to specify which data store goes where. For example:
+        #
+        #   databases:
+        #       master:
+        #           name: psycopg2
+        #           data_stores: ["main"]
+        #           args: {}
+        #       state:
+        #           name: psycopg2
+        #           data_stores: ["state"]
+        #           args: {}
 
-        if self.database_config is None:
-            self.database_config = {"name": "sqlite3", "args": {}}
+        multi_database_config = config.get("databases")
+        database_config = config.get("database")
+
+        if multi_database_config and database_config:
+            raise ConfigError("Can't specify both 'database' and 'datbases' in config")
+
+        if multi_database_config:
+            if config.get("database_path"):
+                raise ConfigError("Can't specify 'database_path' with 'databases'")
+
+            self.databases = [
+                DatabaseConnectionConfig(name, db_conf)
+                for name, db_conf in multi_database_config.items()
+            ]
 
-        name = self.database_config.get("name", None)
-        if name == "psycopg2":
-            pass
-        elif name == "sqlite3":
-            self.database_config.setdefault("args", {}).update(
-                {"cp_min": 1, "cp_max": 1, "check_same_thread": False}
-            )
         else:
-            raise RuntimeError("Unsupported database type '%s'" % (name,))
+            if database_config is None:
+                database_config = {"name": "sqlite3", "args": {}}
 
-        self.set_databasepath(config.get("database_path"))
+            self.databases = [DatabaseConnectionConfig("master", database_config)]
+
+            self.set_databasepath(config.get("database_path"))
 
     def generate_config_section(self, data_dir_path, database_conf, **kwargs):
         if not database_conf:
@@ -76,11 +130,24 @@ class DatabaseConfig(Config):
         self.set_databasepath(args.database_path)
 
     def set_databasepath(self, database_path):
+        if database_path is None:
+            return
+
         if database_path != ":memory:":
             database_path = self.abspath(database_path)
-        if self.database_config.get("name", None) == "sqlite3":
-            if database_path is not None:
-                self.database_config["args"]["database"] = database_path
+
+        # We only support setting a database path if we have a single sqlite3
+        # database.
+        if len(self.databases) != 1:
+            raise ConfigError("Cannot specify 'database_path' with multiple databases")
+
+        database = self.get_single_database()
+        if database.config["name"] != "sqlite3":
+            # We don't raise here as we haven't done so before for this case.
+            logger.warn("Ignoring 'database_path' for non-sqlite3 database")
+            return
+
+        database.config["args"]["database"] = database_path
 
     @staticmethod
     def add_arguments(parser):
@@ -91,3 +158,11 @@ class DatabaseConfig(Config):
             metavar="SQLITE_DATABASE_PATH",
             help="The path to a sqlite database to use.",
         )
+
+    def get_single_database(self) -> DatabaseConnectionConfig:
+        """Returns the database if there is only one, useful for e.g. tests
+        """
+        if len(self.databases) != 1:
+            raise Exception("More than one database exists")
+
+        return self.databases[0]

synapse/config/emailconfig.py

@@ -21,6 +21,7 @@ from __future__ import print_function
 import email.utils
 import os
 from enum import Enum
+from typing import Optional
 
 import pkg_resources
 
@@ -101,7 +102,7 @@ class EmailConfig(Config):
                 # both in RegistrationConfig and here. We should factor this bit out
                 self.account_threepid_delegate_email = self.trusted_third_party_id_servers[
                     0
-                ]
+                ]  # type: Optional[str]
                 self.using_identity_server_from_trusted_list = True
             else:
                 raise ConfigError(

synapse/config/key.py

@@ -108,7 +108,7 @@ class KeyConfig(Config):
             self.signing_key = self.read_signing_keys(signing_key_path, "signing_key")
 
         self.old_signing_keys = self.read_old_signing_keys(
-            config.get("old_signing_keys", {})
+            config.get("old_signing_keys")
         )
         self.key_refresh_interval = self.parse_duration(
             config.get("key_refresh_interval", "1d")
@@ -199,14 +199,19 @@ class KeyConfig(Config):
         signing_key_path: "%(base_key_name)s.signing.key"
 
         # The keys that the server used to sign messages with but won't use
-        # to sign new messages. E.g. it has lost its private key
+        # to sign new messages.
         #
-        #old_signing_keys:
-        #  "ed25519:auto":
-        #    # Base64 encoded public key
-        #    key: "The public part of your old signing key."
-        #    # Millisecond POSIX timestamp when the key expired.
-        #    expired_ts: 123456789123
+        old_signing_keys:
+          # For each key, `key` should be the base64-encoded public key, and
+          # `expired_ts`should be the time (in milliseconds since the unix epoch) that
+          # it was last used.
+          #
+          # It is possible to build an entry from an old signing.key file using the
+          # `export_signing_key` script which is provided with synapse.
+          #
+          # For example:
+          #
+          #"ed25519:id": { key: "base64string", expired_ts: 123456789123 }
 
         # How long key response published by this server is valid for.
         # Used to set the valid_until_ts in /key/v2 APIs.
@@ -290,6 +295,8 @@ class KeyConfig(Config):
             raise ConfigError("Error reading %s: %s" % (name, str(e)))
 
     def read_old_signing_keys(self, old_signing_keys):
+        if old_signing_keys is None:
+            return {}
         keys = {}
         for key_id, key_data in old_signing_keys.items():
             if is_signing_algorithm_supported(key_id):

synapse/config/logger.py

@@ -12,7 +12,7 @@
 # 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 argparse
 import logging
 import logging.config
 import os
@@ -37,10 +37,17 @@ from synapse.logging._structured import (
 from synapse.logging.context import LoggingContextFilter
 from synapse.util.versionstring import get_version_string
 
-from ._base import Config
+from ._base import Config, ConfigError
 
 DEFAULT_LOG_CONFIG = Template(
-    """
+    """\
+# Log configuration for Synapse.
+#
+# This is a YAML file containing a standard Python logging configuration
+# dictionary. See [1] for details on the valid settings.
+#
+# [1]: https://docs.python.org/3.7/library/logging.config.html#configuration-dictionary-schema
+
 version: 1
 
 formatters:
@@ -81,11 +88,18 @@ disable_existing_loggers: false
 """
 )
 
+LOG_FILE_ERROR = """\
+Support for the log_file configuration option and --log-file command-line option was
+removed in Synapse 1.3.0. You should instead set up a separate log configuration file.
+"""
+
 
 class LoggingConfig(Config):
     section = "logging"
 
     def read_config(self, config, **kwargs):
+        if config.get("log_file"):
+            raise ConfigError(LOG_FILE_ERROR)
         self.log_config = self.abspath(config.get("log_config"))
         self.no_redirect_stdio = config.get("no_redirect_stdio", False)
 
@@ -106,6 +120,8 @@ class LoggingConfig(Config):
     def read_arguments(self, args):
         if args.no_redirect_stdio is not None:
             self.no_redirect_stdio = args.no_redirect_stdio
+        if args.log_file is not None:
+            raise ConfigError(LOG_FILE_ERROR)
 
     @staticmethod
     def add_arguments(parser):
@@ -118,6 +134,10 @@ class LoggingConfig(Config):
             help="Do not redirect stdout/stderr to the log",
         )
 
+        logging_group.add_argument(
+            "-f", "--log-file", dest="log_file", help=argparse.SUPPRESS,
+        )
+
     def generate_files(self, config, config_dir_path):
         log_config = config.get("log_config")
         if log_config and not os.path.exists(log_config):

synapse/config/ratelimiting.py

@@ -83,10 +83,9 @@ class RatelimitConfig(Config):
         )
 
         rc_admin_redaction = config.get("rc_admin_redaction")
+        self.rc_admin_redaction = None
         if rc_admin_redaction:
             self.rc_admin_redaction = RateLimitConfig(rc_admin_redaction)
-        else:
-            self.rc_admin_redaction = None
 
     def generate_config_section(self, **kwargs):
         return """\

synapse/config/repository.py

@@ -156,7 +156,6 @@ class ContentRepositoryConfig(Config):
                 (provider_class, parsed_config, wrapper_config)
             )
 
-        self.uploads_path = self.ensure_directory(config.get("uploads_path", "uploads"))
         self.dynamic_thumbnails = config.get("dynamic_thumbnails", False)
         self.thumbnail_requirements = parse_thumbnail_requirements(
             config.get("thumbnail_sizes", DEFAULT_THUMBNAIL_SIZES)
@@ -231,10 +230,6 @@ class ContentRepositoryConfig(Config):
         #    config:
         #       directory: /mnt/some/other/directory
 
-        # Directory where in-progress uploads are stored.
-        #
-        uploads_path: "%(uploads_path)s"
-
         # The largest allowed upload size in bytes
         #
         #max_upload_size: 10M

synapse/config/saml2_config.py

@@ -14,17 +14,19 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import re
+import logging
 
 from synapse.python_dependencies import DependencyException, check_requirements
-from synapse.types import (
-    map_username_to_mxid_localpart,
-    mxid_localpart_allowed_characters,
-)
-from synapse.util.module_loader import load_python_module
+from synapse.util.module_loader import load_module, load_python_module
 
 from ._base import Config, ConfigError
 
+logger = logging.getLogger(__name__)
+
+DEFAULT_USER_MAPPING_PROVIDER = (
+    "synapse.handlers.saml_handler.DefaultSamlMappingProvider"
+)
+
 
 def _dict_merge(merge_dict, into_dict):
     """Do a deep merge of two dicts
@@ -75,15 +77,69 @@ class SAML2Config(Config):
 
         self.saml2_enabled = True
 
-        self.saml2_mxid_source_attribute = saml2_config.get(
-            "mxid_source_attribute", "uid"
-        )
-
         self.saml2_grandfathered_mxid_source_attribute = saml2_config.get(
             "grandfathered_mxid_source_attribute", "uid"
         )
 
-        saml2_config_dict = self._default_saml_config_dict()
+        # user_mapping_provider may be None if the key is present but has no value
+        ump_dict = saml2_config.get("user_mapping_provider") or {}
+
+        # Use the default user mapping provider if not set
+        ump_dict.setdefault("module", DEFAULT_USER_MAPPING_PROVIDER)
+
+        # Ensure a config is present
+        ump_dict["config"] = ump_dict.get("config") or {}
+
+        if ump_dict["module"] == DEFAULT_USER_MAPPING_PROVIDER:
+            # Load deprecated options for use by the default module
+            old_mxid_source_attribute = saml2_config.get("mxid_source_attribute")
+            if old_mxid_source_attribute:
+                logger.warning(
+                    "The config option saml2_config.mxid_source_attribute is deprecated. "
+                    "Please use saml2_config.user_mapping_provider.config"
+                    ".mxid_source_attribute instead."
+                )
+                ump_dict["config"]["mxid_source_attribute"] = old_mxid_source_attribute
+
+            old_mxid_mapping = saml2_config.get("mxid_mapping")
+            if old_mxid_mapping:
+                logger.warning(
+                    "The config option saml2_config.mxid_mapping is deprecated. Please "
+                    "use saml2_config.user_mapping_provider.config.mxid_mapping instead."
+                )
+                ump_dict["config"]["mxid_mapping"] = old_mxid_mapping
+
+        # Retrieve an instance of the module's class
+        # Pass the config dictionary to the module for processing
+        (
+            self.saml2_user_mapping_provider_class,
+            self.saml2_user_mapping_provider_config,
+        ) = load_module(ump_dict)
+
+        # Ensure loaded user mapping module has defined all necessary methods
+        # Note parse_config() is already checked during the call to load_module
+        required_methods = [
+            "get_saml_attributes",
+            "saml_response_to_user_attributes",
+        ]
+        missing_methods = [
+            method
+            for method in required_methods
+            if not hasattr(self.saml2_user_mapping_provider_class, method)
+        ]
+        if missing_methods:
+            raise ConfigError(
+                "Class specified by saml2_config."
+                "user_mapping_provider.module is missing required "
+                "methods: %s" % (", ".join(missing_methods),)
+            )
+
+        # Get the desired saml auth response attributes from the module
+        saml2_config_dict = self._default_saml_config_dict(
+            *self.saml2_user_mapping_provider_class.get_saml_attributes(
+                self.saml2_user_mapping_provider_config
+            )
+        )
         _dict_merge(
             merge_dict=saml2_config.get("sp_config", {}), into_dict=saml2_config_dict
         )
@@ -103,22 +159,27 @@ class SAML2Config(Config):
             saml2_config.get("saml_session_lifetime", "5m")
         )
 
-        mapping = saml2_config.get("mxid_mapping", "hexencode")
-        try:
-            self.saml2_mxid_mapper = MXID_MAPPER_MAP[mapping]
-        except KeyError:
-            raise ConfigError("%s is not a known mxid_mapping" % (mapping,))
+    def _default_saml_config_dict(
+        self, required_attributes: set, optional_attributes: set
+    ):
+        """Generate a configuration dictionary with required and optional attributes that
+        will be needed to process new user registration
 
-    def _default_saml_config_dict(self):
+        Args:
+            required_attributes: SAML auth response attributes that are
+                necessary to function
+            optional_attributes: SAML auth response attributes that can be used to add
+                additional information to Synapse user accounts, but are not required
+
+        Returns:
+            dict: A SAML configuration dictionary
+        """
         import saml2
 
         public_baseurl = self.public_baseurl
         if public_baseurl is None:
             raise ConfigError("saml2_config requires a public_baseurl to be set")
 
-        required_attributes = {"uid", self.saml2_mxid_source_attribute}
-
-        optional_attributes = {"displayName"}
         if self.saml2_grandfathered_mxid_source_attribute:
             optional_attributes.add(self.saml2_grandfathered_mxid_source_attribute)
         optional_attributes -= required_attributes
@@ -207,33 +268,58 @@ class SAML2Config(Config):
           #
           #config_path: "%(config_dir_path)s/sp_conf.py"
 
-          # the lifetime of a SAML session. This defines how long a user has to
+          # The lifetime of a SAML session. This defines how long a user has to
           # complete the authentication process, if allow_unsolicited is unset.
           # The default is 5 minutes.
           #
           #saml_session_lifetime: 5m
 
-          # The SAML attribute (after mapping via the attribute maps) to use to derive
-          # the Matrix ID from. 'uid' by default.
+          # An external module can be provided here as a custom solution to
+          # mapping attributes returned from a saml provider onto a matrix user.
           #
-          #mxid_source_attribute: displayName
+          user_mapping_provider:
+            # The custom module's class. Uncomment to use a custom module.
+            #
+            #module: mapping_provider.SamlMappingProvider
 
-          # The mapping system to use for mapping the saml attribute onto a matrix ID.
-          # Options include:
-          #  * 'hexencode' (which maps unpermitted characters to '=xx')
-          #  * 'dotreplace' (which replaces unpermitted characters with '.').
-          # The default is 'hexencode'.
-          #
-          #mxid_mapping: dotreplace
+            # Custom configuration values for the module. Below options are
+            # intended for the built-in provider, they should be changed if
+            # using a custom module. This section will be passed as a Python
+            # dictionary to the module's `parse_config` method.
+            #
+            config:
+              # The SAML attribute (after mapping via the attribute maps) to use
+              # to derive the Matrix ID from. 'uid' by default.
+              #
+              # Note: This used to be configured by the
+              # saml2_config.mxid_source_attribute option. If that is still
+              # defined, its value will be used instead.
+              #
+              #mxid_source_attribute: displayName
 
-          # In previous versions of synapse, the mapping from SAML attribute to MXID was
-          # always calculated dynamically rather than stored in a table. For backwards-
-          # compatibility, we will look for user_ids matching such a pattern before
-          # creating a new account.
+              # The mapping system to use for mapping the saml attribute onto a
+              # matrix ID.
+              #
+              # Options include:
+              #  * 'hexencode' (which maps unpermitted characters to '=xx')
+              #  * 'dotreplace' (which replaces unpermitted characters with
+              #     '.').
+              # The default is 'hexencode'.
+              #
+              # Note: This used to be configured by the
+              # saml2_config.mxid_mapping option. If that is still defined, its
+              # value will be used instead.
+              #
+              #mxid_mapping: dotreplace
+
+          # In previous versions of synapse, the mapping from SAML attribute to
+          # MXID was always calculated dynamically rather than stored in a
+          # table. For backwards- compatibility, we will look for user_ids
+          # matching such a pattern before creating a new account.
           #
           # This setting controls the SAML attribute which will be used for this
-          # backwards-compatibility lookup. Typically it should be 'uid', but if the
-          # attribute maps are changed, it may be necessary to change it.
+          # backwards-compatibility lookup. Typically it should be 'uid', but if
+          # the attribute maps are changed, it may be necessary to change it.
           #
           # The default is 'uid'.
           #
@@ -241,23 +327,3 @@ class SAML2Config(Config):
         """ % {
             "config_dir_path": config_dir_path
         }
-
-
-DOT_REPLACE_PATTERN = re.compile(
-    ("[^%s]" % (re.escape("".join(mxid_localpart_allowed_characters)),))
-)
-
-
-def dot_replace_for_mxid(username: str) -> str:
-    username = username.lower()
-    username = DOT_REPLACE_PATTERN.sub(".", username)
-
-    # regular mxids aren't allowed to start with an underscore either
-    username = re.sub("^_", "", username)
-    return username
-
-
-MXID_MAPPER_MAP = {
-    "hexencode": map_username_to_mxid_localpart,
-    "dotreplace": dot_replace_for_mxid,
-}

synapse/config/server.py

@@ -102,6 +102,12 @@ class ServerConfig(Config):
             "require_auth_for_profile_requests", False
         )
 
+        # Whether to require sharing a room with a user to retrieve their
+        # profile data
+        self.limit_profile_requests_to_users_who_share_rooms = config.get(
+            "limit_profile_requests_to_users_who_share_rooms", False,
+        )
+
         if "restrict_public_rooms_to_local_users" in config and (
             "allow_public_rooms_without_auth" in config
             or "allow_public_rooms_over_federation" in config
@@ -200,7 +206,7 @@ class ServerConfig(Config):
         self.admin_contact = config.get("admin_contact", None)
 
         # FIXME: federation_domain_whitelist needs sytests
-        self.federation_domain_whitelist = None
+        self.federation_domain_whitelist = None  # type: Optional[dict]
         federation_domain_whitelist = config.get("federation_domain_whitelist", None)
 
         if federation_domain_whitelist is not None:
@@ -621,6 +627,13 @@ class ServerConfig(Config):
         #
         #require_auth_for_profile_requests: true
 
+        # Uncomment to require a user to share a room with another user in order
+        # to retrieve their profile information. Only checked on Client-Server
+        # requests. Profile requests from other servers should be checked by the
+        # requesting server. Defaults to 'false'.
+        #
+        #limit_profile_requests_to_users_who_share_rooms: true
+
         # If set to 'true', removes the need for authentication to access the server's
         # public rooms directory through the client API, meaning that anyone can
         # query the room directory. Defaults to 'false'.

synapse/crypto/event_signing.py

@@ -14,7 +14,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-
+import collections.abc
 import hashlib
 import logging
 
@@ -40,8 +40,11 @@ def check_event_content_hash(event, hash_algorithm=hashlib.sha256):
     # some malformed events lack a 'hashes'. Protect against it being missing
     # or a weird type by basically treating it the same as an unhashed event.
     hashes = event.get("hashes")
-    if not isinstance(hashes, dict):
-        raise SynapseError(400, "Malformed 'hashes'", Codes.UNAUTHORIZED)
+    # nb it might be a frozendict or a dict
+    if not isinstance(hashes, collections.abc.Mapping):
+        raise SynapseError(
+            400, "Malformed 'hashes': %s" % (type(hashes),), Codes.UNAUTHORIZED
+        )
 
     if name not in hashes:
         raise SynapseError(

synapse/crypto/keyring.py

@@ -511,17 +511,18 @@ class BaseV2KeyFetcher(object):
         server_name = response_json["server_name"]
         verified = False
         for key_id in response_json["signatures"].get(server_name, {}):
-            # each of the keys used for the signature must be present in the response
-            # json.
             key = verify_keys.get(key_id)
             if not key:
-                raise KeyLookupError(
-                    "Key response is signed by key id %s:%s but that key is not "
-                    "present in the response" % (server_name, key_id)
-                )
+                # the key may not be present in verify_keys if:
+                #  * we got the key from the notary server, and:
+                #  * the key belongs to the notary server, and:
+                #  * the notary server is using a different key to sign notary
+                #    responses.
+                continue
 
             verify_signed_json(response_json, server_name, key.verify_key)
             verified = True
+            break
 
         if not verified:
             raise KeyLookupError(

synapse/event_auth.py

@@ -43,6 +43,8 @@ def check(room_version, event, auth_events, do_sig_check=True, do_size_check=Tru
     Returns:
          if the auth checks pass.
     """
+    assert isinstance(auth_events, dict)
+
     if do_size_check:
         _check_size_limits(event)
 
@@ -87,12 +89,6 @@ def check(room_version, event, auth_events, do_sig_check=True, do_size_check=Tru
             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)
-        logger.warning("Trusting event: %s", event.event_id)
-        return
-
     if event.type == EventTypes.Create:
         sender_domain = get_domain_from_id(event.sender)
         room_id_domain = get_domain_from_id(event.room_id)

synapse/events/snapshot.py

@@ -149,7 +149,7 @@ class EventContext:
         # the prev_state_ids, so if we're a state event we include the event
         # id that we replaced in the state.
         if event.is_state():
-            prev_state_ids = yield self.get_prev_state_ids(store)
+            prev_state_ids = yield self.get_prev_state_ids()
             prev_state_id = prev_state_ids.get((event.type, event.state_key))
         else:
             prev_state_id = None
@@ -167,12 +167,13 @@ class EventContext:
         }
 
     @staticmethod
-    def deserialize(store, input):
+    def deserialize(storage, input):
         """Converts a dict that was produced by `serialize` back into a
         EventContext.
 
         Args:
-            store (DataStore): Used to convert AS ID to AS object
+            storage (Storage): Used to convert AS ID to AS object and fetch
+                state.
             input (dict): A dict produced by `serialize`
 
         Returns:
@@ -181,6 +182,7 @@ class EventContext:
         context = _AsyncEventContextImpl(
             # We use the state_group and prev_state_id stuff to pull the
             # current_state_ids out of the DB and construct prev_state_ids.
+            storage=storage,
             prev_state_id=input["prev_state_id"],
             event_type=input["event_type"],
             event_state_key=input["event_state_key"],
@@ -193,7 +195,7 @@ class EventContext:
 
         app_service_id = input["app_service_id"]
         if app_service_id:
-            context.app_service = store.get_app_service_by_id(app_service_id)
+            context.app_service = storage.main.get_app_service_by_id(app_service_id)
 
         return context
 
@@ -216,7 +218,7 @@ class EventContext:
         return self._state_group
 
     @defer.inlineCallbacks
-    def get_current_state_ids(self, store):
+    def get_current_state_ids(self):
         """
         Gets the room state map, including this event - ie, the state in ``state_group``
 
@@ -234,11 +236,11 @@ class EventContext:
         if self.rejected:
             raise RuntimeError("Attempt to access state_ids of rejected event")
 
-        yield self._ensure_fetched(store)
+        yield self._ensure_fetched()
         return self._current_state_ids
 
     @defer.inlineCallbacks
-    def get_prev_state_ids(self, store):
+    def get_prev_state_ids(self):
         """
         Gets the room state map, excluding this event.
 
@@ -250,7 +252,7 @@ class EventContext:
                 Maps a (type, state_key) to the event ID of the state event matching
                 this tuple.
         """
-        yield self._ensure_fetched(store)
+        yield self._ensure_fetched()
         return self._prev_state_ids
 
     def get_cached_current_state_ids(self):
@@ -270,7 +272,7 @@ class EventContext:
 
         return self._current_state_ids
 
-    def _ensure_fetched(self, store):
+    def _ensure_fetched(self):
         return defer.succeed(None)
 
 
@@ -282,6 +284,8 @@ class _AsyncEventContextImpl(EventContext):
 
     Attributes:
 
+        _storage (Storage)
+
         _fetching_state_deferred (Deferred|None): Resolves when *_state_ids have
             been calculated. None if we haven't started calculating yet
 
@@ -295,28 +299,30 @@ class _AsyncEventContextImpl(EventContext):
             that was replaced.
     """
 
+    # This needs to have a default as we're inheriting
+    _storage = attr.ib(default=None)
     _prev_state_id = attr.ib(default=None)
     _event_type = attr.ib(default=None)
     _event_state_key = attr.ib(default=None)
     _fetching_state_deferred = attr.ib(default=None)
 
-    def _ensure_fetched(self, store):
+    def _ensure_fetched(self):
         if not self._fetching_state_deferred:
-            self._fetching_state_deferred = run_in_background(
-                self._fill_out_state, store
-            )
+            self._fetching_state_deferred = run_in_background(self._fill_out_state)
 
         return make_deferred_yieldable(self._fetching_state_deferred)
 
     @defer.inlineCallbacks
-    def _fill_out_state(self, store):
+    def _fill_out_state(self):
         """Called to populate the _current_state_ids and _prev_state_ids
         attributes by loading from the database.
         """
         if self.state_group is None:
             return
 
-        self._current_state_ids = yield store.get_state_ids_for_group(self.state_group)
+        self._current_state_ids = yield self._storage.state.get_state_ids_for_group(
+            self.state_group
+        )
         if self._prev_state_id and self._event_state_key is not None:
             self._prev_state_ids = dict(self._current_state_ids)
 

synapse/events/third_party_rules.py

@@ -53,7 +53,7 @@ class ThirdPartyEventRules(object):
         if self.third_party_rules is None:
             return True
 
-        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        prev_state_ids = yield context.get_prev_state_ids()
 
         # Retrieve the state events from the database.
         state_events = {}

synapse/federation/federation_client.py

@@ -526,13 +526,7 @@ class FederationClient(FederationBase):
 
         @defer.inlineCallbacks
         def send_request(destination):
-            time_now = self._clock.time_msec()
-            _, content = yield self.transport_layer.send_join(
-                destination=destination,
-                room_id=pdu.room_id,
-                event_id=pdu.event_id,
-                content=pdu.get_pdu_json(time_now),
-            )
+            content = yield self._do_send_join(destination, pdu)
 
             logger.debug("Got content: %s", content)
 
@@ -599,6 +593,44 @@ class FederationClient(FederationBase):
 
         return self._try_destination_list("send_join", destinations, send_request)
 
+    @defer.inlineCallbacks
+    def _do_send_join(self, destination, pdu):
+        time_now = self._clock.time_msec()
+
+        try:
+            content = yield self.transport_layer.send_join_v2(
+                destination=destination,
+                room_id=pdu.room_id,
+                event_id=pdu.event_id,
+                content=pdu.get_pdu_json(time_now),
+            )
+
+            return content
+        except HttpResponseException as e:
+            if e.code in [400, 404]:
+                err = e.to_synapse_error()
+
+                # If we receive an error response that isn't a generic error, or an
+                # unrecognised endpoint error, we  assume that the remote understands
+                # the v2 invite API and this is a legitimate error.
+                if err.errcode not in [Codes.UNKNOWN, Codes.UNRECOGNIZED]:
+                    raise err
+            else:
+                raise e.to_synapse_error()
+
+        logger.debug("Couldn't send_join with the v2 API, falling back to the v1 API")
+
+        resp = yield self.transport_layer.send_join_v1(
+            destination=destination,
+            room_id=pdu.room_id,
+            event_id=pdu.event_id,
+            content=pdu.get_pdu_json(time_now),
+        )
+
+        # We expect the v1 API to respond with [200, content], so we only return the
+        # content.
+        return resp[1]
+
     @defer.inlineCallbacks
     def send_invite(self, destination, room_id, event_id, pdu):
         room_version = yield self.store.get_room_version(room_id)
@@ -708,18 +740,50 @@ class FederationClient(FederationBase):
 
         @defer.inlineCallbacks
         def send_request(destination):
-            time_now = self._clock.time_msec()
-            _, content = yield self.transport_layer.send_leave(
+            content = yield self._do_send_leave(destination, pdu)
+
+            logger.debug("Got content: %s", content)
+            return None
+
+        return self._try_destination_list("send_leave", destinations, send_request)
+
+    @defer.inlineCallbacks
+    def _do_send_leave(self, destination, pdu):
+        time_now = self._clock.time_msec()
+
+        try:
+            content = yield self.transport_layer.send_leave_v2(
                 destination=destination,
                 room_id=pdu.room_id,
                 event_id=pdu.event_id,
                 content=pdu.get_pdu_json(time_now),
             )
 
-            logger.debug("Got content: %s", content)
-            return None
+            return content
+        except HttpResponseException as e:
+            if e.code in [400, 404]:
+                err = e.to_synapse_error()
 
-        return self._try_destination_list("send_leave", destinations, send_request)
+                # If we receive an error response that isn't a generic error, or an
+                # unrecognised endpoint error, we  assume that the remote understands
+                # the v2 invite API and this is a legitimate error.
+                if err.errcode not in [Codes.UNKNOWN, Codes.UNRECOGNIZED]:
+                    raise err
+            else:
+                raise e.to_synapse_error()
+
+        logger.debug("Couldn't send_leave with the v2 API, falling back to the v1 API")
+
+        resp = yield self.transport_layer.send_leave_v1(
+            destination=destination,
+            room_id=pdu.room_id,
+            event_id=pdu.event_id,
+            content=pdu.get_pdu_json(time_now),
+        )
+
+        # We expect the v1 API to respond with [200, content], so we only return the
+        # content.
+        return resp[1]
 
     def get_public_rooms(
         self,

synapse/federation/federation_server.py

@@ -384,15 +384,10 @@ class FederationServer(FederationBase):
 
         res_pdus = await self.handler.on_send_join_request(origin, pdu)
         time_now = self._clock.time_msec()
-        return (
-            200,
-            {
-                "state": [p.get_pdu_json(time_now) for p in res_pdus["state"]],
-                "auth_chain": [
-                    p.get_pdu_json(time_now) for p in res_pdus["auth_chain"]
-                ],
-            },
-        )
+        return {
+            "state": [p.get_pdu_json(time_now) for p in res_pdus["state"]],
+            "auth_chain": [p.get_pdu_json(time_now) for p in res_pdus["auth_chain"]],
+        }
 
     async def on_make_leave_request(self, origin, room_id, user_id):
         origin_host, _ = parse_server_name(origin)
@@ -419,7 +414,7 @@ class FederationServer(FederationBase):
         pdu = await self._check_sigs_and_hash(room_version, pdu)
 
         await self.handler.on_send_leave_request(origin, pdu)
-        return 200, {}
+        return {}
 
     async def on_event_auth(self, origin, room_id, event_id):
         with (await self._server_linearizer.queue((origin, room_id))):

synapse/federation/transport/client.py

@@ -243,7 +243,7 @@ class TransportLayerClient(object):
 
     @defer.inlineCallbacks
     @log_function
-    def send_join(self, destination, room_id, event_id, content):
+    def send_join_v1(self, destination, room_id, event_id, content):
         path = _create_v1_path("/send_join/%s/%s", room_id, event_id)
 
         response = yield self.client.put_json(
@@ -254,7 +254,18 @@ class TransportLayerClient(object):
 
     @defer.inlineCallbacks
     @log_function
-    def send_leave(self, destination, room_id, event_id, content):
+    def send_join_v2(self, destination, room_id, event_id, content):
+        path = _create_v2_path("/send_join/%s/%s", room_id, event_id)
+
+        response = yield self.client.put_json(
+            destination=destination, path=path, data=content
+        )
+
+        return response
+
+    @defer.inlineCallbacks
+    @log_function
+    def send_leave_v1(self, destination, room_id, event_id, content):
         path = _create_v1_path("/send_leave/%s/%s", room_id, event_id)
 
         response = yield self.client.put_json(
@@ -270,6 +281,24 @@ class TransportLayerClient(object):
 
         return response
 
+    @defer.inlineCallbacks
+    @log_function
+    def send_leave_v2(self, destination, room_id, event_id, content):
+        path = _create_v2_path("/send_leave/%s/%s", room_id, event_id)
+
+        response = yield self.client.put_json(
+            destination=destination,
+            path=path,
+            data=content,
+            # we want to do our best to send this through. The problem is
+            # that if it fails, we won't retry it later, so if the remote
+            # server was just having a momentary blip, the room will be out of
+            # sync.
+            ignore_backoff=True,
+        )
+
+        return response
+
     @defer.inlineCallbacks
     @log_function
     def send_invite_v1(self, destination, room_id, event_id, content):

synapse/federation/transport/server.py

@@ -506,9 +506,19 @@ class FederationMakeLeaveServlet(BaseFederationServlet):
         return 200, content
 
 
-class FederationSendLeaveServlet(BaseFederationServlet):
+class FederationV1SendLeaveServlet(BaseFederationServlet):
     PATH = "/send_leave/(?P<room_id>[^/]*)/(?P<event_id>[^/]*)"
 
+    async def on_PUT(self, origin, content, query, room_id, event_id):
+        content = await self.handler.on_send_leave_request(origin, content, room_id)
+        return 200, (200, content)
+
+
+class FederationV2SendLeaveServlet(BaseFederationServlet):
+    PATH = "/send_leave/(?P<room_id>[^/]*)/(?P<event_id>[^/]*)"
+
+    PREFIX = FEDERATION_V2_PREFIX
+
     async def on_PUT(self, origin, content, query, room_id, event_id):
         content = await self.handler.on_send_leave_request(origin, content, room_id)
         return 200, content
@@ -521,9 +531,21 @@ class FederationEventAuthServlet(BaseFederationServlet):
         return await self.handler.on_event_auth(origin, context, event_id)
 
 
-class FederationSendJoinServlet(BaseFederationServlet):
+class FederationV1SendJoinServlet(BaseFederationServlet):
     PATH = "/send_join/(?P<context>[^/]*)/(?P<event_id>[^/]*)"
 
+    async def on_PUT(self, origin, content, query, context, event_id):
+        # TODO(paul): assert that context/event_id parsed from path actually
+        #   match those given in content
+        content = await self.handler.on_send_join_request(origin, content, context)
+        return 200, (200, content)
+
+
+class FederationV2SendJoinServlet(BaseFederationServlet):
+    PATH = "/send_join/(?P<context>[^/]*)/(?P<event_id>[^/]*)"
+
+    PREFIX = FEDERATION_V2_PREFIX
+
     async def on_PUT(self, origin, content, query, context, event_id):
         # TODO(paul): assert that context/event_id parsed from path actually
         #   match those given in content
@@ -1367,8 +1389,10 @@ FEDERATION_SERVLET_CLASSES = (
     FederationMakeJoinServlet,
     FederationMakeLeaveServlet,
     FederationEventServlet,
-    FederationSendJoinServlet,
-    FederationSendLeaveServlet,
+    FederationV1SendJoinServlet,
+    FederationV2SendJoinServlet,
+    FederationV1SendLeaveServlet,
+    FederationV2SendLeaveServlet,
     FederationV1InviteServlet,
     FederationV2InviteServlet,
     FederationQueryAuthServlet,

synapse/groups/groups_server.py

@@ -773,6 +773,11 @@ class GroupsServerHandler(object):
         if not self.hs.is_mine_id(user_id):
             yield self.store.maybe_delete_remote_profile_cache(user_id)
 
+        # Delete group if the last user has left
+        users = yield self.store.get_users_in_group(group_id, include_private=True)
+        if not users:
+            yield self.store.delete_group(group_id)
+
         return {}
 
     @defer.inlineCallbacks

synapse/handlers/_base.py

@@ -134,7 +134,7 @@ class BaseHandler(object):
             guest_access = event.content.get("guest_access", "forbidden")
             if guest_access != "can_join":
                 if context:
-                    current_state_ids = yield context.get_current_state_ids(self.store)
+                    current_state_ids = yield context.get_current_state_ids()
                     current_state = yield self.store.get_events(
                         list(current_state_ids.values())
                     )

synapse/handlers/account_data.py

@@ -13,8 +13,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from twisted.internet import defer
-
 
 class AccountDataEventSource(object):
     def __init__(self, hs):
@@ -23,15 +21,14 @@ class AccountDataEventSource(object):
     def get_current_key(self, direction="f"):
         return self.store.get_max_account_data_stream_id()
 
-    @defer.inlineCallbacks
-    def get_new_events(self, user, from_key, **kwargs):
+    async def get_new_events(self, user, from_key, **kwargs):
         user_id = user.to_string()
         last_stream_id = from_key
 
-        current_stream_id = yield self.store.get_max_account_data_stream_id()
+        current_stream_id = self.store.get_max_account_data_stream_id()
 
         results = []
-        tags = yield self.store.get_updated_tags(user_id, last_stream_id)
+        tags = await self.store.get_updated_tags(user_id, last_stream_id)
 
         for room_id, room_tags in tags.items():
             results.append(
@@ -41,7 +38,7 @@ class AccountDataEventSource(object):
         (
             account_data,
             room_account_data,
-        ) = yield self.store.get_updated_account_data_for_user(user_id, last_stream_id)
+        ) = await self.store.get_updated_account_data_for_user(user_id, last_stream_id)
 
         for account_data_type, content in account_data.items():
             results.append({"type": account_data_type, "content": content})
@@ -53,7 +50,3 @@ class AccountDataEventSource(object):
                 )
 
         return results, current_stream_id
-
-    @defer.inlineCallbacks
-    def get_pagination_rows(self, user, config, key):
-        return [], config.to_id

synapse/handlers/account_validity.py

@@ -18,8 +18,7 @@ import email.utils
 import logging
 from email.mime.multipart import MIMEMultipart
 from email.mime.text import MIMEText
-
-from twisted.internet import defer
+from typing import List
 
 from synapse.api.errors import StoreError
 from synapse.logging.context import make_deferred_yieldable
@@ -78,42 +77,39 @@ class AccountValidityHandler(object):
                 # run as a background process to make sure that the database transactions
                 # have a logcontext to report to
                 return run_as_background_process(
-                    "send_renewals", self.send_renewal_emails
+                    "send_renewals", self._send_renewal_emails
                 )
 
             self.clock.looping_call(send_emails, 30 * 60 * 1000)
 
-    @defer.inlineCallbacks
-    def send_renewal_emails(self):
+    async def _send_renewal_emails(self):
         """Gets the list of users whose account is expiring in the amount of time
         configured in the ``renew_at`` parameter from the ``account_validity``
         configuration, and sends renewal emails to all of these users as long as they
         have an email 3PID attached to their account.
         """
-        expiring_users = yield self.store.get_users_expiring_soon()
+        expiring_users = await self.store.get_users_expiring_soon()
 
         if expiring_users:
             for user in expiring_users:
-                yield self._send_renewal_email(
+                await self._send_renewal_email(
                     user_id=user["user_id"], expiration_ts=user["expiration_ts_ms"]
                 )
 
-    @defer.inlineCallbacks
-    def send_renewal_email_to_user(self, user_id):
-        expiration_ts = yield self.store.get_expiration_ts_for_user(user_id)
-        yield self._send_renewal_email(user_id, expiration_ts)
+    async def send_renewal_email_to_user(self, user_id: str):
+        expiration_ts = await self.store.get_expiration_ts_for_user(user_id)
+        await self._send_renewal_email(user_id, expiration_ts)
 
-    @defer.inlineCallbacks
-    def _send_renewal_email(self, user_id, expiration_ts):
+    async def _send_renewal_email(self, user_id: str, expiration_ts: int):
         """Sends out a renewal email to every email address attached to the given user
         with a unique link allowing them to renew their account.
 
         Args:
-            user_id (str): ID of the user to send email(s) to.
-            expiration_ts (int): Timestamp in milliseconds for the expiration date of
+            user_id: ID of the user to send email(s) to.
+            expiration_ts: Timestamp in milliseconds for the expiration date of
                 this user's account (used in the email templates).
         """
-        addresses = yield self._get_email_addresses_for_user(user_id)
+        addresses = await self._get_email_addresses_for_user(user_id)
 
         # Stop right here if the user doesn't have at least one email address.
         # In this case, they will have to ask their server admin to renew their
@@ -125,7 +121,7 @@ class AccountValidityHandler(object):
             return
 
         try:
-            user_display_name = yield self.store.get_profile_displayname(
+            user_display_name = await self.store.get_profile_displayname(
                 UserID.from_string(user_id).localpart
             )
             if user_display_name is None:
@@ -133,7 +129,7 @@ class AccountValidityHandler(object):
         except StoreError:
             user_display_name = user_id
 
-        renewal_token = yield self._get_renewal_token(user_id)
+        renewal_token = await self._get_renewal_token(user_id)
         url = "%s_matrix/client/unstable/account_validity/renew?token=%s" % (
             self.hs.config.public_baseurl,
             renewal_token,
@@ -165,7 +161,7 @@ class AccountValidityHandler(object):
 
             logger.info("Sending renewal email to %s", address)
 
-            yield make_deferred_yieldable(
+            await make_deferred_yieldable(
                 self.sendmail(
                     self.hs.config.email_smtp_host,
                     self._raw_from,
@@ -180,19 +176,18 @@ class AccountValidityHandler(object):
                 )
             )
 
-        yield self.store.set_renewal_mail_status(user_id=user_id, email_sent=True)
+        await self.store.set_renewal_mail_status(user_id=user_id, email_sent=True)
 
-    @defer.inlineCallbacks
-    def _get_email_addresses_for_user(self, user_id):
+    async def _get_email_addresses_for_user(self, user_id: str) -> List[str]:
         """Retrieve the list of email addresses attached to a user's account.
 
         Args:
-            user_id (str): ID of the user to lookup email addresses for.
+            user_id: ID of the user to lookup email addresses for.
 
         Returns:
-            defer.Deferred[list[str]]: Email addresses for this account.
+            Email addresses for this account.
         """
-        threepids = yield self.store.user_get_threepids(user_id)
+        threepids = await self.store.user_get_threepids(user_id)
 
         addresses = []
         for threepid in threepids:
@@ -201,16 +196,15 @@ class AccountValidityHandler(object):
 
         return addresses
 
-    @defer.inlineCallbacks
-    def _get_renewal_token(self, user_id):
+    async def _get_renewal_token(self, user_id: str) -> str:
         """Generates a 32-byte long random string that will be inserted into the
         user's renewal email's unique link, then saves it into the database.
 
         Args:
-            user_id (str): ID of the user to generate a string for.
+            user_id: ID of the user to generate a string for.
 
         Returns:
-            defer.Deferred[str]: The generated string.
+            The generated string.
 
         Raises:
             StoreError(500): Couldn't generate a unique string after 5 attempts.
@@ -219,52 +213,52 @@ class AccountValidityHandler(object):
         while attempts < 5:
             try:
                 renewal_token = stringutils.random_string(32)
-                yield self.store.set_renewal_token_for_user(user_id, renewal_token)
+                await self.store.set_renewal_token_for_user(user_id, renewal_token)
                 return renewal_token
             except StoreError:
                 attempts += 1
         raise StoreError(500, "Couldn't generate a unique string as refresh string.")
 
-    @defer.inlineCallbacks
-    def renew_account(self, renewal_token):
+    async def renew_account(self, renewal_token: str) -> bool:
         """Renews the account attached to a given renewal token by pushing back the
         expiration date by the current validity period in the server's configuration.
 
         Args:
-            renewal_token (str): Token sent with the renewal request.
+            renewal_token: Token sent with the renewal request.
         Returns:
-            bool: Whether the provided token is valid.
+            Whether the provided token is valid.
         """
         try:
-            user_id = yield self.store.get_user_from_renewal_token(renewal_token)
+            user_id = await self.store.get_user_from_renewal_token(renewal_token)
         except StoreError:
-            defer.returnValue(False)
+            return False
 
         logger.debug("Renewing an account for user %s", user_id)
-        yield self.renew_account_for_user(user_id)
+        await self.renew_account_for_user(user_id)
 
-        defer.returnValue(True)
+        return True
 
-    @defer.inlineCallbacks
-    def renew_account_for_user(self, user_id, expiration_ts=None, email_sent=False):
+    async def renew_account_for_user(
+        self, user_id: str, expiration_ts: int = None, email_sent: bool = False
+    ) -> int:
         """Renews the account attached to a given user by pushing back the
         expiration date by the current validity period in the server's
         configuration.
 
         Args:
-            renewal_token (str): Token sent with the renewal request.
-            expiration_ts (int): New expiration date. Defaults to now + validity period.
-            email_sent (bool): Whether an email has been sent for this validity period.
+            renewal_token: Token sent with the renewal request.
+            expiration_ts: New expiration date. Defaults to now + validity period.
+            email_sen: Whether an email has been sent for this validity period.
                 Defaults to False.
 
         Returns:
-            defer.Deferred[int]: New expiration date for this account, as a timestamp
-                in milliseconds since epoch.
+            New expiration date for this account, as a timestamp in
+            milliseconds since epoch.
         """
         if expiration_ts is None:
             expiration_ts = self.clock.time_msec() + self._account_validity.period
 
-        yield self.store.set_account_validity_for_user(
+        await self.store.set_account_validity_for_user(
             user_id=user_id, expiration_ts=expiration_ts, email_sent=email_sent
         )
 

synapse/handlers/admin.py

@@ -15,8 +15,6 @@
 
 import logging
 
-from twisted.internet import defer
-
 from synapse.api.constants import Membership
 from synapse.types import RoomStreamToken
 from synapse.visibility import filter_events_for_client
@@ -33,11 +31,10 @@ class AdminHandler(BaseHandler):
         self.storage = hs.get_storage()
         self.state_store = self.storage.state
 
-    @defer.inlineCallbacks
-    def get_whois(self, user):
+    async def get_whois(self, user):
         connections = []
 
-        sessions = yield self.store.get_user_ip_and_agents(user)
+        sessions = await self.store.get_user_ip_and_agents(user)
         for session in sessions:
             connections.append(
                 {
@@ -54,20 +51,18 @@ class AdminHandler(BaseHandler):
 
         return ret
 
-    @defer.inlineCallbacks
-    def get_users(self):
+    async def get_users(self):
         """Function to retrieve a list of users in users table.
 
         Args:
         Returns:
             defer.Deferred: resolves to list[dict[str, Any]]
         """
-        ret = yield self.store.get_users()
+        ret = await self.store.get_users()
 
         return ret
 
-    @defer.inlineCallbacks
-    def get_users_paginate(self, start, limit, name, guests, deactivated):
+    async def get_users_paginate(self, start, limit, name, guests, deactivated):
         """Function to retrieve a paginated list of users from
         users list. This will return a json list of users.
 
@@ -80,14 +75,13 @@ class AdminHandler(BaseHandler):
         Returns:
             defer.Deferred: resolves to json list[dict[str, Any]]
         """
-        ret = yield self.store.get_users_paginate(
+        ret = await self.store.get_users_paginate(
             start, limit, name, guests, deactivated
         )
 
         return ret
 
-    @defer.inlineCallbacks
-    def search_users(self, term):
+    async def search_users(self, term):
         """Function to search users list for one or more users with
         the matched term.
 
@@ -96,7 +90,7 @@ class AdminHandler(BaseHandler):
         Returns:
             defer.Deferred: resolves to list[dict[str, Any]]
         """
-        ret = yield self.store.search_users(term)
+        ret = await self.store.search_users(term)
 
         return ret
 
@@ -119,8 +113,7 @@ class AdminHandler(BaseHandler):
         """
         return self.store.set_server_admin(user, admin)
 
-    @defer.inlineCallbacks
-    def export_user_data(self, user_id, writer):
+    async def export_user_data(self, user_id, writer):
         """Write all data we have on the user to the given writer.
 
         Args:
@@ -132,7 +125,7 @@ class AdminHandler(BaseHandler):
             The returned value is that returned by `writer.finished()`.
         """
         # Get all rooms the user is in or has been in
-        rooms = yield self.store.get_rooms_for_user_where_membership_is(
+        rooms = await self.store.get_rooms_for_user_where_membership_is(
             user_id,
             membership_list=(
                 Membership.JOIN,
@@ -145,7 +138,7 @@ class AdminHandler(BaseHandler):
         # We only try and fetch events for rooms the user has been in. If
         # they've been e.g. invited to a room without joining then we handle
         # those seperately.
-        rooms_user_has_been_in = yield self.store.get_rooms_user_has_been_in(user_id)
+        rooms_user_has_been_in = await self.store.get_rooms_user_has_been_in(user_id)
 
         for index, room in enumerate(rooms):
             room_id = room.room_id
@@ -154,7 +147,7 @@ class AdminHandler(BaseHandler):
                 "[%s] Handling room %s, %d/%d", user_id, room_id, index + 1, len(rooms)
             )
 
-            forgotten = yield self.store.did_forget(user_id, room_id)
+            forgotten = await self.store.did_forget(user_id, room_id)
             if forgotten:
                 logger.info("[%s] User forgot room %d, ignoring", user_id, room_id)
                 continue
@@ -166,7 +159,7 @@ class AdminHandler(BaseHandler):
 
                 if room.membership == Membership.INVITE:
                     event_id = room.event_id
-                    invite = yield self.store.get_event(event_id, allow_none=True)
+                    invite = await self.store.get_event(event_id, allow_none=True)
                     if invite:
                         invited_state = invite.unsigned["invite_room_state"]
                         writer.write_invite(room_id, invite, invited_state)
@@ -177,7 +170,7 @@ class AdminHandler(BaseHandler):
             # were joined. We estimate that point by looking at the
             # stream_ordering of the last membership if it wasn't a join.
             if room.membership == Membership.JOIN:
-                stream_ordering = yield self.store.get_room_max_stream_ordering()
+                stream_ordering = self.store.get_room_max_stream_ordering()
             else:
                 stream_ordering = room.stream_ordering
 
@@ -203,7 +196,7 @@ class AdminHandler(BaseHandler):
             # events that we have and then filtering, this isn't the most
             # efficient method perhaps but it does guarantee we get everything.
             while True:
-                events, _ = yield self.store.paginate_room_events(
+                events, _ = await self.store.paginate_room_events(
                     room_id, from_key, to_key, limit=100, direction="f"
                 )
                 if not events:
@@ -211,7 +204,7 @@ class AdminHandler(BaseHandler):
 
                 from_key = events[-1].internal_metadata.after
 
-                events = yield filter_events_for_client(self.storage, user_id, events)
+                events = await filter_events_for_client(self.storage, user_id, events)
 
                 writer.write_events(room_id, events)
 
@@ -247,7 +240,7 @@ class AdminHandler(BaseHandler):
             for event_id in extremities:
                 if not event_to_unseen_prevs[event_id]:
                     continue
-                state = yield self.state_store.get_state_for_event(event_id)
+                state = await self.state_store.get_state_for_event(event_id)
                 writer.write_state(room_id, event_id, state)
 
         return writer.finished()

synapse/handlers/deactivate_account.py

@@ -15,8 +15,6 @@
 # limitations under the License.
 import logging
 
-from twisted.internet import defer
-
 from synapse.api.errors import SynapseError
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.types import UserID, create_requester
@@ -46,8 +44,7 @@ class DeactivateAccountHandler(BaseHandler):
 
         self._account_validity_enabled = hs.config.account_validity.enabled
 
-    @defer.inlineCallbacks
-    def deactivate_account(self, user_id, erase_data, id_server=None):
+    async def deactivate_account(self, user_id, erase_data, id_server=None):
         """Deactivate a user's account
 
         Args:
@@ -74,11 +71,11 @@ class DeactivateAccountHandler(BaseHandler):
         identity_server_supports_unbinding = True
 
         # Retrieve the 3PIDs this user has bound to an identity server
-        threepids = yield self.store.user_get_bound_threepids(user_id)
+        threepids = await self.store.user_get_bound_threepids(user_id)
 
         for threepid in threepids:
             try:
-                result = yield self._identity_handler.try_unbind_threepid(
+                result = await self._identity_handler.try_unbind_threepid(
                     user_id,
                     {
                         "medium": threepid["medium"],
@@ -91,33 +88,33 @@ class DeactivateAccountHandler(BaseHandler):
                 # Do we want this to be a fatal error or should we carry on?
                 logger.exception("Failed to remove threepid from ID server")
                 raise SynapseError(400, "Failed to remove threepid from ID server")
-            yield self.store.user_delete_threepid(
+            await self.store.user_delete_threepid(
                 user_id, threepid["medium"], threepid["address"]
             )
 
         # Remove all 3PIDs this user has bound to the homeserver
-        yield self.store.user_delete_threepids(user_id)
+        await self.store.user_delete_threepids(user_id)
 
         # delete any devices belonging to the user, which will also
         # delete corresponding access tokens.
-        yield self._device_handler.delete_all_devices_for_user(user_id)
+        await self._device_handler.delete_all_devices_for_user(user_id)
         # then delete any remaining access tokens which weren't associated with
         # a device.
-        yield self._auth_handler.delete_access_tokens_for_user(user_id)
+        await self._auth_handler.delete_access_tokens_for_user(user_id)
 
-        yield self.store.user_set_password_hash(user_id, None)
+        await self.store.user_set_password_hash(user_id, None)
 
         # Add the user to a table of users pending deactivation (ie.
         # removal from all the rooms they're a member of)
-        yield self.store.add_user_pending_deactivation(user_id)
+        await self.store.add_user_pending_deactivation(user_id)
 
         # delete from user directory
-        yield self.user_directory_handler.handle_user_deactivated(user_id)
+        await self.user_directory_handler.handle_user_deactivated(user_id)
 
         # Mark the user as erased, if they asked for that
         if erase_data:
             logger.info("Marking %s as erased", user_id)
-            yield self.store.mark_user_erased(user_id)
+            await self.store.mark_user_erased(user_id)
 
         # Now start the process that goes through that list and
         # parts users from rooms (if it isn't already running)
@@ -125,30 +122,29 @@ class DeactivateAccountHandler(BaseHandler):
 
         # Reject all pending invites for the user, so that the user doesn't show up in the
         # "invited" section of rooms' members list.
-        yield self._reject_pending_invites_for_user(user_id)
+        await self._reject_pending_invites_for_user(user_id)
 
         # Remove all information on the user from the account_validity table.
         if self._account_validity_enabled:
-            yield self.store.delete_account_validity_for_user(user_id)
+            await self.store.delete_account_validity_for_user(user_id)
 
         # Mark the user as deactivated.
-        yield self.store.set_user_deactivated_status(user_id, True)
+        await self.store.set_user_deactivated_status(user_id, True)
 
         return identity_server_supports_unbinding
 
-    @defer.inlineCallbacks
-    def _reject_pending_invites_for_user(self, user_id):
+    async def _reject_pending_invites_for_user(self, user_id):
         """Reject pending invites addressed to a given user ID.
 
         Args:
             user_id (str): The user ID to reject pending invites for.
         """
         user = UserID.from_string(user_id)
-        pending_invites = yield self.store.get_invited_rooms_for_user(user_id)
+        pending_invites = await self.store.get_invited_rooms_for_user(user_id)
 
         for room in pending_invites:
             try:
-                yield self._room_member_handler.update_membership(
+                await self._room_member_handler.update_membership(
                     create_requester(user),
                     user,
                     room.room_id,
@@ -180,8 +176,7 @@ class DeactivateAccountHandler(BaseHandler):
         if not self._user_parter_running:
             run_as_background_process("user_parter_loop", self._user_parter_loop)
 
-    @defer.inlineCallbacks
-    def _user_parter_loop(self):
+    async def _user_parter_loop(self):
         """Loop that parts deactivated users from rooms
 
         Returns:
@@ -191,19 +186,18 @@ class DeactivateAccountHandler(BaseHandler):
         logger.info("Starting user parter")
         try:
             while True:
-                user_id = yield self.store.get_user_pending_deactivation()
+                user_id = await self.store.get_user_pending_deactivation()
                 if user_id is None:
                     break
                 logger.info("User parter parting %r", user_id)
-                yield self._part_user(user_id)
-                yield self.store.del_user_pending_deactivation(user_id)
+                await self._part_user(user_id)
+                await self.store.del_user_pending_deactivation(user_id)
                 logger.info("User parter finished parting %r", user_id)
             logger.info("User parter finished: stopping")
         finally:
             self._user_parter_running = False
 
-    @defer.inlineCallbacks
-    def _part_user(self, user_id):
+    async def _part_user(self, user_id):
         """Causes the given user_id to leave all the rooms they're joined to
 
         Returns:
@@ -211,11 +205,11 @@ class DeactivateAccountHandler(BaseHandler):
         """
         user = UserID.from_string(user_id)
 
-        rooms_for_user = yield self.store.get_rooms_for_user(user_id)
+        rooms_for_user = await self.store.get_rooms_for_user(user_id)
         for room_id in rooms_for_user:
             logger.info("User parter parting %r from %r", user_id, room_id)
             try:
-                yield self._room_member_handler.update_membership(
+                await self._room_member_handler.update_membership(
                     create_requester(user),
                     user,
                     room_id,

synapse/handlers/e2e_keys.py

@@ -264,6 +264,7 @@ class E2eKeysHandler(object):
 
         return ret
 
+    @defer.inlineCallbacks
     def get_cross_signing_keys_from_cache(self, query, from_user_id):
         """Get cross-signing keys for users from the database
 
@@ -283,14 +284,32 @@ class E2eKeysHandler(object):
         self_signing_keys = {}
         user_signing_keys = {}
 
-        # Currently a stub, implementation coming in https://github.com/matrix-org/synapse/pull/6486
-        return defer.succeed(
-            {
-                "master_keys": master_keys,
-                "self_signing_keys": self_signing_keys,
-                "user_signing_keys": user_signing_keys,
-            }
-        )
+        user_ids = list(query)
+
+        keys = yield self.store.get_e2e_cross_signing_keys_bulk(user_ids, from_user_id)
+
+        for user_id, user_info in keys.items():
+            if user_info is None:
+                continue
+            if "master" in user_info:
+                master_keys[user_id] = user_info["master"]
+            if "self_signing" in user_info:
+                self_signing_keys[user_id] = user_info["self_signing"]
+
+        if (
+            from_user_id in keys
+            and keys[from_user_id] is not None
+            and "user_signing" in keys[from_user_id]
+        ):
+            # users can see other users' master and self-signing keys, but can
+            # only see their own user-signing keys
+            user_signing_keys[from_user_id] = keys[from_user_id]["user_signing"]
+
+        return {
+            "master_keys": master_keys,
+            "self_signing_keys": self_signing_keys,
+            "user_signing_keys": user_signing_keys,
+        }
 
     @trace
     @defer.inlineCallbacks

synapse/handlers/federation.py

@@ -19,7 +19,7 @@
 
 import itertools
 import logging
-from typing import Dict, Iterable, Optional, Sequence, Tuple
+from typing import Dict, Iterable, List, Optional, Sequence, Tuple
 
 import six
 from six import iteritems, itervalues
@@ -63,6 +63,7 @@ from synapse.replication.http.federation import (
 )
 from synapse.replication.http.membership import ReplicationUserJoinedLeftRoomRestServlet
 from synapse.state import StateResolutionStore, resolve_events_with_store
+from synapse.storage.data_stores.main.events_worker import EventRedactBehaviour
 from synapse.types import UserID, get_domain_from_id
 from synapse.util.async_helpers import Linearizer, concurrently_execute
 from synapse.util.distributor import user_joined_room
@@ -163,8 +164,7 @@ class FederationHandler(BaseHandler):
 
         self._ephemeral_messages_enabled = hs.config.enable_ephemeral_messages
 
-    @defer.inlineCallbacks
-    def on_receive_pdu(self, origin, pdu, sent_to_us_directly=False):
+    async def on_receive_pdu(self, origin, pdu, sent_to_us_directly=False) -> None:
         """ Process a PDU received via a federation /send/ transaction, or
         via backfill of missing prev_events
 
@@ -174,17 +174,15 @@ class FederationHandler(BaseHandler):
             pdu (FrozenEvent): received PDU
             sent_to_us_directly (bool): True if this event was pushed to us; False if
                 we pulled it as the result of a missing prev_event.
-
-        Returns (Deferred): completes with None
         """
 
         room_id = pdu.room_id
         event_id = pdu.event_id
 
-        logger.info("[%s %s] handling received PDU: %s", room_id, event_id, pdu)
+        logger.info("handling received PDU: %s", pdu)
 
         # We reprocess pdus when we have seen them only as outliers
-        existing = yield self.store.get_event(
+        existing = await self.store.get_event(
             event_id, allow_none=True, allow_rejected=True
         )
 
@@ -228,7 +226,7 @@ class FederationHandler(BaseHandler):
         #
         # Note that if we were never in the room then we would have already
         # dropped the event, since we wouldn't know the room version.
-        is_in_room = yield self.auth.check_host_in_room(room_id, self.server_name)
+        is_in_room = await self.auth.check_host_in_room(room_id, self.server_name)
         if not is_in_room:
             logger.info(
                 "[%s %s] Ignoring PDU from %s as we're not in the room",
@@ -243,20 +241,20 @@ class FederationHandler(BaseHandler):
         # Get missing pdus if necessary.
         if not pdu.internal_metadata.is_outlier():
             # We only backfill backwards to the min depth.
-            min_depth = yield self.get_min_depth_for_context(pdu.room_id)
+            min_depth = await self.get_min_depth_for_context(pdu.room_id)
 
             logger.debug("[%s %s] min_depth: %d", room_id, event_id, min_depth)
 
             prevs = set(pdu.prev_event_ids())
-            seen = yield self.store.have_seen_events(prevs)
+            seen = await self.store.have_seen_events(prevs)
 
-            if min_depth and pdu.depth < min_depth:
+            if min_depth is not None and pdu.depth < min_depth:
                 # This is so that we don't notify the user about this
                 # message, to work around the fact that some events will
                 # reference really really old events we really don't want to
                 # send to the clients.
                 pdu.internal_metadata.outlier = True
-            elif min_depth and pdu.depth > min_depth:
+            elif min_depth is not None and pdu.depth > min_depth:
                 missing_prevs = prevs - seen
                 if sent_to_us_directly and missing_prevs:
                     # If we're missing stuff, ensure we only fetch stuff one
@@ -268,7 +266,7 @@ class FederationHandler(BaseHandler):
                         len(missing_prevs),
                         shortstr(missing_prevs),
                     )
-                    with (yield self._room_pdu_linearizer.queue(pdu.room_id)):
+                    with (await self._room_pdu_linearizer.queue(pdu.room_id)):
                         logger.info(
                             "[%s %s] Acquired room lock to fetch %d missing prev_events",
                             room_id,
@@ -276,13 +274,19 @@ class FederationHandler(BaseHandler):
                             len(missing_prevs),
                         )
 
-                        yield self._get_missing_events_for_pdu(
-                            origin, pdu, prevs, min_depth
-                        )
+                        try:
+                            await self._get_missing_events_for_pdu(
+                                origin, pdu, prevs, min_depth
+                            )
+                        except Exception as e:
+                            raise Exception(
+                                "Error fetching missing prev_events for %s: %s"
+                                % (event_id, e)
+                            )
 
                         # Update the set of things we've seen after trying to
                         # fetch the missing stuff
-                        seen = yield self.store.have_seen_events(prevs)
+                        seen = await self.store.have_seen_events(prevs)
 
                         if not prevs - seen:
                             logger.info(
@@ -290,14 +294,6 @@ class FederationHandler(BaseHandler):
                                 room_id,
                                 event_id,
                             )
-                elif missing_prevs:
-                    logger.info(
-                        "[%s %s] Not recursively fetching %d missing prev_events: %s",
-                        room_id,
-                        event_id,
-                        len(missing_prevs),
-                        shortstr(missing_prevs),
-                    )
 
             if prevs - seen:
                 # We've still not been able to get all of the prev_events for this event.
@@ -342,12 +338,18 @@ class FederationHandler(BaseHandler):
                         affected=pdu.event_id,
                     )
 
+                logger.info(
+                    "Event %s is missing prev_events: calculating state for a "
+                    "backwards extremity",
+                    event_id,
+                )
+
                 # Calculate the state after each of the previous events, and
                 # resolve them to find the correct state at the current event.
                 event_map = {event_id: pdu}
                 try:
                     # Get the state of the events we know about
-                    ours = yield self.state_store.get_state_groups_ids(room_id, seen)
+                    ours = await self.state_store.get_state_groups_ids(room_id, seen)
 
                     # state_maps is a list of mappings from (type, state_key) to event_id
                     state_maps = list(
@@ -361,17 +363,14 @@ class FederationHandler(BaseHandler):
                     # know about
                     for p in prevs - seen:
                         logger.info(
-                            "[%s %s] Requesting state at missing prev_event %s",
-                            room_id,
-                            event_id,
-                            p,
+                            "Requesting state at missing prev_event %s", event_id,
                         )
 
                         with nested_logging_context(p):
                             # note that if any of the missing prevs share missing state or
                             # auth events, the requests to fetch those events are deduped
                             # by the get_pdu_cache in federation_client.
-                            (remote_state, _,) = yield self._get_state_for_room(
+                            (remote_state, _,) = await self._get_state_for_room(
                                 origin, room_id, p, include_event_in_state=True
                             )
 
@@ -383,8 +382,8 @@ class FederationHandler(BaseHandler):
                             for x in remote_state:
                                 event_map[x.event_id] = x
 
-                    room_version = yield self.store.get_room_version(room_id)
-                    state_map = yield resolve_events_with_store(
+                    room_version = await self.store.get_room_version(room_id)
+                    state_map = await resolve_events_with_store(
                         room_id,
                         room_version,
                         state_maps,
@@ -397,10 +396,10 @@ class FederationHandler(BaseHandler):
 
                     # First though we need to fetch all the events that are in
                     # state_map, so we can build up the state below.
-                    evs = yield self.store.get_events(
+                    evs = await self.store.get_events(
                         list(state_map.values()),
                         get_prev_content=False,
-                        check_redacted=False,
+                        redact_behaviour=EventRedactBehaviour.AS_IS,
                     )
                     event_map.update(evs)
 
@@ -420,10 +419,9 @@ class FederationHandler(BaseHandler):
                         affected=event_id,
                     )
 
-        yield self._process_received_pdu(origin, pdu, state=state)
+        await self._process_received_pdu(origin, pdu, state=state)
 
-    @defer.inlineCallbacks
-    def _get_missing_events_for_pdu(self, origin, pdu, prevs, min_depth):
+    async def _get_missing_events_for_pdu(self, origin, pdu, prevs, min_depth):
         """
         Args:
             origin (str): Origin of the pdu. Will be called to get the missing events
@@ -435,12 +433,12 @@ class FederationHandler(BaseHandler):
         room_id = pdu.room_id
         event_id = pdu.event_id
 
-        seen = yield self.store.have_seen_events(prevs)
+        seen = await self.store.have_seen_events(prevs)
 
         if not prevs - seen:
             return
 
-        latest = yield self.store.get_latest_event_ids_in_room(room_id)
+        latest = await self.store.get_latest_event_ids_in_room(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
@@ -504,7 +502,7 @@ class FederationHandler(BaseHandler):
         # All that said: Let's try increasing the timout to 60s and see what happens.
 
         try:
-            missing_events = yield self.federation_client.get_missing_events(
+            missing_events = await self.federation_client.get_missing_events(
                 origin,
                 room_id,
                 earliest_events_ids=list(latest),
@@ -543,7 +541,7 @@ class FederationHandler(BaseHandler):
             )
             with nested_logging_context(ev.event_id):
                 try:
-                    yield self.on_receive_pdu(origin, ev, sent_to_us_directly=False)
+                    await self.on_receive_pdu(origin, ev, sent_to_us_directly=False)
                 except FederationError as e:
                     if e.code == 403:
                         logger.warning(
@@ -555,29 +553,30 @@ class FederationHandler(BaseHandler):
                     else:
                         raise
 
-    @defer.inlineCallbacks
-    @log_function
-    def _get_state_for_room(
-        self, destination, room_id, event_id, include_event_in_state
-    ):
+    async def _get_state_for_room(
+        self,
+        destination: str,
+        room_id: str,
+        event_id: str,
+        include_event_in_state: bool = False,
+    ) -> Tuple[List[EventBase], List[EventBase]]:
         """Requests all of the room state at a given event from a remote homeserver.
 
         Args:
-            destination (str): The remote homeserver to query for the state.
-            room_id (str): The id of the room we're interested in.
-            event_id (str): The id of the event we want the state at.
+            destination: The remote homeserver to query for the state.
+            room_id: The id of the room we're interested in.
+            event_id: The id of the event we want the state at.
             include_event_in_state: if true, the event itself will be included in the
                 returned state event list.
 
         Returns:
-            Deferred[Tuple[List[EventBase], List[EventBase]]]:
-                A list of events in the state, and a list of events in the auth chain
-                for the given event.
+            A list of events in the state, possibly including the event itself, and
+            a list of events in the auth chain for the given event.
         """
         (
             state_event_ids,
             auth_event_ids,
-        ) = yield self.federation_client.get_room_state_ids(
+        ) = await self.federation_client.get_room_state_ids(
             destination, room_id, event_id=event_id
         )
 
@@ -586,15 +585,15 @@ class FederationHandler(BaseHandler):
         if include_event_in_state:
             desired_events.add(event_id)
 
-        event_map = yield self._get_events_from_store_or_dest(
+        event_map = await self._get_events_from_store_or_dest(
             destination, room_id, desired_events
         )
 
         failed_to_fetch = desired_events - event_map.keys()
         if failed_to_fetch:
             logger.warning(
-                "Failed to fetch missing state/auth events for %s: %s",
-                room_id,
+                "Failed to fetch missing state/auth events for %s %s",
+                event_id,
                 failed_to_fetch,
             )
 
@@ -614,15 +613,11 @@ class FederationHandler(BaseHandler):
 
         return remote_state, auth_chain
 
-    @defer.inlineCallbacks
-    def _get_events_from_store_or_dest(self, destination, room_id, event_ids):
+    async def _get_events_from_store_or_dest(
+        self, destination: str, room_id: str, event_ids: Iterable[str]
+    ) -> Dict[str, EventBase]:
         """Fetch events from a remote destination, checking if we already have them.
 
-        Args:
-            destination (str)
-            room_id (str)
-            event_ids (Iterable[str])
-
         Persists any events we don't already have as outliers.
 
         If we fail to fetch any of the events, a warning will be logged, and the event
@@ -630,10 +625,9 @@ class FederationHandler(BaseHandler):
         be in the given room.
 
         Returns:
-            Deferred[dict[str, EventBase]]: A deferred resolving to a map
-            from event_id to event
+            map from event_id to event
         """
-        fetched_events = yield self.store.get_events(event_ids, allow_rejected=True)
+        fetched_events = await self.store.get_events(event_ids, allow_rejected=True)
 
         missing_events = set(event_ids) - fetched_events.keys()
 
@@ -644,14 +638,14 @@ class FederationHandler(BaseHandler):
                 room_id,
             )
 
-            yield self._get_events_and_persist(
+            await self._get_events_and_persist(
                 destination=destination, room_id=room_id, events=missing_events
             )
 
             # we need to make sure we re-load from the database to get the rejected
             # state correct.
             fetched_events.update(
-                (yield self.store.get_events(missing_events, allow_rejected=True))
+                (await self.store.get_events(missing_events, allow_rejected=True))
             )
 
         # check for events which were in the wrong room.
@@ -677,12 +671,14 @@ class FederationHandler(BaseHandler):
                 bad_room_id,
                 room_id,
             )
+
             del fetched_events[bad_event_id]
 
         return fetched_events
 
-    @defer.inlineCallbacks
-    def _process_received_pdu(self, origin, event, state):
+    async def _process_received_pdu(
+        self, origin: str, event: EventBase, state: Optional[Iterable[EventBase]],
+    ):
         """ Called when we have a new pdu. We need to do auth checks and put it
         through the StateHandler.
 
@@ -701,15 +697,15 @@ class FederationHandler(BaseHandler):
         logger.debug("[%s %s] Processing event: %s", room_id, event_id, event)
 
         try:
-            context = yield self._handle_new_event(origin, event, state=state)
+            context = await self._handle_new_event(origin, event, state=state)
         except AuthError as e:
             raise FederationError("ERROR", e.code, e.msg, affected=event.event_id)
 
-        room = yield self.store.get_room(room_id)
+        room = await self.store.get_room(room_id)
 
         if not room:
             try:
-                yield self.store.store_room(
+                await self.store.store_room(
                     room_id=room_id, room_creator_user_id="", is_public=False
                 )
             except StoreError:
@@ -722,11 +718,11 @@ class FederationHandler(BaseHandler):
                 # changing their profile info.
                 newly_joined = True
 
-                prev_state_ids = yield context.get_prev_state_ids(self.store)
+                prev_state_ids = await context.get_prev_state_ids()
 
                 prev_state_id = prev_state_ids.get((event.type, event.state_key))
                 if prev_state_id:
-                    prev_state = yield self.store.get_event(
+                    prev_state = await self.store.get_event(
                         prev_state_id, allow_none=True
                     )
                     if prev_state and prev_state.membership == Membership.JOIN:
@@ -734,11 +730,10 @@ class FederationHandler(BaseHandler):
 
                 if newly_joined:
                     user = UserID.from_string(event.state_key)
-                    yield self.user_joined_room(user, room_id)
+                    await self.user_joined_room(user, room_id)
 
     @log_function
-    @defer.inlineCallbacks
-    def backfill(self, dest, room_id, limit, extremities):
+    async 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. If the other side
@@ -755,7 +750,7 @@ class FederationHandler(BaseHandler):
         if dest == self.server_name:
             raise SynapseError(400, "Can't backfill from self.")
 
-        events = yield self.federation_client.backfill(
+        events = await self.federation_client.backfill(
             dest, room_id, limit=limit, extremities=extremities
         )
 
@@ -770,7 +765,7 @@ class FederationHandler(BaseHandler):
         #     self._sanity_check_event(ev)
 
         # Don't bother processing events we already have.
-        seen_events = yield self.store.have_events_in_timeline(
+        seen_events = await self.store.have_events_in_timeline(
             set(e.event_id for e in events)
         )
 
@@ -796,7 +791,7 @@ class FederationHandler(BaseHandler):
         state_events = {}
         events_to_state = {}
         for e_id in edges:
-            state, auth = yield self._get_state_for_room(
+            state, auth = await self._get_state_for_room(
                 destination=dest,
                 room_id=room_id,
                 event_id=e_id,
@@ -843,7 +838,7 @@ class FederationHandler(BaseHandler):
                 )
             )
 
-        yield self._handle_new_events(dest, ev_infos, backfilled=True)
+        await self._handle_new_events(dest, ev_infos, backfilled=True)
 
         # Step 2: Persist the rest of the events in the chunk one by one
         events.sort(key=lambda e: e.depth)
@@ -859,16 +854,15 @@ class FederationHandler(BaseHandler):
             # 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)
+            await self._handle_new_event(dest, event, backfilled=True)
 
         return events
 
-    @defer.inlineCallbacks
-    def maybe_backfill(self, room_id, current_depth):
+    async def maybe_backfill(self, room_id, current_depth):
         """Checks the database to see if we should backfill before paginating,
         and if so do.
         """
-        extremities = yield self.store.get_oldest_events_with_depth_in_room(room_id)
+        extremities = await self.store.get_oldest_events_with_depth_in_room(room_id)
 
         if not extremities:
             logger.debug("Not backfilling as no extremeties found.")
@@ -900,15 +894,17 @@ class FederationHandler(BaseHandler):
         #   state *before* the event, ignoring the special casing certain event
         #   types have.
 
-        forward_events = yield self.store.get_successor_events(list(extremities))
+        forward_events = await self.store.get_successor_events(list(extremities))
 
-        extremities_events = yield self.store.get_events(
-            forward_events, check_redacted=False, get_prev_content=False
+        extremities_events = await self.store.get_events(
+            forward_events,
+            redact_behaviour=EventRedactBehaviour.AS_IS,
+            get_prev_content=False,
         )
 
         # We set `check_history_visibility_only` as we might otherwise get false
         # positives from users having been erased.
-        filtered_extremities = yield filter_events_for_server(
+        filtered_extremities = await filter_events_for_server(
             self.storage,
             self.server_name,
             list(extremities_events.values()),
@@ -938,7 +934,7 @@ class FederationHandler(BaseHandler):
         # First we try hosts that are already in the room
         # TODO: HEURISTIC ALERT.
 
-        curr_state = yield self.state_handler.get_current_state(room_id)
+        curr_state = await self.state_handler.get_current_state(room_id)
 
         def get_domains_from_state(state):
             """Get joined domains from state
@@ -977,12 +973,11 @@ class FederationHandler(BaseHandler):
             domain for domain, depth in curr_domains if domain != self.server_name
         ]
 
-        @defer.inlineCallbacks
-        def try_backfill(domains):
+        async def try_backfill(domains):
             # TODO: Should we try multiple of these at a time?
             for dom in domains:
                 try:
-                    yield self.backfill(
+                    await self.backfill(
                         dom, room_id, limit=100, extremities=extremities
                     )
                     # If this succeeded then we probably already have the
@@ -1013,7 +1008,7 @@ class FederationHandler(BaseHandler):
 
             return False
 
-        success = yield try_backfill(likely_domains)
+        success = await try_backfill(likely_domains)
         if success:
             return True
 
@@ -1027,7 +1022,7 @@ class FederationHandler(BaseHandler):
 
         logger.debug("calling resolve_state_groups in _maybe_backfill")
         resolve = preserve_fn(self.state_handler.resolve_state_groups_for_events)
-        states = yield make_deferred_yieldable(
+        states = await make_deferred_yieldable(
             defer.gatherResults(
                 [resolve(room_id, [e]) for e in event_ids], consumeErrors=True
             )
@@ -1037,7 +1032,7 @@ class FederationHandler(BaseHandler):
         # event_ids.
         states = dict(zip(event_ids, [s.state for s in states]))
 
-        state_map = yield self.store.get_events(
+        state_map = await self.store.get_events(
             [e_id for ids in itervalues(states) for e_id in itervalues(ids)],
             get_prev_content=False,
         )
@@ -1053,7 +1048,7 @@ class FederationHandler(BaseHandler):
         for e_id, _ in sorted_extremeties_tuple:
             likely_domains = get_domains_from_state(states[e_id])
 
-            success = yield try_backfill(
+            success = await try_backfill(
                 [dom for dom, _ in likely_domains if dom not in tried_domains]
             )
             if success:
@@ -1063,8 +1058,7 @@ class FederationHandler(BaseHandler):
 
         return False
 
-    @defer.inlineCallbacks
-    def _get_events_and_persist(
+    async def _get_events_and_persist(
         self, destination: str, room_id: str, events: Iterable[str]
     ):
         """Fetch the given events from a server, and persist them as outliers.
@@ -1072,7 +1066,7 @@ class FederationHandler(BaseHandler):
         Logs a warning if we can't find the given event.
         """
 
-        room_version = yield self.store.get_room_version(room_id)
+        room_version = await self.store.get_room_version(room_id)
 
         event_infos = []
 
@@ -1108,9 +1102,9 @@ class FederationHandler(BaseHandler):
                         e,
                     )
 
-        yield concurrently_execute(get_event, events, 5)
+        await concurrently_execute(get_event, events, 5)
 
-        yield self._handle_new_events(
+        await self._handle_new_events(
             destination, event_infos,
         )
 
@@ -1253,7 +1247,7 @@ class FederationHandler(BaseHandler):
             # Check whether this room is the result of an upgrade of a room we already know
             # about. If so, migrate over user information
             predecessor = yield self.store.get_room_predecessor(room_id)
-            if not predecessor:
+            if not predecessor or not isinstance(predecessor.get("room_id"), str):
                 return
             old_room_id = predecessor["room_id"]
             logger.debug(
@@ -1281,8 +1275,7 @@ class FederationHandler(BaseHandler):
 
         return True
 
-    @defer.inlineCallbacks
-    def _handle_queued_pdus(self, room_queue):
+    async def _handle_queued_pdus(self, room_queue):
         """Process PDUs which got queued up while we were busy send_joining.
 
         Args:
@@ -1298,7 +1291,7 @@ class FederationHandler(BaseHandler):
                     p.room_id,
                 )
                 with nested_logging_context(p.event_id):
-                    yield self.on_receive_pdu(origin, p, sent_to_us_directly=True)
+                    await self.on_receive_pdu(origin, p, sent_to_us_directly=True)
             except Exception as e:
                 logger.warning(
                     "Error handling queued PDU %s from %s: %s", p.event_id, origin, e
@@ -1428,7 +1421,7 @@ class FederationHandler(BaseHandler):
                 user = UserID.from_string(event.state_key)
                 yield self.user_joined_room(user, event.room_id)
 
-        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        prev_state_ids = yield context.get_prev_state_ids()
 
         state_ids = list(prev_state_ids.values())
         auth_chain = yield self.store.get_auth_chain(state_ids)
@@ -1496,7 +1489,7 @@ class FederationHandler(BaseHandler):
     @defer.inlineCallbacks
     def do_remotely_reject_invite(self, target_hosts, room_id, user_id, content):
         origin, event, event_format_version = yield self._make_and_verify_event(
-            target_hosts, room_id, user_id, "leave", content=content,
+            target_hosts, room_id, user_id, "leave", content=content
         )
         # Mark as outlier as we don't have any state for this event; we're not
         # even in the room.
@@ -1937,7 +1930,7 @@ class FederationHandler(BaseHandler):
         context = yield self.state_handler.compute_event_context(event, old_state=state)
 
         if not auth_events:
-            prev_state_ids = yield context.get_prev_state_ids(self.store)
+            prev_state_ids = yield context.get_prev_state_ids()
             auth_events_ids = yield self.auth.compute_auth_events(
                 event, prev_state_ids, for_verification=True
             )
@@ -2346,12 +2339,12 @@ class FederationHandler(BaseHandler):
             k: a.event_id for k, a in iteritems(auth_events) if k != event_key
         }
 
-        current_state_ids = yield context.get_current_state_ids(self.store)
+        current_state_ids = yield context.get_current_state_ids()
         current_state_ids = dict(current_state_ids)
 
         current_state_ids.update(state_updates)
 
-        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        prev_state_ids = yield context.get_prev_state_ids()
         prev_state_ids = dict(prev_state_ids)
 
         prev_state_ids.update({k: a.event_id for k, a in iteritems(auth_events)})
@@ -2635,7 +2628,7 @@ class FederationHandler(BaseHandler):
             event.content["third_party_invite"]["signed"]["token"],
         )
         original_invite = None
-        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        prev_state_ids = yield context.get_prev_state_ids()
         original_invite_id = prev_state_ids.get(key)
         if original_invite_id:
             original_invite = yield self.store.get_event(
@@ -2683,7 +2676,7 @@ class FederationHandler(BaseHandler):
         signed = event.content["third_party_invite"]["signed"]
         token = signed["token"]
 
-        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        prev_state_ids = yield context.get_prev_state_ids()
         invite_event_id = prev_state_ids.get((EventTypes.ThirdPartyInvite, token))
 
         invite_event = None
@@ -2857,7 +2850,7 @@ class FederationHandler(BaseHandler):
                 room_id=room_id, user_id=user.to_string(), change="joined"
             )
         else:
-            return user_joined_room(self.distributor, user, room_id)
+            return defer.succeed(user_joined_room(self.distributor, user, room_id))
 
     @defer.inlineCallbacks
     def get_room_complexity(self, remote_room_hosts, room_id):

synapse/handlers/initial_sync.py

@@ -26,7 +26,7 @@ from synapse.streams.config import PaginationConfig
 from synapse.types import StreamToken, UserID
 from synapse.util import unwrapFirstError
 from synapse.util.async_helpers import concurrently_execute
-from synapse.util.caches.snapshot_cache import SnapshotCache
+from synapse.util.caches.response_cache import ResponseCache
 from synapse.visibility import filter_events_for_client
 
 from ._base import BaseHandler
@@ -41,7 +41,7 @@ class InitialSyncHandler(BaseHandler):
         self.state = hs.get_state_handler()
         self.clock = hs.get_clock()
         self.validator = EventValidator()
-        self.snapshot_cache = SnapshotCache()
+        self.snapshot_cache = ResponseCache(hs, "initial_sync_cache")
         self._event_serializer = hs.get_event_client_serializer()
         self.storage = hs.get_storage()
         self.state_store = self.storage.state
@@ -79,21 +79,17 @@ class InitialSyncHandler(BaseHandler):
             as_client_event,
             include_archived,
         )
-        now_ms = self.clock.time_msec()
-        result = self.snapshot_cache.get(now_ms, key)
-        if result is not None:
-            return result
 
-        return self.snapshot_cache.set(
-            now_ms,
+        return self.snapshot_cache.wrap(
             key,
-            self._snapshot_all_rooms(
-                user_id, pagin_config, as_client_event, include_archived
-            ),
+            self._snapshot_all_rooms,
+            user_id,
+            pagin_config,
+            as_client_event,
+            include_archived,
         )
 
-    @defer.inlineCallbacks
-    def _snapshot_all_rooms(
+    async def _snapshot_all_rooms(
         self,
         user_id=None,
         pagin_config=None,
@@ -105,7 +101,7 @@ class InitialSyncHandler(BaseHandler):
         if include_archived:
             memberships.append(Membership.LEAVE)
 
-        room_list = yield self.store.get_rooms_for_user_where_membership_is(
+        room_list = await self.store.get_rooms_for_user_where_membership_is(
             user_id=user_id, membership_list=memberships
         )
 
@@ -113,33 +109,32 @@ class InitialSyncHandler(BaseHandler):
 
         rooms_ret = []
 
-        now_token = yield self.hs.get_event_sources().get_current_token()
+        now_token = await self.hs.get_event_sources().get_current_token()
 
         presence_stream = self.hs.get_event_sources().sources["presence"]
         pagination_config = PaginationConfig(from_token=now_token)
-        presence, _ = yield presence_stream.get_pagination_rows(
+        presence, _ = await presence_stream.get_pagination_rows(
             user, pagination_config.get_source_config("presence"), None
         )
 
         receipt_stream = self.hs.get_event_sources().sources["receipt"]
-        receipt, _ = yield receipt_stream.get_pagination_rows(
+        receipt, _ = await receipt_stream.get_pagination_rows(
             user, pagination_config.get_source_config("receipt"), None
         )
 
-        tags_by_room = yield self.store.get_tags_for_user(user_id)
+        tags_by_room = await self.store.get_tags_for_user(user_id)
 
-        account_data, account_data_by_room = yield self.store.get_account_data_for_user(
+        account_data, account_data_by_room = await self.store.get_account_data_for_user(
             user_id
         )
 
-        public_room_ids = yield self.store.get_public_room_ids()
+        public_room_ids = await self.store.get_public_room_ids()
 
         limit = pagin_config.limit
         if limit is None:
             limit = 10
 
-        @defer.inlineCallbacks
-        def handle_room(event):
+        async def handle_room(event):
             d = {
                 "room_id": event.room_id,
                 "membership": event.membership,
@@ -152,8 +147,8 @@ class InitialSyncHandler(BaseHandler):
                 time_now = self.clock.time_msec()
                 d["inviter"] = event.sender
 
-                invite_event = yield self.store.get_event(event.event_id)
-                d["invite"] = yield self._event_serializer.serialize_event(
+                invite_event = await self.store.get_event(event.event_id)
+                d["invite"] = await self._event_serializer.serialize_event(
                     invite_event, time_now, as_client_event
                 )
 
@@ -177,7 +172,7 @@ class InitialSyncHandler(BaseHandler):
                         lambda states: states[event.event_id]
                     )
 
-                (messages, token), current_state = yield make_deferred_yieldable(
+                (messages, token), current_state = await make_deferred_yieldable(
                     defer.gatherResults(
                         [
                             run_in_background(
@@ -191,7 +186,7 @@ class InitialSyncHandler(BaseHandler):
                     )
                 ).addErrback(unwrapFirstError)
 
-                messages = yield filter_events_for_client(
+                messages = await filter_events_for_client(
                     self.storage, user_id, messages
                 )
 
@@ -201,7 +196,7 @@ class InitialSyncHandler(BaseHandler):
 
                 d["messages"] = {
                     "chunk": (
-                        yield self._event_serializer.serialize_events(
+                        await self._event_serializer.serialize_events(
                             messages, time_now=time_now, as_client_event=as_client_event
                         )
                     ),
@@ -209,7 +204,7 @@ class InitialSyncHandler(BaseHandler):
                     "end": end_token.to_string(),
                 }
 
-                d["state"] = yield self._event_serializer.serialize_events(
+                d["state"] = await self._event_serializer.serialize_events(
                     current_state.values(),
                     time_now=time_now,
                     as_client_event=as_client_event,
@@ -232,7 +227,7 @@ class InitialSyncHandler(BaseHandler):
             except Exception:
                 logger.exception("Failed to get snapshot")
 
-        yield concurrently_execute(handle_room, room_list, 10)
+        await concurrently_execute(handle_room, room_list, 10)
 
         account_data_events = []
         for account_data_type, content in account_data.items():
@@ -256,8 +251,7 @@ class InitialSyncHandler(BaseHandler):
 
         return ret
 
-    @defer.inlineCallbacks
-    def room_initial_sync(self, requester, room_id, pagin_config=None):
+    async def room_initial_sync(self, requester, room_id, pagin_config=None):
         """Capture the a snapshot of a room. If user is currently a member of
         the room this will be what is currently in the room. If the user left
         the room this will be what was in the room when they left.
@@ -274,32 +268,32 @@ class InitialSyncHandler(BaseHandler):
             A JSON serialisable dict with the snapshot of the room.
         """
 
-        blocked = yield self.store.is_room_blocked(room_id)
+        blocked = await self.store.is_room_blocked(room_id)
         if blocked:
             raise SynapseError(403, "This room has been blocked on this server")
 
         user_id = requester.user.to_string()
 
-        membership, member_event_id = yield self._check_in_room_or_world_readable(
+        membership, member_event_id = await self._check_in_room_or_world_readable(
             room_id, user_id
         )
         is_peeking = member_event_id is None
 
         if membership == Membership.JOIN:
-            result = yield self._room_initial_sync_joined(
+            result = await self._room_initial_sync_joined(
                 user_id, room_id, pagin_config, membership, is_peeking
             )
         elif membership == Membership.LEAVE:
-            result = yield self._room_initial_sync_parted(
+            result = await self._room_initial_sync_parted(
                 user_id, room_id, pagin_config, membership, member_event_id, is_peeking
             )
 
         account_data_events = []
-        tags = yield self.store.get_tags_for_room(user_id, room_id)
+        tags = await self.store.get_tags_for_room(user_id, room_id)
         if tags:
             account_data_events.append({"type": "m.tag", "content": {"tags": tags}})
 
-        account_data = yield self.store.get_account_data_for_room(user_id, room_id)
+        account_data = await self.store.get_account_data_for_room(user_id, room_id)
         for account_data_type, content in account_data.items():
             account_data_events.append({"type": account_data_type, "content": content})
 
@@ -307,11 +301,10 @@ class InitialSyncHandler(BaseHandler):
 
         return result
 
-    @defer.inlineCallbacks
-    def _room_initial_sync_parted(
+    async def _room_initial_sync_parted(
         self, user_id, room_id, pagin_config, membership, member_event_id, is_peeking
     ):
-        room_state = yield self.state_store.get_state_for_events([member_event_id])
+        room_state = await self.state_store.get_state_for_events([member_event_id])
 
         room_state = room_state[member_event_id]
 
@@ -319,13 +312,13 @@ class InitialSyncHandler(BaseHandler):
         if limit is None:
             limit = 10
 
-        stream_token = yield self.store.get_stream_token_for_event(member_event_id)
+        stream_token = await self.store.get_stream_token_for_event(member_event_id)
 
-        messages, token = yield self.store.get_recent_events_for_room(
+        messages, token = await self.store.get_recent_events_for_room(
             room_id, limit=limit, end_token=stream_token
         )
 
-        messages = yield filter_events_for_client(
+        messages = await filter_events_for_client(
             self.storage, user_id, messages, is_peeking=is_peeking
         )
 
@@ -339,13 +332,13 @@ class InitialSyncHandler(BaseHandler):
             "room_id": room_id,
             "messages": {
                 "chunk": (
-                    yield self._event_serializer.serialize_events(messages, time_now)
+                    await self._event_serializer.serialize_events(messages, time_now)
                 ),
                 "start": start_token.to_string(),
                 "end": end_token.to_string(),
             },
             "state": (
-                yield self._event_serializer.serialize_events(
+                await self._event_serializer.serialize_events(
                     room_state.values(), time_now
                 )
             ),
@@ -353,19 +346,18 @@ class InitialSyncHandler(BaseHandler):
             "receipts": [],
         }
 
-    @defer.inlineCallbacks
-    def _room_initial_sync_joined(
+    async def _room_initial_sync_joined(
         self, user_id, room_id, pagin_config, membership, is_peeking
     ):
-        current_state = yield self.state.get_current_state(room_id=room_id)
+        current_state = await self.state.get_current_state(room_id=room_id)
 
         # TODO: These concurrently
         time_now = self.clock.time_msec()
-        state = yield self._event_serializer.serialize_events(
+        state = await self._event_serializer.serialize_events(
             current_state.values(), time_now
         )
 
-        now_token = yield self.hs.get_event_sources().get_current_token()
+        now_token = await self.hs.get_event_sources().get_current_token()
 
         limit = pagin_config.limit if pagin_config else None
         if limit is None:
@@ -380,28 +372,26 @@ class InitialSyncHandler(BaseHandler):
 
         presence_handler = self.hs.get_presence_handler()
 
-        @defer.inlineCallbacks
-        def get_presence():
+        async def get_presence():
             # If presence is disabled, return an empty list
             if not self.hs.config.use_presence:
                 return []
 
-            states = yield presence_handler.get_states(
+            states = await presence_handler.get_states(
                 [m.user_id for m in room_members], as_event=True
             )
 
             return states
 
-        @defer.inlineCallbacks
-        def get_receipts():
-            receipts = yield self.store.get_linearized_receipts_for_room(
+        async def get_receipts():
+            receipts = await self.store.get_linearized_receipts_for_room(
                 room_id, to_key=now_token.receipt_key
             )
             if not receipts:
                 receipts = []
             return receipts
 
-        presence, receipts, (messages, token) = yield make_deferred_yieldable(
+        presence, receipts, (messages, token) = await make_deferred_yieldable(
             defer.gatherResults(
                 [
                     run_in_background(get_presence),
@@ -417,7 +407,7 @@ class InitialSyncHandler(BaseHandler):
             ).addErrback(unwrapFirstError)
         )
 
-        messages = yield filter_events_for_client(
+        messages = await filter_events_for_client(
             self.storage, user_id, messages, is_peeking=is_peeking
         )
 
@@ -430,7 +420,7 @@ class InitialSyncHandler(BaseHandler):
             "room_id": room_id,
             "messages": {
                 "chunk": (
-                    yield self._event_serializer.serialize_events(messages, time_now)
+                    await self._event_serializer.serialize_events(messages, time_now)
                 ),
                 "start": start_token.to_string(),
                 "end": end_token.to_string(),
@@ -444,18 +434,17 @@ class InitialSyncHandler(BaseHandler):
 
         return ret
 
-    @defer.inlineCallbacks
-    def _check_in_room_or_world_readable(self, room_id, user_id):
+    async def _check_in_room_or_world_readable(self, room_id, user_id):
         try:
             # check_user_was_in_room will return the most recent membership
             # event for the user if:
             #  * The user is a non-guest user, and was ever in the room
             #  * The user is a guest user, and has joined the room
             # else it will throw.
-            member_event = yield self.auth.check_user_was_in_room(room_id, user_id)
+            member_event = await self.auth.check_user_was_in_room(room_id, user_id)
             return member_event.membership, member_event.event_id
         except AuthError:
-            visibility = yield self.state_handler.get_current_state(
+            visibility = await self.state_handler.get_current_state(
                 room_id, EventTypes.RoomHistoryVisibility, ""
             )
             if (

synapse/handlers/message.py

@@ -46,8 +46,9 @@ from synapse.events.validator import EventValidator
 from synapse.logging.context import run_in_background
 from synapse.metrics.background_process_metrics import run_as_background_process
 from synapse.replication.http.send_event import ReplicationSendEventRestServlet
+from synapse.storage.data_stores.main.events_worker import EventRedactBehaviour
 from synapse.storage.state import StateFilter
-from synapse.types import RoomAlias, UserID, create_requester
+from synapse.types import Collection, RoomAlias, UserID, create_requester
 from synapse.util.async_helpers import Linearizer
 from synapse.util.frozenutils import frozendict_json_encoder
 from synapse.util.metrics import measure_func
@@ -421,7 +422,7 @@ class EventCreationHandler(object):
         event_dict,
         token_id=None,
         txn_id=None,
-        prev_events_and_hashes=None,
+        prev_event_ids: Optional[Collection[str]] = None,
         require_consent=True,
     ):
         """
@@ -438,10 +439,9 @@ class EventCreationHandler(object):
             token_id (str)
             txn_id (str)
 
-            prev_events_and_hashes (list[(str, dict[str, str], int)]|None):
+            prev_event_ids:
                 the forward extremities to use as the prev_events for the
-                new event. For each event, a tuple of (event_id, hashes, depth)
-                where *hashes* is a map from algorithm to hash.
+                new event.
 
                 If None, they will be requested from the database.
 
@@ -497,9 +497,7 @@ class EventCreationHandler(object):
             builder.internal_metadata.txn_id = txn_id
 
         event, context = yield self.create_new_client_event(
-            builder=builder,
-            requester=requester,
-            prev_events_and_hashes=prev_events_and_hashes,
+            builder=builder, requester=requester, prev_event_ids=prev_event_ids,
         )
 
         # In an ideal world we wouldn't need the second part of this condition. However,
@@ -514,7 +512,7 @@ class EventCreationHandler(object):
             # federation as well as those created locally. As of room v3, aliases events
             # can be created by users that are not in the room, therefore we have to
             # tolerate them in event_auth.check().
-            prev_state_ids = yield context.get_prev_state_ids(self.store)
+            prev_state_ids = yield context.get_prev_state_ids()
             prev_event_id = prev_state_ids.get((EventTypes.Member, event.sender))
             prev_event = (
                 yield self.store.get_event(prev_event_id, allow_none=True)
@@ -664,7 +662,7 @@ class EventCreationHandler(object):
         If so, returns the version of the event in context.
         Otherwise, returns None.
         """
-        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        prev_state_ids = yield context.get_prev_state_ids()
         prev_event_id = prev_state_ids.get((event.type, event.state_key))
         if not prev_event_id:
             return
@@ -713,7 +711,7 @@ class EventCreationHandler(object):
     @measure_func("create_new_client_event")
     @defer.inlineCallbacks
     def create_new_client_event(
-        self, builder, requester=None, prev_events_and_hashes=None
+        self, builder, requester=None, prev_event_ids: Optional[Collection[str]] = None
     ):
         """Create a new event for a local client
 
@@ -722,10 +720,9 @@ class EventCreationHandler(object):
 
             requester (synapse.types.Requester|None):
 
-            prev_events_and_hashes (list[(str, dict[str, str], int)]|None):
+            prev_event_ids:
                 the forward extremities to use as the prev_events for the
-                new event. For each event, a tuple of (event_id, hashes, depth)
-                where *hashes* is a map from algorithm to hash.
+                new event.
 
                 If None, they will be requested from the database.
 
@@ -733,22 +730,15 @@ class EventCreationHandler(object):
             Deferred[(synapse.events.EventBase, synapse.events.snapshot.EventContext)]
         """
 
-        if prev_events_and_hashes is not None:
-            assert len(prev_events_and_hashes) <= 10, (
+        if prev_event_ids is not None:
+            assert len(prev_event_ids) <= 10, (
                 "Attempting to create an event with %i prev_events"
-                % (len(prev_events_and_hashes),)
+                % (len(prev_event_ids),)
             )
         else:
-            prev_events_and_hashes = yield self.store.get_prev_events_for_room(
-                builder.room_id
-            )
+            prev_event_ids = yield self.store.get_prev_events_for_room(builder.room_id)
 
-        prev_events = [
-            (event_id, prev_hashes)
-            for event_id, prev_hashes, _ in prev_events_and_hashes
-        ]
-
-        event = yield builder.build(prev_event_ids=[p for p, _ in prev_events])
+        event = yield builder.build(prev_event_ids=prev_event_ids)
         context = yield self.state.compute_event_context(event)
         if requester:
             context.app_service = requester.app_service
@@ -875,7 +865,7 @@ class EventCreationHandler(object):
             if event.type == EventTypes.Redaction:
                 original_event = yield self.store.get_event(
                     event.redacts,
-                    check_redacted=False,
+                    redact_behaviour=EventRedactBehaviour.AS_IS,
                     get_prev_content=False,
                     allow_rejected=False,
                     allow_none=True,
@@ -913,7 +903,7 @@ class EventCreationHandler(object):
                 def is_inviter_member_event(e):
                     return e.type == EventTypes.Member and e.sender == event.sender
 
-                current_state_ids = yield context.get_current_state_ids(self.store)
+                current_state_ids = yield context.get_current_state_ids()
 
                 state_to_include_ids = [
                     e_id
@@ -952,7 +942,7 @@ class EventCreationHandler(object):
         if event.type == EventTypes.Redaction:
             original_event = yield self.store.get_event(
                 event.redacts,
-                check_redacted=False,
+                redact_behaviour=EventRedactBehaviour.AS_IS,
                 get_prev_content=False,
                 allow_rejected=False,
                 allow_none=True,
@@ -966,7 +956,7 @@ class EventCreationHandler(object):
                 if original_event.room_id != event.room_id:
                     raise SynapseError(400, "Cannot redact event from a different room")
 
-            prev_state_ids = yield context.get_prev_state_ids(self.store)
+            prev_state_ids = yield context.get_prev_state_ids()
             auth_events_ids = yield self.auth.compute_auth_events(
                 event, prev_state_ids, for_verification=True
             )
@@ -988,7 +978,7 @@ class EventCreationHandler(object):
                 event.internal_metadata.recheck_redaction = False
 
         if event.type == EventTypes.Create:
-            prev_state_ids = yield context.get_prev_state_ids(self.store)
+            prev_state_ids = yield context.get_prev_state_ids()
             if prev_state_ids:
                 raise AuthError(403, "Changing the room create event is forbidden")
 
@@ -1041,9 +1031,7 @@ class EventCreationHandler(object):
             # For each room we need to find a joined member we can use to send
             # the dummy event with.
 
-            prev_events_and_hashes = yield self.store.get_prev_events_for_room(room_id)
-
-            latest_event_ids = (event_id for (event_id, _, _) in prev_events_and_hashes)
+            latest_event_ids = yield self.store.get_prev_events_for_room(room_id)
 
             members = yield self.state.get_current_users_in_room(
                 room_id, latest_event_ids=latest_event_ids
@@ -1062,7 +1050,7 @@ class EventCreationHandler(object):
                             "room_id": room_id,
                             "sender": user_id,
                         },
-                        prev_events_and_hashes=prev_events_and_hashes,
+                        prev_event_ids=latest_event_ids,
                     )
 
                     event.internal_metadata.proactively_send = False

synapse/handlers/pagination.py

@@ -280,8 +280,7 @@ class PaginationHandler(object):
 
             await self.storage.purge_events.purge_room(room_id)
 
-    @defer.inlineCallbacks
-    def get_messages(
+    async def get_messages(
         self,
         requester,
         room_id=None,
@@ -307,7 +306,7 @@ class PaginationHandler(object):
             room_token = pagin_config.from_token.room_key
         else:
             pagin_config.from_token = (
-                yield self.hs.get_event_sources().get_current_token_for_pagination()
+                await self.hs.get_event_sources().get_current_token_for_pagination()
             )
             room_token = pagin_config.from_token.room_key
 
@@ -319,11 +318,11 @@ class PaginationHandler(object):
 
         source_config = pagin_config.get_source_config("room")
 
-        with (yield self.pagination_lock.read(room_id)):
+        with (await self.pagination_lock.read(room_id)):
             (
                 membership,
                 member_event_id,
-            ) = yield self.auth.check_in_room_or_world_readable(room_id, user_id)
+            ) = await self.auth.check_in_room_or_world_readable(room_id, user_id)
 
             if source_config.direction == "b":
                 # if we're going backwards, we might need to backfill. This
@@ -331,7 +330,7 @@ class PaginationHandler(object):
                 if room_token.topological:
                     max_topo = room_token.topological
                 else:
-                    max_topo = yield self.store.get_max_topological_token(
+                    max_topo = await self.store.get_max_topological_token(
                         room_id, room_token.stream
                     )
 
@@ -339,18 +338,18 @@ class PaginationHandler(object):
                     # If they have left the room then clamp the token to be before
                     # they left the room, to save the effort of loading from the
                     # database.
-                    leave_token = yield self.store.get_topological_token_for_event(
+                    leave_token = await self.store.get_topological_token_for_event(
                         member_event_id
                     )
                     leave_token = RoomStreamToken.parse(leave_token)
                     if leave_token.topological < max_topo:
                         source_config.from_key = str(leave_token)
 
-                yield self.hs.get_handlers().federation_handler.maybe_backfill(
+                await self.hs.get_handlers().federation_handler.maybe_backfill(
                     room_id, max_topo
                 )
 
-            events, next_key = yield self.store.paginate_room_events(
+            events, next_key = await self.store.paginate_room_events(
                 room_id=room_id,
                 from_key=source_config.from_key,
                 to_key=source_config.to_key,
@@ -365,7 +364,7 @@ class PaginationHandler(object):
             if event_filter:
                 events = event_filter.filter(events)
 
-            events = yield filter_events_for_client(
+            events = await filter_events_for_client(
                 self.storage, user_id, events, is_peeking=(member_event_id is None)
             )
 
@@ -385,19 +384,19 @@ class PaginationHandler(object):
                 (EventTypes.Member, event.sender) for event in events
             )
 
-            state_ids = yield self.state_store.get_state_ids_for_event(
+            state_ids = await self.state_store.get_state_ids_for_event(
                 events[0].event_id, state_filter=state_filter
             )
 
             if state_ids:
-                state = yield self.store.get_events(list(state_ids.values()))
+                state = await self.store.get_events(list(state_ids.values()))
                 state = state.values()
 
         time_now = self.clock.time_msec()
 
         chunk = {
             "chunk": (
-                yield self._event_serializer.serialize_events(
+                await self._event_serializer.serialize_events(
                     events, time_now, as_client_event=as_client_event
                 )
             ),
@@ -406,7 +405,7 @@ class PaginationHandler(object):
         }
 
         if state:
-            chunk["state"] = yield self._event_serializer.serialize_events(
+            chunk["state"] = await self._event_serializer.serialize_events(
                 state, time_now, as_client_event=as_client_event
             )
 

synapse/handlers/presence.py

@@ -95,12 +95,7 @@ assert LAST_ACTIVE_GRANULARITY < IDLE_TIMER
 
 
 class PresenceHandler(object):
-    def __init__(self, hs):
-        """
-
-        Args:
-            hs (synapse.server.HomeServer):
-        """
+    def __init__(self, hs: "synapse.server.HomeServer"):
         self.hs = hs
         self.is_mine = hs.is_mine
         self.is_mine_id = hs.is_mine_id
@@ -230,7 +225,7 @@ class PresenceHandler(object):
         is some spurious presence changes that will self-correct.
         """
         # If the DB pool has already terminated, don't try updating
-        if not self.hs.get_db_pool().running:
+        if not self.store.db.is_running():
             return
 
         logger.info(

synapse/handlers/profile.py

@@ -295,12 +295,16 @@ class BaseProfileHandler(BaseHandler):
                 be found to be in any room the server is in, and therefore the query
                 is denied.
         """
+
         # Implementation of MSC1301: don't allow looking up profiles if the
         # requester isn't in the same room as the target. We expect requester to
         # be None when this function is called outside of a profile query, e.g.
         # when building a membership event. In this case, we must allow the
         # lookup.
-        if not self.hs.config.require_auth_for_profile_requests or not requester:
+        if (
+            not self.hs.config.limit_profile_requests_to_users_who_share_rooms
+            or not requester
+        ):
             return
 
         # Always allow the user to query their own profile.

synapse/handlers/room.py

@@ -16,6 +16,7 @@
 # limitations under the License.
 
 """Contains functions for performing events on rooms."""
+
 import itertools
 import logging
 import math
@@ -184,7 +185,7 @@ class RoomCreationHandler(BaseHandler):
             requester, tombstone_event, tombstone_context
         )
 
-        old_room_state = yield tombstone_context.get_current_state_ids(self.store)
+        old_room_state = yield tombstone_context.get_current_state_ids()
 
         # update any aliases
         yield self._move_aliases_to_new_room(
@@ -271,7 +272,7 @@ class RoomCreationHandler(BaseHandler):
             except AuthError as e:
                 logger.warning("Unable to update PLs in old room: %s", e)
 
-        logger.info("Setting correct PLs in new room")
+        logger.info("Setting correct PLs in new room to %s", old_room_pl_state.content)
         yield self.event_creation_handler.create_and_send_nonmember_event(
             requester,
             {
@@ -365,13 +366,18 @@ class RoomCreationHandler(BaseHandler):
         needed_power_level = max(state_default, ban, max(event_power_levels.values()))
 
         # Raise the requester's power level in the new room if necessary
-        current_power_level = power_levels["users"][requester.user.to_string()]
+        current_power_level = power_levels["users"][user_id]
         if current_power_level < needed_power_level:
-            # Assign this power level to the requester
-            power_levels["users"][requester.user.to_string()] = needed_power_level
+            # make sure we copy the event content rather than overwriting it.
+            # note that if frozen_dicts are enabled, `power_levels` will be a frozen
+            # dict so we can't just copy.deepcopy it.
 
-        # Set the power levels to the modified state
-        initial_state[(EventTypes.PowerLevels, "")] = power_levels
+            new_power_levels = {k: v for k, v in power_levels.items() if k != "users"}
+            new_power_levels["users"] = {
+                k: v for k, v in power_levels.get("users", {}).items() if k != user_id
+            }
+            new_power_levels["users"][user_id] = needed_power_level
+            initial_state[(EventTypes.PowerLevels, "")] = new_power_levels
 
         yield self._send_events_for_new_room(
             requester,
@@ -733,7 +739,7 @@ class RoomCreationHandler(BaseHandler):
         initial_state,
         creation_content,
         room_alias=None,
-        power_level_content_override=None,
+        power_level_content_override=None,  # Doesn't apply when initial state has power level state event content
         creator_join_profile=None,
     ):
         def create(etype, content, **kwargs):
@@ -1011,15 +1017,3 @@ class RoomEventSource(object):
 
     def get_current_key_for_room(self, room_id):
         return self.store.get_room_events_max_id(room_id)
-
-    @defer.inlineCallbacks
-    def get_pagination_rows(self, user, config, key):
-        events, next_key = yield self.store.paginate_room_events(
-            room_id=key,
-            from_key=config.from_key,
-            to_key=config.to_key,
-            direction=config.direction,
-            limit=config.limit,
-        )
-
-        return (events, next_key)

synapse/handlers/room_member.py

@@ -25,7 +25,7 @@ from twisted.internet import defer
 from synapse import types
 from synapse.api.constants import EventTypes, Membership
 from synapse.api.errors import AuthError, Codes, SynapseError
-from synapse.types import RoomID, UserID
+from synapse.types import Collection, RoomID, UserID
 from synapse.util.async_helpers import Linearizer
 from synapse.util.distributor import user_joined_room, user_left_room
 
@@ -149,7 +149,7 @@ class RoomMemberHandler(object):
         target,
         room_id,
         membership,
-        prev_events_and_hashes,
+        prev_event_ids: Collection[str],
         txn_id=None,
         ratelimit=True,
         content=None,
@@ -177,7 +177,7 @@ class RoomMemberHandler(object):
             },
             token_id=requester.access_token_id,
             txn_id=txn_id,
-            prev_events_and_hashes=prev_events_and_hashes,
+            prev_event_ids=prev_event_ids,
             require_consent=require_consent,
         )
 
@@ -193,7 +193,7 @@ class RoomMemberHandler(object):
             requester, event, context, extra_users=[target], ratelimit=ratelimit
         )
 
-        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        prev_state_ids = yield context.get_prev_state_ids()
 
         prev_member_event_id = prev_state_ids.get((EventTypes.Member, user_id), None)
 
@@ -370,8 +370,7 @@ class RoomMemberHandler(object):
             if block_invite:
                 raise SynapseError(403, "Invites have been disabled on this server")
 
-        prev_events_and_hashes = yield self.store.get_prev_events_for_room(room_id)
-        latest_event_ids = (event_id for (event_id, _, _) in prev_events_and_hashes)
+        latest_event_ids = yield self.store.get_prev_events_for_room(room_id)
 
         current_state_ids = yield self.state_handler.get_current_state_ids(
             room_id, latest_event_ids=latest_event_ids
@@ -485,7 +484,7 @@ class RoomMemberHandler(object):
             membership=effective_membership_state,
             txn_id=txn_id,
             ratelimit=ratelimit,
-            prev_events_and_hashes=prev_events_and_hashes,
+            prev_event_ids=latest_event_ids,
             content=content,
             require_consent=require_consent,
         )
@@ -507,6 +506,8 @@ class RoomMemberHandler(object):
         Returns:
             Deferred
         """
+        logger.info("Transferring room state from %s to %s", old_room_id, room_id)
+
         # Find all local users that were in the old room and copy over each user's state
         users = yield self.store.get_users_in_room(old_room_id)
         yield self.copy_user_state_on_room_upgrade(old_room_id, room_id, users)
@@ -601,7 +602,7 @@ class RoomMemberHandler(object):
         if prev_event is not None:
             return
 
-        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        prev_state_ids = yield context.get_prev_state_ids()
         if event.membership == Membership.JOIN:
             if requester.is_guest:
                 guest_can_join = yield self._can_guest_join(prev_state_ids)

synapse/handlers/saml_handler.py

@@ -13,20 +13,36 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import logging
+import re
+from typing import Tuple
 
 import attr
 import saml2
+import saml2.response
 from saml2.client import Saml2Client
 
 from synapse.api.errors import SynapseError
+from synapse.config import ConfigError
 from synapse.http.servlet import parse_string
 from synapse.rest.client.v1.login import SSOAuthHandler
-from synapse.types import UserID, map_username_to_mxid_localpart
+from synapse.types import (
+    UserID,
+    map_username_to_mxid_localpart,
+    mxid_localpart_allowed_characters,
+)
 from synapse.util.async_helpers import Linearizer
 
 logger = logging.getLogger(__name__)
 
 
+@attr.s
+class Saml2SessionData:
+    """Data we track about SAML2 sessions"""
+
+    # time the session was created, in milliseconds
+    creation_time = attr.ib()
+
+
 class SamlHandler:
     def __init__(self, hs):
         self._saml_client = Saml2Client(hs.config.saml2_sp_config)
@@ -37,11 +53,14 @@ class SamlHandler:
         self._datastore = hs.get_datastore()
         self._hostname = hs.hostname
         self._saml2_session_lifetime = hs.config.saml2_session_lifetime
-        self._mxid_source_attribute = hs.config.saml2_mxid_source_attribute
         self._grandfathered_mxid_source_attribute = (
             hs.config.saml2_grandfathered_mxid_source_attribute
         )
-        self._mxid_mapper = hs.config.saml2_mxid_mapper
+
+        # plugin to do custom mapping from saml response to mxid
+        self._user_mapping_provider = hs.config.saml2_user_mapping_provider_class(
+            hs.config.saml2_user_mapping_provider_config
+        )
 
         # identifier for the external_ids table
         self._auth_provider_id = "saml"
@@ -118,22 +137,10 @@ class SamlHandler:
             remote_user_id = saml2_auth.ava["uid"][0]
         except KeyError:
             logger.warning("SAML2 response lacks a 'uid' attestation")
-            raise SynapseError(400, "uid not in SAML2 response")
-
-        try:
-            mxid_source = saml2_auth.ava[self._mxid_source_attribute][0]
-        except KeyError:
-            logger.warning(
-                "SAML2 response lacks a '%s' attestation", self._mxid_source_attribute
-            )
-            raise SynapseError(
-                400, "%s not in SAML2 response" % (self._mxid_source_attribute,)
-            )
+            raise SynapseError(400, "'uid' not in SAML2 response")
 
         self._outstanding_requests_dict.pop(saml2_auth.in_response_to, None)
 
-        displayName = saml2_auth.ava.get("displayName", [None])[0]
-
         with (await self._mapping_lock.queue(self._auth_provider_id)):
             # first of all, check if we already have a mapping for this user
             logger.info(
@@ -173,22 +180,46 @@ class SamlHandler:
                     )
                     return registered_user_id
 
-            # figure out a new mxid for this user
-            base_mxid_localpart = self._mxid_mapper(mxid_source)
+            # Map saml response to user attributes using the configured mapping provider
+            for i in range(1000):
+                attribute_dict = self._user_mapping_provider.saml_response_to_user_attributes(
+                    saml2_auth, i
+                )
 
-            suffix = 0
-            while True:
-                localpart = base_mxid_localpart + (str(suffix) if suffix else "")
+                logger.debug(
+                    "Retrieved SAML attributes from user mapping provider: %s "
+                    "(attempt %d)",
+                    attribute_dict,
+                    i,
+                )
+
+                localpart = attribute_dict.get("mxid_localpart")
+                if not localpart:
+                    logger.error(
+                        "SAML mapping provider plugin did not return a "
+                        "mxid_localpart object"
+                    )
+                    raise SynapseError(500, "Error parsing SAML2 response")
+
+                displayname = attribute_dict.get("displayname")
+
+                # Check if this mxid already exists
                 if not await self._datastore.get_users_by_id_case_insensitive(
                     UserID(localpart, self._hostname).to_string()
                 ):
+                    # This mxid is free
                     break
-                suffix += 1
-            logger.info("Allocating mxid for new user with localpart %s", localpart)
+            else:
+                # Unable to generate a username in 1000 iterations
+                # Break and return error to the user
+                raise SynapseError(
+                    500, "Unable to generate a Matrix ID from the SAML response"
+                )
 
             registered_user_id = await self._registration_handler.register_user(
-                localpart=localpart, default_display_name=displayName
+                localpart=localpart, default_display_name=displayname
             )
+
             await self._datastore.record_user_external_id(
                 self._auth_provider_id, remote_user_id, registered_user_id
             )
@@ -205,9 +236,120 @@ class SamlHandler:
             del self._outstanding_requests_dict[reqid]
 
 
-@attr.s
-class Saml2SessionData:
-    """Data we track about SAML2 sessions"""
+DOT_REPLACE_PATTERN = re.compile(
+    ("[^%s]" % (re.escape("".join(mxid_localpart_allowed_characters)),))
+)
 
-    # time the session was created, in milliseconds
-    creation_time = attr.ib()
+
+def dot_replace_for_mxid(username: str) -> str:
+    username = username.lower()
+    username = DOT_REPLACE_PATTERN.sub(".", username)
+
+    # regular mxids aren't allowed to start with an underscore either
+    username = re.sub("^_", "", username)
+    return username
+
+
+MXID_MAPPER_MAP = {
+    "hexencode": map_username_to_mxid_localpart,
+    "dotreplace": dot_replace_for_mxid,
+}
+
+
+@attr.s
+class SamlConfig(object):
+    mxid_source_attribute = attr.ib()
+    mxid_mapper = attr.ib()
+
+
+class DefaultSamlMappingProvider(object):
+    __version__ = "0.0.1"
+
+    def __init__(self, parsed_config: SamlConfig):
+        """The default SAML user mapping provider
+
+        Args:
+            parsed_config: Module configuration
+        """
+        self._mxid_source_attribute = parsed_config.mxid_source_attribute
+        self._mxid_mapper = parsed_config.mxid_mapper
+
+    def saml_response_to_user_attributes(
+        self, saml_response: saml2.response.AuthnResponse, failures: int = 0,
+    ) -> dict:
+        """Maps some text from a SAML response to attributes of a new user
+
+        Args:
+            saml_response: A SAML auth response object
+
+            failures: How many times a call to this function with this
+                saml_response has resulted in a failure
+
+        Returns:
+            dict: A dict containing new user attributes. Possible keys:
+                * mxid_localpart (str): Required. The localpart of the user's mxid
+                * displayname (str): The displayname of the user
+        """
+        try:
+            mxid_source = saml_response.ava[self._mxid_source_attribute][0]
+        except KeyError:
+            logger.warning(
+                "SAML2 response lacks a '%s' attestation", self._mxid_source_attribute,
+            )
+            raise SynapseError(
+                400, "%s not in SAML2 response" % (self._mxid_source_attribute,)
+            )
+
+        # Use the configured mapper for this mxid_source
+        base_mxid_localpart = self._mxid_mapper(mxid_source)
+
+        # Append suffix integer if last call to this function failed to produce
+        # a usable mxid
+        localpart = base_mxid_localpart + (str(failures) if failures else "")
+
+        # Retrieve the display name from the saml response
+        # If displayname is None, the mxid_localpart will be used instead
+        displayname = saml_response.ava.get("displayName", [None])[0]
+
+        return {
+            "mxid_localpart": localpart,
+            "displayname": displayname,
+        }
+
+    @staticmethod
+    def parse_config(config: dict) -> SamlConfig:
+        """Parse the dict provided by the homeserver's config
+        Args:
+            config: A dictionary containing configuration options for this provider
+        Returns:
+            SamlConfig: A custom config object for this module
+        """
+        # Parse config options and use defaults where necessary
+        mxid_source_attribute = config.get("mxid_source_attribute", "uid")
+        mapping_type = config.get("mxid_mapping", "hexencode")
+
+        # Retrieve the associating mapping function
+        try:
+            mxid_mapper = MXID_MAPPER_MAP[mapping_type]
+        except KeyError:
+            raise ConfigError(
+                "saml2_config.user_mapping_provider.config: '%s' is not a valid "
+                "mxid_mapping value" % (mapping_type,)
+            )
+
+        return SamlConfig(mxid_source_attribute, mxid_mapper)
+
+    @staticmethod
+    def get_saml_attributes(config: SamlConfig) -> Tuple[set, set]:
+        """Returns the required attributes of a SAML
+
+        Args:
+            config: A SamlConfig object containing configuration params for this provider
+
+        Returns:
+            tuple[set,set]: The first set equates to the saml auth response
+                attributes that are required for the module to function, whereas the
+                second set consists of those attributes which can be used if
+                available, but are not necessary
+        """
+        return {"uid", config.mxid_source_attribute}, {"displayName"}

synapse/handlers/search.py

@@ -21,7 +21,7 @@ from unpaddedbase64 import decode_base64, encode_base64
 from twisted.internet import defer
 
 from synapse.api.constants import EventTypes, Membership
-from synapse.api.errors import SynapseError
+from synapse.api.errors import NotFoundError, SynapseError
 from synapse.api.filtering import Filter
 from synapse.storage.state import StateFilter
 from synapse.visibility import filter_events_for_client
@@ -37,6 +37,7 @@ class SearchHandler(BaseHandler):
         self._event_serializer = hs.get_event_client_serializer()
         self.storage = hs.get_storage()
         self.state_store = self.storage.state
+        self.auth = hs.get_auth()
 
     @defer.inlineCallbacks
     def get_old_rooms_from_upgraded_room(self, room_id):
@@ -53,23 +54,38 @@ class SearchHandler(BaseHandler):
             room_id (str): id of the room to search through.
 
         Returns:
-            Deferred[iterable[unicode]]: predecessor room ids
+            Deferred[iterable[str]]: predecessor room ids
         """
 
         historical_room_ids = []
 
-        while True:
-            predecessor = yield self.store.get_room_predecessor(room_id)
+        # The initial room must have been known for us to get this far
+        predecessor = yield self.store.get_room_predecessor(room_id)
 
-            # If no predecessor, assume we've hit a dead end
+        while True:
             if not predecessor:
+                # We have reached the end of the chain of predecessors
                 break
 
-            # Add predecessor's room ID
-            historical_room_ids.append(predecessor["room_id"])
+            if not isinstance(predecessor.get("room_id"), str):
+                # This predecessor object is malformed. Exit here
+                break
 
-            # Scan through the old room for further predecessors
-            room_id = predecessor["room_id"]
+            predecessor_room_id = predecessor["room_id"]
+
+            # Don't add it to the list until we have checked that we are in the room
+            try:
+                next_predecessor_room = yield self.store.get_room_predecessor(
+                    predecessor_room_id
+                )
+            except NotFoundError:
+                # The predecessor is not a known room, so we are done here
+                break
+
+            historical_room_ids.append(predecessor_room_id)
+
+            # And repeat
+            predecessor = next_predecessor_room
 
         return historical_room_ids
 

synapse/handlers/typing.py

@@ -317,6 +317,3 @@ class TypingNotificationEventSource(object):
 
     def get_current_key(self):
         return self.get_typing_handler()._latest_room_serial
-
-    def get_pagination_rows(self, user, pagination_config, key):
-        return [], pagination_config.from_key

synapse/http/site.py

@@ -47,9 +47,9 @@ class SynapseRequest(Request):
         logcontext(LoggingContext) : the log context for this request
     """
 
-    def __init__(self, site, channel, *args, **kw):
+    def __init__(self, channel, *args, **kw):
         Request.__init__(self, channel, *args, **kw)
-        self.site = site
+        self.site = channel.site
         self._channel = channel  # this is used by the tests
         self.authenticated_entity = None
         self.start_time = 0
@@ -331,18 +331,6 @@ class XForwardedForRequest(SynapseRequest):
         )
 
 
-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
@@ -364,7 +352,7 @@ class SynapseSite(Site):
         self.site_tag = site_tag
 
         proxied = config.get("x_forwarded", False)
-        self.requestFactory = SynapseRequestFactory(self, proxied)
+        self.requestFactory = XForwardedForRequest if proxied else SynapseRequest
         self.access_logger = logging.getLogger(logger_name)
         self.server_version_string = server_version_string.encode("ascii")
 

synapse/logging/_terse_json.py

@@ -171,7 +171,7 @@ class LogProducer(object):
 
     def stopProducing(self):
         self._paused = True
-        self._buffer = None
+        self._buffer = deque()
 
     def resumeProducing(self):
         self._paused = False

synapse/logging/context.py

@@ -23,6 +23,7 @@ them.
 See doc/log_contexts.rst for details on how this works.
 """
 
+import inspect
 import logging
 import threading
 import types
@@ -404,6 +405,9 @@ class LoggingContext(object):
         """
         current = get_thread_resource_usage()
 
+        # Indicate to mypy that we know that self.usage_start is None.
+        assert self.usage_start is not None
+
         utime_delta = current.ru_utime - self.usage_start.ru_utime
         stime_delta = current.ru_stime - self.usage_start.ru_stime
 
@@ -612,7 +616,8 @@ def run_in_background(f, *args, **kwargs):
 
 
 def make_deferred_yieldable(deferred):
-    """Given a deferred, make it follow the Synapse logcontext rules:
+    """Given a deferred (or coroutine), make it follow the Synapse logcontext
+    rules:
 
     If the deferred has completed (or is not actually a Deferred), essentially
     does nothing (just returns another completed deferred with the
@@ -624,6 +629,13 @@ def make_deferred_yieldable(deferred):
 
     (This is more-or-less the opposite operation to run_in_background.)
     """
+    if inspect.isawaitable(deferred):
+        # If we're given a coroutine we convert it to a deferred so that we
+        # run it and find out if it immediately finishes, it it does then we
+        # don't need to fiddle with log contexts at all and can return
+        # immediately.
+        deferred = defer.ensureDeferred(deferred)
+
     if not isinstance(deferred, defer.Deferred):
         return deferred
 

synapse/push/bulk_push_rule_evaluator.py

@@ -116,7 +116,7 @@ class BulkPushRuleEvaluator(object):
 
     @defer.inlineCallbacks
     def _get_power_levels_and_sender_level(self, event, context):
-        prev_state_ids = yield context.get_prev_state_ids(self.store)
+        prev_state_ids = yield context.get_prev_state_ids()
         pl_event_id = prev_state_ids.get(POWER_KEY)
         if pl_event_id:
             # fastpath: if there's a power level event, that's all we need, and
@@ -304,7 +304,7 @@ class RulesForRoom(object):
 
                 push_rules_delta_state_cache_metric.inc_hits()
             else:
-                current_state_ids = yield context.get_current_state_ids(self.store)
+                current_state_ids = yield context.get_current_state_ids()
                 push_rules_delta_state_cache_metric.inc_misses()
 
             push_rules_state_size_counter.inc(len(current_state_ids))

synapse/push/pusher.py

@@ -80,9 +80,11 @@ class PusherFactory(object):
         return EmailPusher(self.hs, pusherdict, mailer)
 
     def _app_name_from_pusherdict(self, pusherdict):
-        if "data" in pusherdict and "brand" in pusherdict["data"]:
-            app_name = pusherdict["data"]["brand"]
-        else:
-            app_name = self.config.email_app_name
+        data = pusherdict["data"]
 
-        return app_name
+        if isinstance(data, dict):
+            brand = data.get("brand")
+            if isinstance(brand, str):
+                return brand
+
+        return self.config.email_app_name

synapse/push/pusherpool.py

@@ -232,7 +232,6 @@ class PusherPool:
             Deferred
         """
         pushers = yield self.store.get_all_pushers()
-        logger.info("Starting %d pushers", len(pushers))
 
         # Stagger starting up the pushers so we don't completely drown the
         # process on start up.
@@ -245,7 +244,7 @@ class PusherPool:
         """Start the given pusher
 
         Args:
-            pusherdict (dict):
+            pusherdict (dict): dict with the values pulled from the db table
 
         Returns:
             Deferred[EmailPusher|HttpPusher]
@@ -254,7 +253,8 @@ class PusherPool:
             p = self.pusher_factory.create_pusher(pusherdict)
         except PusherConfigException as e:
             logger.warning(
-                "Pusher incorrectly configured user=%s, appid=%s, pushkey=%s: %s",
+                "Pusher incorrectly configured id=%i, user=%s, appid=%s, pushkey=%s: %s",
+                pusherdict["id"],
                 pusherdict.get("user_name"),
                 pusherdict.get("app_id"),
                 pusherdict.get("pushkey"),
@@ -262,7 +262,9 @@ class PusherPool:
             )
             return
         except Exception:
-            logger.exception("Couldn't start a pusher: caught Exception")
+            logger.exception(
+                "Couldn't start pusher id %i: caught Exception", pusherdict["id"],
+            )
             return
 
         if not p:

synapse/replication/http/federation.py

@@ -51,6 +51,7 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint):
         super(ReplicationFederationSendEventsRestServlet, self).__init__(hs)
 
         self.store = hs.get_datastore()
+        self.storage = hs.get_storage()
         self.clock = hs.get_clock()
         self.federation_handler = hs.get_handlers().federation_handler
 
@@ -100,7 +101,9 @@ class ReplicationFederationSendEventsRestServlet(ReplicationEndpoint):
                 EventType = event_type_from_format_version(format_ver)
                 event = EventType(event_dict, internal_metadata, rejected_reason)
 
-                context = EventContext.deserialize(self.store, event_payload["context"])
+                context = EventContext.deserialize(
+                    self.storage, event_payload["context"]
+                )
 
                 event_and_contexts.append((event, context))
 

synapse/replication/http/send_event.py

@@ -54,6 +54,7 @@ class ReplicationSendEventRestServlet(ReplicationEndpoint):
 
         self.event_creation_handler = hs.get_event_creation_handler()
         self.store = hs.get_datastore()
+        self.storage = hs.get_storage()
         self.clock = hs.get_clock()
 
     @staticmethod
@@ -100,7 +101,7 @@ class ReplicationSendEventRestServlet(ReplicationEndpoint):
             event = EventType(event_dict, internal_metadata, rejected_reason)
 
             requester = Requester.deserialize(self.store, content["requester"])
-            context = EventContext.deserialize(self.store, content["context"])
+            context = EventContext.deserialize(self.storage, content["context"])
 
             ratelimit = content["ratelimit"]
             extra_users = [UserID.from_string(u) for u in content["extra_users"]]

synapse/replication/tcp/client.py

@@ -46,7 +46,8 @@ class ReplicationClientFactory(ReconnectingClientFactory):
     is required.
     """
 
-    maxDelay = 30  # Try at least once every N seconds
+    initialDelay = 0.1
+    maxDelay = 1  # Try at least once every N seconds
 
     def __init__(self, hs, client_name, handler: AbstractReplicationClientHandler):
         self.client_name = client_name

synapse/rest/client/v1/pusher.py

@@ -28,6 +28,17 @@ from synapse.rest.client.v2_alpha._base import client_patterns
 
 logger = logging.getLogger(__name__)
 
+ALLOWED_KEYS = {
+    "app_display_name",
+    "app_id",
+    "data",
+    "device_display_name",
+    "kind",
+    "lang",
+    "profile_tag",
+    "pushkey",
+}
+
 
 class PushersRestServlet(RestServlet):
     PATTERNS = client_patterns("/pushers$", v1=True)
@@ -43,23 +54,11 @@ class PushersRestServlet(RestServlet):
 
         pushers = await self.hs.get_datastore().get_pushers_by_user_id(user.to_string())
 
-        allowed_keys = [
-            "app_display_name",
-            "app_id",
-            "data",
-            "device_display_name",
-            "kind",
-            "lang",
-            "profile_tag",
-            "pushkey",
-        ]
+        filtered_pushers = list(
+            {k: v for k, v in p.items() if k in ALLOWED_KEYS} for p in pushers
+        )
 
-        for p in pushers:
-            for k, v in list(p.items()):
-                if k not in allowed_keys:
-                    del p[k]
-
-        return 200, {"pushers": pushers}
+        return 200, {"pushers": filtered_pushers}
 
     def on_OPTIONS(self, _):
         return 200, {}

synapse/rest/media/v0/content_repository.py

@@ -1,103 +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.
-
-import base64
-import logging
-import os
-import re
-
-from canonicaljson import json
-
-from twisted.protocols.basic import FileSender
-from twisted.web import resource, server
-
-from synapse.api.errors import Codes, cs_error
-from synapse.http.server import finish_request, respond_with_json_bytes
-
-logger = logging.getLogger(__name__)
-
-
-class ContentRepoResource(resource.Resource):
-    """Provides file uploading and downloading.
-
-    Uploads are POSTed to wherever this Resource is linked to. This resource
-    returns a "content token" which can be used to GET this content again. The
-    token is typically a path, but it may not be. Tokens can expire, be
-    one-time uses, etc.
-
-    In this case, the token is a path to the file and contains 3 interesting
-    sections:
-        - User ID base64d (for namespacing content to each user)
-        - random 24 char string
-        - Content type base64d (so we can return it when clients GET it)
-
-    """
-
-    isLeaf = True
-
-    def __init__(self, hs, directory):
-        resource.Resource.__init__(self)
-        self.hs = hs
-        self.directory = directory
-
-    def render_GET(self, request):
-        # no auth here on purpose, to allow anyone to view, even across home
-        # servers.
-
-        # TODO: A little crude here, we could do this better.
-        filename = request.path.decode("ascii").split("/")[-1]
-        # be paranoid
-        filename = re.sub("[^0-9A-z.-_]", "", filename)
-
-        file_path = self.directory + "/" + filename
-
-        logger.debug("Searching for %s", file_path)
-
-        if os.path.isfile(file_path):
-            # filename has the content type
-            base64_contentype = filename.split(".")[1]
-            content_type = base64.urlsafe_b64decode(base64_contentype)
-            logger.info("Sending file %s", file_path)
-            f = open(file_path, "rb")
-            request.setHeader("Content-Type", content_type)
-
-            # cache for at least a day.
-            # XXX: we might want to turn this off for data we don't want to
-            # recommend caching as it's sensitive or private - or at least
-            # select private. don't bother setting Expires as all our matrix
-            # clients are smart enough to be happy with Cache-Control (right?)
-            request.setHeader(b"Cache-Control", b"public,max-age=86400,s-maxage=86400")
-
-            d = FileSender().beginFileTransfer(f, request)
-
-            # after the file has been sent, clean up and finish the request
-            def cbFinished(ignored):
-                f.close()
-                finish_request(request)
-
-            d.addCallback(cbFinished)
-        else:
-            respond_with_json_bytes(
-                request,
-                404,
-                json.dumps(cs_error("Not found", code=Codes.NOT_FOUND)),
-                send_cors=True,
-            )
-
-        return server.NOT_DONE_YET
-
-    def render_OPTIONS(self, request):
-        respond_with_json_bytes(request, 200, {}, send_cors=True)
-        return server.NOT_DONE_YET

synapse/server.py

@@ -25,7 +25,6 @@ import abc
 import logging
 import os
 
-from twisted.enterprise import adbapi
 from twisted.mail.smtp import sendmail
 from twisted.web.client import BrowserLikePolicyForHTTPS
 
@@ -34,6 +33,7 @@ from synapse.api.filtering import Filtering
 from synapse.api.ratelimiting import Ratelimiter
 from synapse.appservice.api import ApplicationServiceApi
 from synapse.appservice.scheduler import ApplicationServiceScheduler
+from synapse.config.homeserver import HomeServerConfig
 from synapse.crypto import context_factory
 from synapse.crypto.keyring import Keyring
 from synapse.events.builder import EventBuilderFactory
@@ -132,7 +132,6 @@ class HomeServer(object):
 
     DEPENDENCIES = [
         "http_client",
-        "db_pool",
         "federation_client",
         "federation_server",
         "handlers",
@@ -209,16 +208,18 @@ class HomeServer(object):
     # instantiated during setup() for future return by get_datastore()
     DATASTORE_CLASS = abc.abstractproperty()
 
-    def __init__(self, hostname, reactor=None, **kwargs):
+    def __init__(self, hostname: str, config: HomeServerConfig, reactor=None, **kwargs):
         """
         Args:
             hostname : The hostname for the server.
+            config: The full config for the homeserver.
         """
         if not reactor:
             from twisted.internet import reactor
 
         self._reactor = reactor
         self.hostname = hostname
+        self.config = config
         self._building = {}
         self._listening_services = []
         self.start_time = None
@@ -237,10 +238,8 @@ class HomeServer(object):
 
     def setup(self):
         logger.info("Setting up.")
-        with self.get_db_conn() as conn:
-            self.datastores = DataStores(self.DATASTORE_CLASS, conn, self)
-            conn.commit()
         self.start_time = int(self.get_clock().time())
+        self.datastores = DataStores(self.DATASTORE_CLASS, self)
         logger.info("Finished setting up.")
 
     def setup_master(self):
@@ -274,6 +273,9 @@ class HomeServer(object):
     def get_datastore(self):
         return self.datastores.main
 
+    def get_datastores(self):
+        return self.datastores
+
     def get_config(self):
         return self.config
 
@@ -423,31 +425,6 @@ class HomeServer(object):
         )
         return MatrixFederationHttpClient(self, tls_client_options_factory)
 
-    def build_db_pool(self):
-        name = self.db_config["name"]
-
-        return adbapi.ConnectionPool(
-            name, cp_reactor=self.get_reactor(), **self.db_config.get("args", {})
-        )
-
-    def get_db_conn(self, run_new_connection=True):
-        """Makes a new connection to the database, skipping the db pool
-
-        Returns:
-            Connection: a connection object implementing the PEP-249 spec
-        """
-        # 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 build_media_repository_resource(self):
         # build the media repo resource. This indirects through the HomeServer
         # to ensure that we only have a single instance of

synapse/state/__init__.py

@@ -32,6 +32,7 @@ from synapse.events import EventBase
 from synapse.events.snapshot import EventContext
 from synapse.logging.utils import log_function
 from synapse.state import v1, v2
+from synapse.storage.data_stores.main.events_worker import EventRedactBehaviour
 from synapse.util.async_helpers import Linearizer
 from synapse.util.caches import get_cache_factor_for
 from synapse.util.caches.expiringcache import ExpiringCache
@@ -655,7 +656,7 @@ class StateResolutionStore(object):
 
         return self.store.get_events(
             event_ids,
-            check_redacted=False,
+            redact_behaviour=EventRedactBehaviour.AS_IS,
             get_prev_content=False,
             allow_rejected=allow_rejected,
         )

synapse/state/v2.py

@@ -183,16 +183,20 @@ def _get_power_level_for_sender(room_id, event_id, event_map, state_res_store):
 
     pl = None
     for aid in event.auth_event_ids():
-        aev = yield _get_event(room_id, aid, event_map, state_res_store)
-        if (aev.type, aev.state_key) == (EventTypes.PowerLevels, ""):
+        aev = yield _get_event(
+            room_id, aid, event_map, state_res_store, allow_none=True
+        )
+        if aev and (aev.type, aev.state_key) == (EventTypes.PowerLevels, ""):
             pl = aev
             break
 
     if pl is None:
         # Couldn't find power level. Check if they're the creator of the room
         for aid in event.auth_event_ids():
-            aev = yield _get_event(room_id, aid, event_map, state_res_store)
-            if (aev.type, aev.state_key) == (EventTypes.Create, ""):
+            aev = yield _get_event(
+                room_id, aid, event_map, state_res_store, allow_none=True
+            )
+            if aev and (aev.type, aev.state_key) == (EventTypes.Create, ""):
                 if aev.content.get("creator") == event.sender:
                     return 100
                 break
@@ -403,10 +407,17 @@ def _iterative_auth_checks(
 
         auth_events = {}
         for aid in event.auth_event_ids():
-            ev = yield _get_event(room_id, aid, event_map, state_res_store)
+            ev = yield _get_event(
+                room_id, aid, event_map, state_res_store, allow_none=True
+            )
 
-            if ev.rejected_reason is None:
-                auth_events[(ev.type, ev.state_key)] = ev
+            if not ev:
+                logger.warning(
+                    "auth_event id %s for event %s is missing", aid, event_id
+                )
+            else:
+                if ev.rejected_reason is None:
+                    auth_events[(ev.type, ev.state_key)] = ev
 
         for key in event_auth.auth_types_for_event(event):
             if key in resolved_state:
@@ -457,8 +468,10 @@ def _mainline_sort(
         auth_events = pl_ev.auth_event_ids()
         pl = None
         for aid in auth_events:
-            ev = yield _get_event(room_id, aid, event_map, state_res_store)
-            if (ev.type, ev.state_key) == (EventTypes.PowerLevels, ""):
+            ev = yield _get_event(
+                room_id, aid, event_map, state_res_store, allow_none=True
+            )
+            if ev and (ev.type, ev.state_key) == (EventTypes.PowerLevels, ""):
                 pl = aid
                 break
 
@@ -506,8 +519,10 @@ def _get_mainline_depth_for_event(event, mainline_map, event_map, state_res_stor
         event = None
 
         for aid in auth_events:
-            aev = yield _get_event(room_id, aid, event_map, state_res_store)
-            if (aev.type, aev.state_key) == (EventTypes.PowerLevels, ""):
+            aev = yield _get_event(
+                room_id, aid, event_map, state_res_store, allow_none=True
+            )
+            if aev and (aev.type, aev.state_key) == (EventTypes.PowerLevels, ""):
                 event = aev
                 break
 
@@ -516,7 +531,7 @@ def _get_mainline_depth_for_event(event, mainline_map, event_map, state_res_stor
 
 
 @defer.inlineCallbacks
-def _get_event(room_id, event_id, event_map, state_res_store):
+def _get_event(room_id, event_id, event_map, state_res_store, allow_none=False):
     """Helper function to look up event in event_map, falling back to looking
     it up in the store
 
@@ -525,15 +540,22 @@ def _get_event(room_id, event_id, event_map, state_res_store):
         event_id (str)
         event_map (dict[str,FrozenEvent])
         state_res_store (StateResolutionStore)
+        allow_none (bool): if the event is not found, return None rather than raising
+            an exception
 
     Returns:
-        Deferred[FrozenEvent]
+        Deferred[Optional[FrozenEvent]]
     """
     if event_id not in event_map:
         events = yield state_res_store.get_events([event_id], allow_rejected=True)
         event_map.update(events)
-    event = event_map[event_id]
-    assert event is not None
+    event = event_map.get(event_id)
+
+    if event is None:
+        if allow_none:
+            return None
+        raise Exception("Unknown event %s" % (event_id,))
+
     if event.room_id != room_id:
         raise Exception(
             "In state res for room %s, event %s is in %s"

synapse/storage/_base.py

@@ -16,6 +16,7 @@
 # limitations under the License.
 import logging
 import random
+from abc import ABCMeta
 
 from six import PY2
 from six.moves import builtins
@@ -30,7 +31,8 @@ from synapse.types import get_domain_from_id
 logger = logging.getLogger(__name__)
 
 
-class SQLBaseStore(object):
+# some of our subclasses have abstract methods, so we use the ABCMeta metaclass.
+class SQLBaseStore(metaclass=ABCMeta):
     """Base class for data stores that holds helper functions.
 
     Note that multiple instances of this class will exist as there will be one
@@ -40,7 +42,7 @@ class SQLBaseStore(object):
     def __init__(self, database: Database, db_conn, hs):
         self.hs = hs
         self._clock = hs.get_clock()
-        self.database_engine = hs.database_engine
+        self.database_engine = database.engine
         self.db = database
         self.rand = random.SystemRandom()
 

synapse/storage/background_updates.py

@@ -14,6 +14,7 @@
 # limitations under the License.
 
 import logging
+from typing import Optional
 
 from canonicaljson import json
 
@@ -97,15 +98,14 @@ class BackgroundUpdater(object):
     def start_doing_background_updates(self):
         run_as_background_process("background_updates", self.run_background_updates)
 
-    @defer.inlineCallbacks
-    def run_background_updates(self, sleep=True):
+    async def run_background_updates(self, sleep=True):
         logger.info("Starting background schema updates")
         while True:
             if sleep:
-                yield self._clock.sleep(self.BACKGROUND_UPDATE_INTERVAL_MS / 1000.0)
+                await self._clock.sleep(self.BACKGROUND_UPDATE_INTERVAL_MS / 1000.0)
 
             try:
-                result = yield self.do_next_background_update(
+                result = await self.do_next_background_update(
                     self.BACKGROUND_UPDATE_DURATION_MS
                 )
             except Exception:
@@ -170,20 +170,21 @@ class BackgroundUpdater(object):
 
         return not update_exists
 
-    @defer.inlineCallbacks
-    def do_next_background_update(self, desired_duration_ms):
+    async def do_next_background_update(
+        self, desired_duration_ms: float
+    ) -> Optional[int]:
         """Does some amount of work on the next queued background update
 
+        Returns once some amount of work is done.
+
         Args:
             desired_duration_ms(float): How long we want to spend
                 updating.
         Returns:
-            A deferred that completes once some amount of work is done.
-            The deferred will have a value of None if there is currently
-            no more work to do.
+            None if there is no more work to do, otherwise an int
         """
         if not self._background_update_queue:
-            updates = yield self.db.simple_select_list(
+            updates = await self.db.simple_select_list(
                 "background_updates",
                 keyvalues=None,
                 retcols=("update_name", "depends_on"),
@@ -201,11 +202,12 @@ class BackgroundUpdater(object):
         update_name = self._background_update_queue.pop(0)
         self._background_update_queue.append(update_name)
 
-        res = yield self._do_background_update(update_name, desired_duration_ms)
+        res = await self._do_background_update(update_name, desired_duration_ms)
         return res
 
-    @defer.inlineCallbacks
-    def _do_background_update(self, update_name, desired_duration_ms):
+    async def _do_background_update(
+        self, update_name: str, desired_duration_ms: float
+    ) -> int:
         logger.info("Starting update batch on background update '%s'", update_name)
 
         update_handler = self._background_update_handlers[update_name]
@@ -225,7 +227,7 @@ class BackgroundUpdater(object):
         else:
             batch_size = self.DEFAULT_BACKGROUND_BATCH_SIZE
 
-        progress_json = yield self.db.simple_select_one_onecol(
+        progress_json = await self.db.simple_select_one_onecol(
             "background_updates",
             keyvalues={"update_name": update_name},
             retcol="progress_json",
@@ -234,7 +236,7 @@ class BackgroundUpdater(object):
         progress = json.loads(progress_json)
 
         time_start = self._clock.time_msec()
-        items_updated = yield update_handler(progress, batch_size)
+        items_updated = await update_handler(progress, batch_size)
         time_stop = self._clock.time_msec()
 
         duration_ms = time_stop - time_start
@@ -263,7 +265,9 @@ class BackgroundUpdater(object):
         * A dict of the current progress
         * An integer count of the number of items to update in this batch.
 
-        The handler should return a deferred integer count of items updated.
+        The handler should return a deferred or coroutine which returns an integer count
+        of items updated.
+
         The handler is responsible for updating the progress of the update.
 
         Args:
@@ -432,6 +436,21 @@ class BackgroundUpdater(object):
             "background_updates", keyvalues={"update_name": update_name}
         )
 
+    def _background_update_progress(self, update_name: str, progress: dict):
+        """Update the progress of a background update
+
+        Args:
+            update_name: The name of the background update task
+            progress: The progress of the update.
+        """
+
+        return self.db.runInteraction(
+            "background_update_progress",
+            self._background_update_progress_txn,
+            update_name,
+            progress,
+        )
+
     def _background_update_progress_txn(self, txn, update_name, progress):
         """Update the progress of a background update
 

synapse/storage/data_stores/__init__.py

@@ -13,24 +13,76 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from synapse.storage.database import Database
+import logging
+
+from synapse.storage.data_stores.state import StateGroupDataStore
+from synapse.storage.database import Database, make_conn
+from synapse.storage.engines import create_engine
 from synapse.storage.prepare_database import prepare_database
 
+logger = logging.getLogger(__name__)
+
 
 class DataStores(object):
     """The various data stores.
 
     These are low level interfaces to physical databases.
+
+    Attributes:
+        main (DataStore)
     """
 
-    def __init__(self, main_store_class, db_conn, hs):
+    def __init__(self, main_store_class, hs):
         # Note we pass in the main store class here as workers use a different main
         # store.
-        database = Database(hs)
 
-        # Check that db is correctly configured.
-        database.engine.check_database(db_conn.cursor())
+        self.databases = []
+        self.main = None
+        self.state = None
 
-        prepare_database(db_conn, database.engine, config=hs.config)
+        for database_config in hs.config.database.databases:
+            db_name = database_config.name
+            engine = create_engine(database_config.config)
 
-        self.main = main_store_class(database, db_conn, hs)
+            with make_conn(database_config, engine) as db_conn:
+                logger.info("Preparing database %r...", db_name)
+
+                engine.check_database(db_conn.cursor())
+                prepare_database(
+                    db_conn, engine, hs.config, data_stores=database_config.data_stores,
+                )
+
+                database = Database(hs, database_config, engine)
+
+                if "main" in database_config.data_stores:
+                    logger.info("Starting 'main' data store")
+
+                    # Sanity check we don't try and configure the main store on
+                    # multiple databases.
+                    if self.main:
+                        raise Exception("'main' data store already configured")
+
+                    self.main = main_store_class(database, db_conn, hs)
+
+                if "state" in database_config.data_stores:
+                    logger.info("Starting 'state' data store")
+
+                    # Sanity check we don't try and configure the state store on
+                    # multiple databases.
+                    if self.state:
+                        raise Exception("'state' data store already configured")
+
+                    self.state = StateGroupDataStore(database, db_conn, hs)
+
+                db_conn.commit()
+
+                self.databases.append(database)
+
+                logger.info("Database %r prepared", db_name)
+
+        # Sanity check that we have actually configured all the required stores.
+        if not self.main:
+            raise Exception("No 'main' data store configured")
+
+        if not self.state:
+            raise Exception("No 'main' data store configured")

synapse/storage/data_stores/main/__init__.py

@@ -526,9 +526,9 @@ class DataStore(
 
         attr_filter = {}
         if not guests:
-            attr_filter["is_guest"] = False
+            attr_filter["is_guest"] = 0
         if not deactivated:
-            attr_filter["deactivated"] = False
+            attr_filter["deactivated"] = 0
 
         return self.db.simple_select_list_paginate(
             desc="get_users_paginate",

synapse/storage/data_stores/main/client_ips.py

@@ -412,7 +412,7 @@ class ClientIpStore(ClientIpBackgroundUpdateStore):
     def _update_client_ips_batch(self):
 
         # If the DB pool has already terminated, don't try updating
-        if not self.hs.get_db_pool().running:
+        if not self.db.is_running():
             return
 
         to_update = self._batch_row_update
@@ -451,16 +451,18 @@ class ClientIpStore(ClientIpBackgroundUpdateStore):
                 # Technically an access token might not be associated with
                 # a device so we need to check.
                 if device_id:
-                    self.db.simple_upsert_txn(
+                    # this is always an update rather than an upsert: the row should
+                    # already exist, and if it doesn't, that may be because it has been
+                    # deleted, and we don't want to re-create it.
+                    self.db.simple_update_txn(
                         txn,
                         table="devices",
                         keyvalues={"user_id": user_id, "device_id": device_id},
-                        values={
+                        updatevalues={
                             "user_agent": user_agent,
                             "last_seen": last_seen,
                             "ip": ip,
                         },
-                        lock=False,
                     )
             except Exception as e:
                 # Failed to upsert, log and continue

synapse/storage/data_stores/main/deviceinbox.py

@@ -358,21 +358,8 @@ class DeviceInboxStore(DeviceInboxWorkerStore, DeviceInboxBackgroundUpdateStore)
     def _add_messages_to_local_device_inbox_txn(
         self, txn, stream_id, messages_by_user_then_device
     ):
-        # Compatible method of performing an upsert
-        sql = "SELECT stream_id FROM device_max_stream_id"
-
-        txn.execute(sql)
-        rows = txn.fetchone()
-        if rows:
-            db_stream_id = rows[0]
-            if db_stream_id < stream_id:
-                # Insert the new stream_id
-                sql = "UPDATE device_max_stream_id SET stream_id = ?"
-        else:
-            # No rows, perform an insert
-            sql = "INSERT INTO device_max_stream_id (stream_id) VALUES (?)"
-
-        txn.execute(sql, (stream_id,))
+        sql = "UPDATE device_max_stream_id" " SET stream_id = ?" " WHERE stream_id < ?"
+        txn.execute(sql, (stream_id, stream_id))
 
         local_by_user_then_device = {}
         for user_id, messages_by_device in messages_by_user_then_device.items():

synapse/storage/data_stores/main/end_to_end_keys.py

@@ -14,15 +14,18 @@
 # 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 typing import Dict, List
+
 from six import iteritems
 
 from canonicaljson import encode_canonical_json, json
 
+from twisted.enterprise.adbapi import Connection
 from twisted.internet import defer
 
 from synapse.logging.opentracing import log_kv, set_tag, trace
 from synapse.storage._base import SQLBaseStore, db_to_json
-from synapse.util.caches.descriptors import cached
+from synapse.util.caches.descriptors import cached, cachedList
 
 
 class EndToEndKeyWorkerStore(SQLBaseStore):
@@ -271,7 +274,7 @@ class EndToEndKeyWorkerStore(SQLBaseStore):
         Args:
             txn (twisted.enterprise.adbapi.Connection): db connection
             user_id (str): the user whose key is being requested
-            key_type (str): the type of key that is being set: either 'master'
+            key_type (str): the type of key that is being requested: either 'master'
                 for a master key, 'self_signing' for a self-signing key, or
                 'user_signing' for a user-signing key
             from_user_id (str): if specified, signatures made by this user on
@@ -316,8 +319,10 @@ class EndToEndKeyWorkerStore(SQLBaseStore):
         """Returns a user's cross-signing key.
 
         Args:
-            user_id (str): the user whose self-signing key is being requested
-            key_type (str): the type of cross-signing key to get
+            user_id (str): the user whose key is being requested
+            key_type (str): the type of key that is being requested: either 'master'
+                for a master key, 'self_signing' for a self-signing key, or
+                'user_signing' for a user-signing key
             from_user_id (str): if specified, signatures made by this user on
                 the self-signing key will be included in the result
 
@@ -332,6 +337,206 @@ class EndToEndKeyWorkerStore(SQLBaseStore):
             from_user_id,
         )
 
+    @cached(num_args=1)
+    def _get_bare_e2e_cross_signing_keys(self, user_id):
+        """Dummy function.  Only used to make a cache for
+        _get_bare_e2e_cross_signing_keys_bulk.
+        """
+        raise NotImplementedError()
+
+    @cachedList(
+        cached_method_name="_get_bare_e2e_cross_signing_keys",
+        list_name="user_ids",
+        num_args=1,
+    )
+    def _get_bare_e2e_cross_signing_keys_bulk(
+        self, user_ids: List[str]
+    ) -> Dict[str, Dict[str, dict]]:
+        """Returns the cross-signing keys for a set of users.  The output of this
+        function should be passed to _get_e2e_cross_signing_signatures_txn if
+        the signatures for the calling user need to be fetched.
+
+        Args:
+            user_ids (list[str]): the users whose keys are being requested
+
+        Returns:
+            dict[str, dict[str, dict]]: mapping from user ID to key type to key
+                data.  If a user's cross-signing keys were not found, either
+                their user ID will not be in the dict, or their user ID will map
+                to None.
+
+        """
+        return self.db.runInteraction(
+            "get_bare_e2e_cross_signing_keys_bulk",
+            self._get_bare_e2e_cross_signing_keys_bulk_txn,
+            user_ids,
+        )
+
+    def _get_bare_e2e_cross_signing_keys_bulk_txn(
+        self, txn: Connection, user_ids: List[str],
+    ) -> Dict[str, Dict[str, dict]]:
+        """Returns the cross-signing keys for a set of users.  The output of this
+        function should be passed to _get_e2e_cross_signing_signatures_txn if
+        the signatures for the calling user need to be fetched.
+
+        Args:
+            txn (twisted.enterprise.adbapi.Connection): db connection
+            user_ids (list[str]): the users whose keys are being requested
+
+        Returns:
+            dict[str, dict[str, dict]]: mapping from user ID to key type to key
+                data.  If a user's cross-signing keys were not found, their user
+                ID will not be in the dict.
+
+        """
+        result = {}
+
+        batch_size = 100
+        chunks = [
+            user_ids[i : i + batch_size] for i in range(0, len(user_ids), batch_size)
+        ]
+        for user_chunk in chunks:
+            sql = """
+                SELECT k.user_id, k.keytype, k.keydata, k.stream_id
+                  FROM e2e_cross_signing_keys k
+                  INNER JOIN (SELECT user_id, keytype, MAX(stream_id) AS stream_id
+                                FROM e2e_cross_signing_keys
+                               GROUP BY user_id, keytype) s
+                 USING (user_id, stream_id, keytype)
+                 WHERE k.user_id IN (%s)
+            """ % (
+                ",".join("?" for u in user_chunk),
+            )
+            query_params = []
+            query_params.extend(user_chunk)
+
+            txn.execute(sql, query_params)
+            rows = self.db.cursor_to_dict(txn)
+
+            for row in rows:
+                user_id = row["user_id"]
+                key_type = row["keytype"]
+                key = json.loads(row["keydata"])
+                user_info = result.setdefault(user_id, {})
+                user_info[key_type] = key
+
+        return result
+
+    def _get_e2e_cross_signing_signatures_txn(
+        self, txn: Connection, keys: Dict[str, Dict[str, dict]], from_user_id: str,
+    ) -> Dict[str, Dict[str, dict]]:
+        """Returns the cross-signing signatures made by a user on a set of keys.
+
+        Args:
+            txn (twisted.enterprise.adbapi.Connection): db connection
+            keys (dict[str, dict[str, dict]]): a map of user ID to key type to
+                key data.  This dict will be modified to add signatures.
+            from_user_id (str): fetch the signatures made by this user
+
+        Returns:
+            dict[str, dict[str, dict]]: mapping from user ID to key type to key
+                data.  The return value will be the same as the keys argument,
+                with the modifications included.
+        """
+
+        # find out what cross-signing keys (a.k.a. devices) we need to get
+        # signatures for.  This is a map of (user_id, device_id) to key type
+        # (device_id is the key's public part).
+        devices = {}
+
+        for user_id, user_info in keys.items():
+            if user_info is None:
+                continue
+            for key_type, key in user_info.items():
+                device_id = None
+                for k in key["keys"].values():
+                    device_id = k
+                devices[(user_id, device_id)] = key_type
+
+        device_list = list(devices)
+
+        # split into batches
+        batch_size = 100
+        chunks = [
+            device_list[i : i + batch_size]
+            for i in range(0, len(device_list), batch_size)
+        ]
+        for user_chunk in chunks:
+            sql = """
+                SELECT target_user_id, target_device_id, key_id, signature
+                  FROM e2e_cross_signing_signatures
+                 WHERE user_id = ?
+                   AND (%s)
+            """ % (
+                " OR ".join(
+                    "(target_user_id = ? AND target_device_id = ?)" for d in devices
+                )
+            )
+            query_params = [from_user_id]
+            for item in devices:
+                # item is a (user_id, device_id) tuple
+                query_params.extend(item)
+
+            txn.execute(sql, query_params)
+            rows = self.db.cursor_to_dict(txn)
+
+            # and add the signatures to the appropriate keys
+            for row in rows:
+                key_id = row["key_id"]
+                target_user_id = row["target_user_id"]
+                target_device_id = row["target_device_id"]
+                key_type = devices[(target_user_id, target_device_id)]
+                # We need to copy everything, because the result may have come
+                # from the cache.  dict.copy only does a shallow copy, so we
+                # need to recursively copy the dicts that will be modified.
+                user_info = keys[target_user_id] = keys[target_user_id].copy()
+                target_user_key = user_info[key_type] = user_info[key_type].copy()
+                if "signatures" in target_user_key:
+                    signatures = target_user_key["signatures"] = target_user_key[
+                        "signatures"
+                    ].copy()
+                    if from_user_id in signatures:
+                        user_sigs = signatures[from_user_id] = signatures[from_user_id]
+                        user_sigs[key_id] = row["signature"]
+                    else:
+                        signatures[from_user_id] = {key_id: row["signature"]}
+                else:
+                    target_user_key["signatures"] = {
+                        from_user_id: {key_id: row["signature"]}
+                    }
+
+        return keys
+
+    @defer.inlineCallbacks
+    def get_e2e_cross_signing_keys_bulk(
+        self, user_ids: List[str], from_user_id: str = None
+    ) -> defer.Deferred:
+        """Returns the cross-signing keys for a set of users.
+
+        Args:
+            user_ids (list[str]): the users whose keys are being requested
+            from_user_id (str): if specified, signatures made by this user on
+                the self-signing keys will be included in the result
+
+        Returns:
+            Deferred[dict[str, dict[str, dict]]]: map of user ID to key type to
+                key data.  If a user's cross-signing keys were not found, either
+                their user ID will not be in the dict, or their user ID will map
+                to None.
+        """
+
+        result = yield self._get_bare_e2e_cross_signing_keys_bulk(user_ids)
+
+        if from_user_id:
+            result = yield self.db.runInteraction(
+                "get_e2e_cross_signing_signatures",
+                self._get_e2e_cross_signing_signatures_txn,
+                result,
+                from_user_id,
+            )
+
+        return result
+
     def get_all_user_signature_changes_for_remotes(self, from_key, to_key):
         """Return a list of changes from the user signature stream to notify remotes.
         Note that the user signature stream represents when a user signs their
@@ -520,6 +725,10 @@ class EndToEndKeyStore(EndToEndKeyWorkerStore, SQLBaseStore):
                 },
             )
 
+        self._invalidate_cache_and_stream(
+            txn, self._get_bare_e2e_cross_signing_keys, (user_id,)
+        )
+
     def set_e2e_cross_signing_key(self, user_id, key_type, key):
         """Set a user's cross-signing key.
 

synapse/storage/data_stores/main/event_federation.py

@@ -14,13 +14,10 @@
 # limitations under the License.
 import itertools
 import logging
-import random
 
 from six.moves import range
 from six.moves.queue import Empty, PriorityQueue
 
-from unpaddedbase64 import encode_base64
-
 from twisted.internet import defer
 
 from synapse.api.errors import StoreError
@@ -148,8 +145,7 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas
             retcol="event_id",
         )
 
-    @defer.inlineCallbacks
-    def get_prev_events_for_room(self, room_id):
+    def get_prev_events_for_room(self, room_id: str):
         """
         Gets a subset of the current forward extremities in the given room.
 
@@ -160,41 +156,30 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas
             room_id (str): room_id
 
         Returns:
-            Deferred[list[(str, dict[str, str], int)]]
-                for each event, a tuple of (event_id, hashes, depth)
-                where *hashes* is a map from algorithm to hash.
-        """
-        res = yield self.get_latest_event_ids_and_hashes_in_room(room_id)
-        if len(res) > 10:
-            # Sort by reverse depth, so we point to the most recent.
-            res.sort(key=lambda a: -a[2])
+            Deferred[List[str]]: the event ids of the forward extremites
 
-            # we use half of the limit for the actual most recent events, and
-            # the other half to randomly point to some of the older events, to
-            # make sure that we don't completely ignore the older events.
-            res = res[0:5] + random.sample(res[5:], 5)
-
-        return res
-
-    def get_latest_event_ids_and_hashes_in_room(self, room_id):
-        """
-        Gets the current forward extremities in the given room
-
-        Args:
-            room_id (str): room_id
-
-        Returns:
-            Deferred[list[(str, dict[str, str], int)]]
-                for each event, a tuple of (event_id, hashes, depth)
-                where *hashes* is a map from algorithm to hash.
         """
 
         return self.db.runInteraction(
-            "get_latest_event_ids_and_hashes_in_room",
-            self._get_latest_event_ids_and_hashes_in_room,
-            room_id,
+            "get_prev_events_for_room", self._get_prev_events_for_room_txn, room_id
         )
 
+    def _get_prev_events_for_room_txn(self, txn, room_id: str):
+        # we just use the 10 newest events. Older events will become
+        # prev_events of future events.
+
+        sql = """
+            SELECT e.event_id FROM event_forward_extremities AS f
+            INNER JOIN events AS e USING (event_id)
+            WHERE f.room_id = ?
+            ORDER BY e.depth DESC
+            LIMIT 10
+        """
+
+        txn.execute(sql, (room_id,))
+
+        return [row[0] for row in txn]
+
     def get_rooms_with_many_extremities(self, min_count, limit, room_id_filter):
         """Get the top rooms with at least N extremities.
 
@@ -243,27 +228,6 @@ class EventFederationWorkerStore(EventsWorkerStore, SignatureWorkerStore, SQLBas
             desc="get_latest_event_ids_in_room",
         )
 
-    def _get_latest_event_ids_and_hashes_in_room(self, txn, room_id):
-        sql = (
-            "SELECT e.event_id, e.depth FROM events as e "
-            "INNER JOIN event_forward_extremities as f "
-            "ON e.event_id = f.event_id "
-            "AND e.room_id = f.room_id "
-            "WHERE f.room_id = ?"
-        )
-
-        txn.execute(sql, (room_id,))
-
-        results = []
-        for event_id, depth in txn.fetchall():
-            hashes = self._get_event_reference_hashes_txn(txn, event_id)
-            prev_hashes = {
-                k: encode_base64(v) for k, v in hashes.items() if k == "sha256"
-            }
-            results.append((event_id, prev_hashes, depth))
-
-        return results
-
     def get_min_depth(self, room_id):
         """ For hte given room, get the minimum depth we have seen for it.
         """
@@ -506,7 +470,7 @@ class EventFederationStore(EventFederationWorkerStore):
     def _update_min_depth_for_room_txn(self, txn, room_id, depth):
         min_depth = self._get_min_depth_interaction(txn, room_id)
 
-        if min_depth and depth >= min_depth:
+        if min_depth is not None and depth >= min_depth:
             return
 
         self.db.simple_upsert_txn(

synapse/storage/data_stores/main/events.py

@@ -1757,163 +1757,6 @@ class EventsStore(
 
         return state_groups
 
-    def purge_unreferenced_state_groups(
-        self, room_id: str, state_groups_to_delete
-    ) -> defer.Deferred:
-        """Deletes no longer referenced state groups and de-deltas any state
-        groups that reference them.
-
-        Args:
-            room_id: The room the state groups belong to (must all be in the
-                same room).
-            state_groups_to_delete (Collection[int]): Set of all state groups
-                to delete.
-        """
-
-        return self.db.runInteraction(
-            "purge_unreferenced_state_groups",
-            self._purge_unreferenced_state_groups,
-            room_id,
-            state_groups_to_delete,
-        )
-
-    def _purge_unreferenced_state_groups(self, txn, room_id, state_groups_to_delete):
-        logger.info(
-            "[purge] found %i state groups to delete", len(state_groups_to_delete)
-        )
-
-        rows = self.db.simple_select_many_txn(
-            txn,
-            table="state_group_edges",
-            column="prev_state_group",
-            iterable=state_groups_to_delete,
-            keyvalues={},
-            retcols=("state_group",),
-        )
-
-        remaining_state_groups = set(
-            row["state_group"]
-            for row in rows
-            if row["state_group"] not in state_groups_to_delete
-        )
-
-        logger.info(
-            "[purge] de-delta-ing %i remaining state groups",
-            len(remaining_state_groups),
-        )
-
-        # Now we turn the state groups that reference to-be-deleted state
-        # groups to non delta versions.
-        for sg in remaining_state_groups:
-            logger.info("[purge] de-delta-ing remaining state group %s", sg)
-            curr_state = self._get_state_groups_from_groups_txn(txn, [sg])
-            curr_state = curr_state[sg]
-
-            self.db.simple_delete_txn(
-                txn, table="state_groups_state", keyvalues={"state_group": sg}
-            )
-
-            self.db.simple_delete_txn(
-                txn, table="state_group_edges", keyvalues={"state_group": sg}
-            )
-
-            self.db.simple_insert_many_txn(
-                txn,
-                table="state_groups_state",
-                values=[
-                    {
-                        "state_group": sg,
-                        "room_id": room_id,
-                        "type": key[0],
-                        "state_key": key[1],
-                        "event_id": state_id,
-                    }
-                    for key, state_id in iteritems(curr_state)
-                ],
-            )
-
-        logger.info("[purge] removing redundant state groups")
-        txn.executemany(
-            "DELETE FROM state_groups_state WHERE state_group = ?",
-            ((sg,) for sg in state_groups_to_delete),
-        )
-        txn.executemany(
-            "DELETE FROM state_groups WHERE id = ?",
-            ((sg,) for sg in state_groups_to_delete),
-        )
-
-    @defer.inlineCallbacks
-    def get_previous_state_groups(self, state_groups):
-        """Fetch the previous groups of the given state groups.
-
-        Args:
-            state_groups (Iterable[int])
-
-        Returns:
-            Deferred[dict[int, int]]: mapping from state group to previous
-            state group.
-        """
-
-        rows = yield self.db.simple_select_many_batch(
-            table="state_group_edges",
-            column="prev_state_group",
-            iterable=state_groups,
-            keyvalues={},
-            retcols=("prev_state_group", "state_group"),
-            desc="get_previous_state_groups",
-        )
-
-        return {row["state_group"]: row["prev_state_group"] for row in rows}
-
-    def purge_room_state(self, room_id, state_groups_to_delete):
-        """Deletes all record of a room from state tables
-
-        Args:
-            room_id (str):
-            state_groups_to_delete (list[int]): State groups to delete
-        """
-
-        return self.db.runInteraction(
-            "purge_room_state",
-            self._purge_room_state_txn,
-            room_id,
-            state_groups_to_delete,
-        )
-
-    def _purge_room_state_txn(self, txn, room_id, state_groups_to_delete):
-        # first we have to delete the state groups states
-        logger.info("[purge] removing %s from state_groups_state", room_id)
-
-        self.db.simple_delete_many_txn(
-            txn,
-            table="state_groups_state",
-            column="state_group",
-            iterable=state_groups_to_delete,
-            keyvalues={},
-        )
-
-        # ... and the state group edges
-        logger.info("[purge] removing %s from state_group_edges", room_id)
-
-        self.db.simple_delete_many_txn(
-            txn,
-            table="state_group_edges",
-            column="state_group",
-            iterable=state_groups_to_delete,
-            keyvalues={},
-        )
-
-        # ... and the state groups
-        logger.info("[purge] removing %s from state_groups", room_id)
-
-        self.db.simple_delete_many_txn(
-            txn,
-            table="state_groups",
-            column="id",
-            iterable=state_groups_to_delete,
-            keyvalues={},
-        )
-
     async def is_event_after(self, event_id1, event_id2):
         """Returns True if event_id1 is after event_id2 in the stream
         """

synapse/storage/data_stores/main/events_worker.py

@@ -19,8 +19,10 @@ import itertools
 import logging
 import threading
 from collections import namedtuple
+from typing import List, Optional
 
 from canonicaljson import json
+from constantly import NamedConstant, Names
 
 from twisted.internet import defer
 
@@ -55,6 +57,16 @@ EVENT_QUEUE_TIMEOUT_S = 0.1  # Timeout when waiting for requests for events
 _EventCacheEntry = namedtuple("_EventCacheEntry", ("event", "redacted_event"))
 
 
+class EventRedactBehaviour(Names):
+    """
+    What to do when retrieving a redacted event from the database.
+    """
+
+    AS_IS = NamedConstant()
+    REDACT = NamedConstant()
+    BLOCK = NamedConstant()
+
+
 class EventsWorkerStore(SQLBaseStore):
     def __init__(self, database: Database, db_conn, hs):
         super(EventsWorkerStore, self).__init__(database, db_conn, hs)
@@ -125,25 +137,34 @@ class EventsWorkerStore(SQLBaseStore):
     @defer.inlineCallbacks
     def get_event(
         self,
-        event_id,
-        check_redacted=True,
-        get_prev_content=False,
-        allow_rejected=False,
-        allow_none=False,
-        check_room_id=None,
+        event_id: str,
+        redact_behaviour: EventRedactBehaviour = EventRedactBehaviour.REDACT,
+        get_prev_content: bool = False,
+        allow_rejected: bool = False,
+        allow_none: bool = False,
+        check_room_id: Optional[str] = None,
     ):
         """Get an event from the database by event_id.
 
         Args:
-            event_id (str): The event_id of the event to fetch
-            check_redacted (bool): If True, check if event has been redacted
-                and redact it.
-            get_prev_content (bool): If True and event is a state event,
+            event_id: The event_id of the event to fetch
+
+            redact_behaviour: Determine what to do with a redacted event. Possible values:
+                * AS_IS - Return the full event body with no redacted content
+                * REDACT - Return the event but with a redacted body
+                * DISALLOW - Do not return redacted events (behave as per allow_none
+                    if the event is redacted)
+
+            get_prev_content: If True and event is a state event,
                 include the previous states content in the unsigned field.
-            allow_rejected (bool): If True return rejected events.
-            allow_none (bool): If True, return None if no event found, if
+
+            allow_rejected: If True, return rejected events. Otherwise,
+                behave as per allow_none.
+
+            allow_none: If True, return None if no event found, if
                 False throw a NotFoundError
-            check_room_id (str|None): if not None, check the room of the found event.
+
+            check_room_id: if not None, check the room of the found event.
                 If there is a mismatch, behave as per allow_none.
 
         Returns:
@@ -154,7 +175,7 @@ class EventsWorkerStore(SQLBaseStore):
 
         events = yield self.get_events_as_list(
             [event_id],
-            check_redacted=check_redacted,
+            redact_behaviour=redact_behaviour,
             get_prev_content=get_prev_content,
             allow_rejected=allow_rejected,
         )
@@ -173,27 +194,34 @@ class EventsWorkerStore(SQLBaseStore):
     @defer.inlineCallbacks
     def get_events(
         self,
-        event_ids,
-        check_redacted=True,
-        get_prev_content=False,
-        allow_rejected=False,
+        event_ids: List[str],
+        redact_behaviour: EventRedactBehaviour = EventRedactBehaviour.REDACT,
+        get_prev_content: bool = False,
+        allow_rejected: bool = False,
     ):
         """Get events from the database
 
         Args:
-            event_ids (list): The event_ids of the events to fetch
-            check_redacted (bool): If True, check if event has been redacted
-                and redact it.
-            get_prev_content (bool): If True and event is a state event,
+            event_ids: The event_ids of the events to fetch
+
+            redact_behaviour: Determine what to do with a redacted event. Possible
+                values:
+                * AS_IS - Return the full event body with no redacted content
+                * REDACT - Return the event but with a redacted body
+                * DISALLOW - Do not return redacted events (omit them from the response)
+
+            get_prev_content: If True and event is a state event,
                 include the previous states content in the unsigned field.
-            allow_rejected (bool): If True return rejected events.
+
+            allow_rejected: If True, return rejected events. Otherwise,
+                omits rejeted events from the response.
 
         Returns:
             Deferred : Dict from event_id to event.
         """
         events = yield self.get_events_as_list(
             event_ids,
-            check_redacted=check_redacted,
+            redact_behaviour=redact_behaviour,
             get_prev_content=get_prev_content,
             allow_rejected=allow_rejected,
         )
@@ -203,21 +231,29 @@ class EventsWorkerStore(SQLBaseStore):
     @defer.inlineCallbacks
     def get_events_as_list(
         self,
-        event_ids,
-        check_redacted=True,
-        get_prev_content=False,
-        allow_rejected=False,
+        event_ids: List[str],
+        redact_behaviour: EventRedactBehaviour = EventRedactBehaviour.REDACT,
+        get_prev_content: bool = False,
+        allow_rejected: bool = False,
     ):
         """Get events from the database and return in a list in the same order
         as given by `event_ids` arg.
 
+        Unknown events will be omitted from the response.
+
         Args:
-            event_ids (list): The event_ids of the events to fetch
-            check_redacted (bool): If True, check if event has been redacted
-                and redact it.
-            get_prev_content (bool): If True and event is a state event,
+            event_ids: The event_ids of the events to fetch
+
+            redact_behaviour: Determine what to do with a redacted event. Possible values:
+                * AS_IS - Return the full event body with no redacted content
+                * REDACT - Return the event but with a redacted body
+                * DISALLOW - Do not return redacted events (omit them from the response)
+
+            get_prev_content: If True and event is a state event,
                 include the previous states content in the unsigned field.
-            allow_rejected (bool): If True return rejected events.
+
+            allow_rejected: If True, return rejected events. Otherwise,
+                omits rejected events from the response.
 
         Returns:
             Deferred[list[EventBase]]: List of events fetched from the database. The
@@ -319,10 +355,14 @@ class EventsWorkerStore(SQLBaseStore):
                     # Update the cache to save doing the checks again.
                     entry.event.internal_metadata.recheck_redaction = False
 
-            if check_redacted and entry.redacted_event:
-                event = entry.redacted_event
-            else:
-                event = entry.event
+            event = entry.event
+
+            if entry.redacted_event:
+                if redact_behaviour == EventRedactBehaviour.BLOCK:
+                    # Skip this event
+                    continue
+                elif redact_behaviour == EventRedactBehaviour.REDACT:
+                    event = entry.redacted_event
 
             events.append(event)
 
@@ -346,9 +386,14 @@ class EventsWorkerStore(SQLBaseStore):
 
         If events are pulled from the database, they will be cached for future lookups.
 
+        Unknown events are omitted from the response.
+
         Args:
+
             event_ids (Iterable[str]): The event_ids of the events to fetch
-            allow_rejected (bool): Whether to include rejected events
+
+            allow_rejected (bool): Whether to include rejected events. If False,
+                rejected events are omitted from the response.
 
         Returns:
             Deferred[Dict[str, _EventCacheEntry]]:
@@ -483,9 +528,13 @@ class EventsWorkerStore(SQLBaseStore):
 
         Returned events will be added to the cache for future lookups.
 
+        Unknown events are omitted from the response.
+
         Args:
             event_ids (Iterable[str]): The event_ids of the events to fetch
-            allow_rejected (bool): Whether to include rejected events
+
+            allow_rejected (bool): Whether to include rejected events. If False,
+                rejected events are omitted from the response.
 
         Returns:
             Deferred[Dict[str, _EventCacheEntry]]:

synapse/storage/data_stores/main/push_rule.py

@@ -244,7 +244,7 @@ class PushRulesWorkerStore(
             # To do this we set the state_group to a new object as object() != object()
             state_group = object()
 
-        current_state_ids = yield context.get_current_state_ids(self)
+        current_state_ids = yield context.get_current_state_ids()
         result = yield self._bulk_get_push_rules_for_room(
             event.room_id, state_group, current_state_ids, event=event
         )

synapse/storage/data_stores/main/pusher.py

@@ -15,8 +15,7 @@
 # limitations under the License.
 
 import logging
-
-import six
+from typing import Iterable, Iterator
 
 from canonicaljson import encode_canonical_json, json
 
@@ -27,21 +26,16 @@ from synapse.util.caches.descriptors import cachedInlineCallbacks, cachedList
 
 logger = logging.getLogger(__name__)
 
-if six.PY2:
-    db_binary_type = six.moves.builtins.buffer
-else:
-    db_binary_type = memoryview
-
 
 class PusherWorkerStore(SQLBaseStore):
-    def _decode_pushers_rows(self, rows):
+    def _decode_pushers_rows(self, rows: Iterable[dict]) -> Iterator[dict]:
+        """JSON-decode the data in the rows returned from the `pushers` table
+
+        Drops any rows whose data cannot be decoded
+        """
         for r in rows:
             dataJson = r["data"]
-            r["data"] = None
             try:
-                if isinstance(dataJson, db_binary_type):
-                    dataJson = str(dataJson).decode("UTF8")
-
                 r["data"] = json.loads(dataJson)
             except Exception as e:
                 logger.warning(
@@ -50,12 +44,9 @@ class PusherWorkerStore(SQLBaseStore):
                     dataJson,
                     e.args[0],
                 )
-                pass
+                continue
 
-            if isinstance(r["pushkey"], db_binary_type):
-                r["pushkey"] = str(r["pushkey"]).decode("UTF8")
-
-        return rows
+            yield r
 
     @defer.inlineCallbacks
     def user_has_pusher(self, user_id):