Merge commit '3f58fc848' into anoa/dinsic_release_1_31_0

tests/rest/admin/test_room.py

@@ -1445,6 +1445,90 @@ class JoinAliasRoomTestCase(unittest.HomeserverTestCase):
         self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"])
         self.assertEqual(private_room_id, channel.json_body["joined_rooms"][0])
 
+    def test_context_as_non_admin(self):
+        """
+        Test that, without being admin, one cannot use the context admin API
+        """
+        # Create a room.
+        user_id = self.register_user("test", "test")
+        user_tok = self.login("test", "test")
+
+        self.register_user("test_2", "test")
+        user_tok_2 = self.login("test_2", "test")
+
+        room_id = self.helper.create_room_as(user_id, tok=user_tok)
+
+        # Populate the room with events.
+        events = []
+        for i in range(30):
+            events.append(
+                self.helper.send_event(
+                    room_id, "com.example.test", content={"index": i}, tok=user_tok
+                )
+            )
+
+        # Now attempt to find the context using the admin API without being admin.
+        midway = (len(events) - 1) // 2
+        for tok in [user_tok, user_tok_2]:
+            channel = self.make_request(
+                "GET",
+                "/_synapse/admin/v1/rooms/%s/context/%s"
+                % (room_id, events[midway]["event_id"]),
+                access_token=tok,
+            )
+            self.assertEquals(
+                403, int(channel.result["code"]), msg=channel.result["body"]
+            )
+            self.assertEqual(Codes.FORBIDDEN, channel.json_body["errcode"])
+
+    def test_context_as_admin(self):
+        """
+        Test that, as admin, we can find the context of an event without having joined the room.
+        """
+
+        # Create a room. We're not part of it.
+        user_id = self.register_user("test", "test")
+        user_tok = self.login("test", "test")
+        room_id = self.helper.create_room_as(user_id, tok=user_tok)
+
+        # Populate the room with events.
+        events = []
+        for i in range(30):
+            events.append(
+                self.helper.send_event(
+                    room_id, "com.example.test", content={"index": i}, tok=user_tok
+                )
+            )
+
+        # Now let's fetch the context for this room.
+        midway = (len(events) - 1) // 2
+        channel = self.make_request(
+            "GET",
+            "/_synapse/admin/v1/rooms/%s/context/%s"
+            % (room_id, events[midway]["event_id"]),
+            access_token=self.admin_user_tok,
+        )
+        self.assertEquals(200, int(channel.result["code"]), msg=channel.result["body"])
+        self.assertEquals(
+            channel.json_body["event"]["event_id"], events[midway]["event_id"]
+        )
+
+        for i, found_event in enumerate(channel.json_body["events_before"]):
+            for j, posted_event in enumerate(events):
+                if found_event["event_id"] == posted_event["event_id"]:
+                    self.assertTrue(j < midway)
+                    break
+            else:
+                self.fail("Event %s from events_before not found" % j)
+
+        for i, found_event in enumerate(channel.json_body["events_after"]):
+            for j, posted_event in enumerate(events):
+                if found_event["event_id"] == posted_event["event_id"]:
+                    self.assertTrue(j > midway)
+                    break
+            else:
+                self.fail("Event %s from events_after not found" % j)
+
 
 class MakeRoomAdminTestCase(unittest.HomeserverTestCase):
     servlets = [

tests/test_preview.py

@@ -15,6 +15,7 @@
 
 from synapse.rest.media.v1.preview_url_resource import (
     decode_and_calc_og,
+    get_html_media_encoding,
     summarize_paragraphs,
 )
 
@@ -26,7 +27,7 @@ except ImportError:
     lxml = None
 
 
-class PreviewTestCase(unittest.TestCase):
+class SummarizeTestCase(unittest.TestCase):
     if not lxml:
         skip = "url preview feature requires lxml"
 
@@ -144,12 +145,12 @@ class PreviewTestCase(unittest.TestCase):
         )
 
 
-class PreviewUrlTestCase(unittest.TestCase):
+class CalcOgTestCase(unittest.TestCase):
     if not lxml:
         skip = "url preview feature requires lxml"
 
     def test_simple(self):
-        html = """
+        html = b"""
         <html>
         <head><title>Foo</title></head>
         <body>
@@ -163,7 +164,7 @@ class PreviewUrlTestCase(unittest.TestCase):
         self.assertEqual(og, {"og:title": "Foo", "og:description": "Some text."})
 
     def test_comment(self):
-        html = """
+        html = b"""
         <html>
         <head><title>Foo</title></head>
         <body>
@@ -178,7 +179,7 @@ class PreviewUrlTestCase(unittest.TestCase):
         self.assertEqual(og, {"og:title": "Foo", "og:description": "Some text."})
 
     def test_comment2(self):
-        html = """
+        html = b"""
         <html>
         <head><title>Foo</title></head>
         <body>
@@ -202,7 +203,7 @@ class PreviewUrlTestCase(unittest.TestCase):
         )
 
     def test_script(self):
-        html = """
+        html = b"""
         <html>
         <head><title>Foo</title></head>
         <body>
@@ -217,7 +218,7 @@ class PreviewUrlTestCase(unittest.TestCase):
         self.assertEqual(og, {"og:title": "Foo", "og:description": "Some text."})
 
     def test_missing_title(self):
-        html = """
+        html = b"""
         <html>
         <body>
         Some text.
@@ -230,7 +231,7 @@ class PreviewUrlTestCase(unittest.TestCase):
         self.assertEqual(og, {"og:title": None, "og:description": "Some text."})
 
     def test_h1_as_title(self):
-        html = """
+        html = b"""
         <html>
         <meta property="og:description" content="Some text."/>
         <body>
@@ -244,7 +245,7 @@ class PreviewUrlTestCase(unittest.TestCase):
         self.assertEqual(og, {"og:title": "Title", "og:description": "Some text."})
 
     def test_missing_title_and_broken_h1(self):
-        html = """
+        html = b"""
         <html>
         <body>
         <h1><a href="foo"/></h1>
@@ -258,13 +259,20 @@ class PreviewUrlTestCase(unittest.TestCase):
         self.assertEqual(og, {"og:title": None, "og:description": "Some text."})
 
     def test_empty(self):
-        html = ""
+        """Test a body with no data in it."""
+        html = b""
+        og = decode_and_calc_og(html, "http://example.com/test.html")
+        self.assertEqual(og, {})
+
+    def test_no_tree(self):
+        """A valid body with no tree in it."""
+        html = b"\x00"
         og = decode_and_calc_og(html, "http://example.com/test.html")
         self.assertEqual(og, {})
 
     def test_invalid_encoding(self):
         """An invalid character encoding should be ignored and treated as UTF-8, if possible."""
-        html = """
+        html = b"""
         <html>
         <head><title>Foo</title></head>
         <body>
@@ -290,3 +298,76 @@ class PreviewUrlTestCase(unittest.TestCase):
         """
         og = decode_and_calc_og(html, "http://example.com/test.html")
         self.assertEqual(og, {"og:title": "ÿÿ Foo", "og:description": "Some text."})
+
+
+class MediaEncodingTestCase(unittest.TestCase):
+    def test_meta_charset(self):
+        """A character encoding is found via the meta tag."""
+        encoding = get_html_media_encoding(
+            b"""
+        <html>
+        <head><meta charset="ascii">
+        </head>
+        </html>
+        """,
+            "text/html",
+        )
+        self.assertEqual(encoding, "ascii")
+
+        # A less well-formed version.
+        encoding = get_html_media_encoding(
+            b"""
+        <html>
+        <head>< meta charset = ascii>
+        </head>
+        </html>
+        """,
+            "text/html",
+        )
+        self.assertEqual(encoding, "ascii")
+
+    def test_xml_encoding(self):
+        """A character encoding is found via the meta tag."""
+        encoding = get_html_media_encoding(
+            b"""
+        <?xml version="1.0" encoding="ascii"?>
+        <html>
+        </html>
+        """,
+            "text/html",
+        )
+        self.assertEqual(encoding, "ascii")
+
+    def test_meta_xml_encoding(self):
+        """Meta tags take precedence over XML encoding."""
+        encoding = get_html_media_encoding(
+            b"""
+        <?xml version="1.0" encoding="ascii"?>
+        <html>
+        <head><meta charset="UTF-16">
+        </head>
+        </html>
+        """,
+            "text/html",
+        )
+        self.assertEqual(encoding, "UTF-16")
+
+    def test_content_type(self):
+        """A character encoding is found via the Content-Type header."""
+        # Test a few variations of the header.
+        headers = (
+            'text/html; charset="ascii";',
+            "text/html;charset=ascii;",
+            'text/html;  charset="ascii"',
+            "text/html; charset=ascii",
+            'text/html; charset="ascii;',
+            'text/html; charset=ascii";',
+        )
+        for header in headers:
+            encoding = get_html_media_encoding(b"", header)
+            self.assertEqual(encoding, "ascii")
+
+    def test_fallback(self):
+        """A character encoding cannot be found in the body or header."""
+        encoding = get_html_media_encoding(b"", "text/html")
+        self.assertEqual(encoding, "utf-8")