Merge tag 'v1.60.0rc2' into develop

Synapse 1.60.0rc2 (2022-05-27)
==============================

This release of Synapse adds a unique index to the `state_group_edges` table, in
order to prevent accidentally introducing duplicate information (for example,
because a database backup was restored multiple times). If your Synapse database
already has duplicate rows in this table, this could fail with an error and
require manual remediation.

Additionally, the signature of the `check_event_for_spam` module callback has changed.
The previous signature has been deprecated and remains working for now. Module authors
should update their modules to use the new signature where possible.

See [the upgrade notes](https://github.com/matrix-org/synapse/blob/develop/docs/upgrade.md#upgrading-to-v1600)
for more details.

Features
--------

- Add an option allowing users to use their password to reauthenticate for privileged actions even though password login is disabled. ([\#12883](https://github.com/matrix-org/synapse/issues/12883))

Bugfixes
--------

- Explicitly close `ijson` coroutines once we are done with them, instead of leaving the garbage collector to close them. ([\#12875](https://github.com/matrix-org/synapse/issues/12875))

Internal Changes
----------------

- Improve URL previews by not including the content of media tags in the generated description. ([\#12887](https://github.com/matrix-org/synapse/issues/12887))

5 files changed, 62 insertions, 14 deletions

diff --git a/synapse/config/auth.py b/synapse/config/auth.py
index bb417a2359..265a554a5d 100644
--- a/synapse/config/auth.py
+++ b/synapse/config/auth.py
@@ -29,7 +29,18 @@ class AuthConfig(Config):
         if password_config is None:
             password_config = {}
 
-        self.password_enabled = password_config.get("enabled", True)
+        passwords_enabled = password_config.get("enabled", True)
+        # 'only_for_reauth' allows users who have previously set a password to use it,
+        # even though passwords would otherwise be disabled.
+        passwords_for_reauth_only = passwords_enabled == "only_for_reauth"
+
+        self.password_enabled_for_login = (
+            passwords_enabled and not passwords_for_reauth_only
+        )
+        self.password_enabled_for_reauth = (
+            passwords_for_reauth_only or passwords_enabled
+        )
+
         self.password_localdb_enabled = password_config.get("localdb_enabled", True)
         self.password_pepper = password_config.get("pepper", "")
 
@@ -46,7 +57,9 @@ class AuthConfig(Config):
     def generate_config_section(self, **kwargs: Any) -> str:
         return """\
         password_config:
-           # Uncomment to disable password login
+           # Uncomment to disable password login.
+           # Set to `only_for_reauth` to permit reauthentication for users that
+           # have passwords and are already logged in.
            #
            #enabled: false
 
diff --git a/synapse/federation/transport/client.py b/synapse/federation/transport/client.py
index 25df1905c6..9da80176a5 100644
--- a/synapse/federation/transport/client.py
+++ b/synapse/federation/transport/client.py
@@ -1361,7 +1361,7 @@ class SendJoinParser(ByteParser[SendJoinResponse]):
     def __init__(self, room_version: RoomVersion, v1_api: bool):
         self._response = SendJoinResponse([], [], event_dict={})
         self._room_version = room_version
-        self._coros = []
+        self._coros: List[Generator[None, bytes, None]] = []
 
         # The V1 API has the shape of `[200, {...}]`, which we handle by
         # prefixing with `item.*`.
@@ -1409,6 +1409,9 @@ class SendJoinParser(ByteParser[SendJoinResponse]):
         return len(data)
 
     def finish(self) -> SendJoinResponse:
+        for c in self._coros:
+            c.close()
+
         if self._response.event_dict:
             self._response.event = make_event_from_dict(
                 self._response.event_dict, self._room_version
@@ -1431,7 +1434,7 @@ class _StateParser(ByteParser[StateRequestResponse]):
     def __init__(self, room_version: RoomVersion):
         self._response = StateRequestResponse([], [])
         self._room_version = room_version
-        self._coros = [
+        self._coros: List[Generator[None, bytes, None]] = [
             ijson.items_coro(
                 _event_list_parser(room_version, self._response.state),
                 "pdus.item",
@@ -1450,4 +1453,6 @@ class _StateParser(ByteParser[StateRequestResponse]):
         return len(data)
 
     def finish(self) -> StateRequestResponse:
+        for c in self._coros:
+            c.close()
         return self._response
diff --git a/synapse/handlers/auth.py b/synapse/handlers/auth.py
index 1b9050ea96..fbafbbee6b 100644
--- a/synapse/handlers/auth.py
+++ b/synapse/handlers/auth.py
@@ -210,7 +210,8 @@ class AuthHandler:
 
         self.hs = hs  # FIXME better possibility to access registrationHandler later?
         self.macaroon_gen = hs.get_macaroon_generator()
-        self._password_enabled = hs.config.auth.password_enabled
+        self._password_enabled_for_login = hs.config.auth.password_enabled_for_login
+        self._password_enabled_for_reauth = hs.config.auth.password_enabled_for_reauth
         self._password_localdb_enabled = hs.config.auth.password_localdb_enabled
         self._third_party_rules = hs.get_third_party_event_rules()
 
@@ -387,13 +388,13 @@ class AuthHandler:
         return params, session_id
 
     async def _get_available_ui_auth_types(self, user: UserID) -> Iterable[str]:
-        """Get a list of the authentication types this user can use"""
+        """Get a list of the user-interactive authentication types this user can use."""
 
         ui_auth_types = set()
 
         # if the HS supports password auth, and the user has a non-null password, we
         # support password auth
-        if self._password_localdb_enabled and self._password_enabled:
+        if self._password_localdb_enabled and self._password_enabled_for_reauth:
             lookupres = await self._find_user_id_and_pwd_hash(user.to_string())
             if lookupres:
                 _, password_hash = lookupres
@@ -402,7 +403,7 @@ class AuthHandler:
 
         # also allow auth from password providers
         for t in self.password_auth_provider.get_supported_login_types().keys():
-            if t == LoginType.PASSWORD and not self._password_enabled:
+            if t == LoginType.PASSWORD and not self._password_enabled_for_reauth:
                 continue
             ui_auth_types.add(t)
 
@@ -710,7 +711,7 @@ class AuthHandler:
             return res
 
         # fall back to the v1 login flow
-        canonical_id, _ = await self.validate_login(authdict)
+        canonical_id, _ = await self.validate_login(authdict, is_reauth=True)
         return canonical_id
 
     def _get_params_recaptcha(self) -> dict:
@@ -1064,7 +1065,7 @@ class AuthHandler:
         Returns:
             Whether users on this server are allowed to change or set a password
         """
-        return self._password_enabled and self._password_localdb_enabled
+        return self._password_enabled_for_login and self._password_localdb_enabled
 
     def get_supported_login_types(self) -> Iterable[str]:
         """Get a the login types supported for the /login API
@@ -1089,9 +1090,9 @@ class AuthHandler:
         # that comes first, where it's present.
         if LoginType.PASSWORD in types:
             types.remove(LoginType.PASSWORD)
-            if self._password_enabled:
+            if self._password_enabled_for_login:
                 types.insert(0, LoginType.PASSWORD)
-        elif self._password_localdb_enabled and self._password_enabled:
+        elif self._password_localdb_enabled and self._password_enabled_for_login:
             types.insert(0, LoginType.PASSWORD)
 
         return types
@@ -1100,6 +1101,7 @@ class AuthHandler:
         self,
         login_submission: Dict[str, Any],
         ratelimit: bool = False,
+        is_reauth: bool = False,
     ) -> Tuple[str, Optional[Callable[["LoginResponse"], Awaitable[None]]]]:
         """Authenticates the user for the /login API
 
@@ -1110,6 +1112,9 @@ class AuthHandler:
             login_submission: the whole of the login submission
                 (including 'type' and other relevant fields)
             ratelimit: whether to apply the failed_login_attempt ratelimiter
+            is_reauth: whether this is part of a User-Interactive Authorisation
+                flow to reauthenticate for a privileged action (rather than a
+                new login)
         Returns:
             A tuple of the canonical user id, and optional callback
                 to be called once the access token and device id are issued
@@ -1132,8 +1137,14 @@ class AuthHandler:
         # special case to check for "password" for the check_password interface
         # for the auth providers
         password = login_submission.get("password")
+
         if login_type == LoginType.PASSWORD:
-            if not self._password_enabled:
+            if is_reauth:
+                passwords_allowed_here = self._password_enabled_for_reauth
+            else:
+                passwords_allowed_here = self._password_enabled_for_login
+
+            if not passwords_allowed_here:
                 raise SynapseError(400, "Password login has been disabled.")
             if not isinstance(password, str):
                 raise SynapseError(400, "Bad parameter: password", Codes.INVALID_PARAM)
diff --git a/synapse/http/matrixfederationclient.py b/synapse/http/matrixfederationclient.py
index db44721ef5..776ed43f03 100644
--- a/synapse/http/matrixfederationclient.py
+++ b/synapse/http/matrixfederationclient.py
@@ -223,6 +223,7 @@ async def _handle_response(
 
     max_response_size = parser.MAX_RESPONSE_SIZE
 
+    finished = False
     try:
         check_content_type_is(response.headers, parser.CONTENT_TYPE)
 
@@ -231,6 +232,7 @@ async def _handle_response(
 
         length = await make_deferred_yieldable(d)
 
+        finished = True
         value = parser.finish()
     except BodyExceededMaxSize as e:
         # The response was too big.
@@ -281,6 +283,15 @@ async def _handle_response(
             e,
         )
         raise
+    finally:
+        if not finished:
+            # There was an exception and we didn't `finish()` the parse.
+            # Let the parser know that it can free up any resources.
+            try:
+                parser.finish()
+            except Exception:
+                # Ignore any additional exceptions.
+                pass
 
     time_taken_secs = reactor.seconds() - start_ms / 1000
 
diff --git a/synapse/rest/media/v1/preview_html.py b/synapse/rest/media/v1/preview_html.py
index e72c8987cc..13ec7ab533 100644
--- a/synapse/rest/media/v1/preview_html.py
+++ b/synapse/rest/media/v1/preview_html.py
@@ -246,7 +246,9 @@ def parse_html_description(tree: "etree.Element") -> Optional[str]:
 
     Grabs any text nodes which are inside the <body/> tag, unless they are within
     an HTML5 semantic markup tag (<header/>, <nav/>, <aside/>, <footer/>), or
-    if they are within a <script/> or <style/> tag.
+    if they are within a <script/>, <svg/> or <style/> tag, or if they are within
+    a tag whose content is usually only shown to old browsers
+    (<iframe/>, <video/>, <canvas/>, <picture/>).
 
     This is a very very very coarse approximation to a plain text render of the page.
 
@@ -268,6 +270,12 @@ def parse_html_description(tree: "etree.Element") -> Optional[str]:
         "script",
         "noscript",
         "style",
+        "svg",
+        "iframe",
+        "video",
+        "canvas",
+        "img",
+        "picture",
         etree.Comment,
     )