Add information on uploaded media to user export command. (#15107)

2 files changed, 48 insertions, 0 deletions

diff --git a/synapse/app/admin_cmd.py b/synapse/app/admin_cmd.py
index 5003777f0d..b05fe2c589 100644
--- a/synapse/app/admin_cmd.py
+++ b/synapse/app/admin_cmd.py
@@ -44,6 +44,7 @@ from synapse.storage.databases.main.event_push_actions import (
 )
 from synapse.storage.databases.main.events_worker import EventsWorkerStore
 from synapse.storage.databases.main.filtering import FilteringWorkerStore
+from synapse.storage.databases.main.media_repository import MediaRepositoryStore
 from synapse.storage.databases.main.profile import ProfileWorkerStore
 from synapse.storage.databases.main.push_rule import PushRulesWorkerStore
 from synapse.storage.databases.main.receipts import ReceiptsWorkerStore
@@ -86,6 +87,7 @@ class AdminCmdSlavedStore(
     RegistrationWorkerStore,
     RoomWorkerStore,
     ProfileWorkerStore,
+    MediaRepositoryStore,
 ):
     def __init__(
         self,
@@ -235,6 +237,14 @@ class FileExfiltrationWriter(ExfiltrationWriter):
         with open(account_data_file, "a") as f:
             json.dump(account_data, fp=f)
 
+    def write_media_id(self, media_id: str, media_metadata: JsonDict) -> None:
+        file_directory = os.path.join(self.base_directory, "media_ids")
+        os.makedirs(file_directory, exist_ok=True)
+        media_id_file = os.path.join(file_directory, media_id)
+
+        with open(media_id_file, "w") as f:
+            json.dump(media_metadata, fp=f)
+
     def finished(self) -> str:
         return self.base_directory
 
diff --git a/synapse/handlers/admin.py b/synapse/handlers/admin.py
index 8b7760b2cc..b06f25b03c 100644
--- a/synapse/handlers/admin.py
+++ b/synapse/handlers/admin.py
@@ -252,16 +252,19 @@ class AdminHandler:
         profile = await self.get_user(UserID.from_string(user_id))
         if profile is not None:
             writer.write_profile(profile)
+            logger.info("[%s] Written profile", user_id)
 
         # Get all devices the user has
         devices = await self._device_handler.get_devices_by_user(user_id)
         writer.write_devices(devices)
+        logger.info("[%s] Written %s devices", user_id, len(devices))
 
         # Get all connections the user has
         connections = await self.get_whois(UserID.from_string(user_id))
         writer.write_connections(
             connections["devices"][""]["sessions"][0]["connections"]
         )
+        logger.info("[%s] Written %s connections", user_id, len(connections))
 
         # Get all account data the user has global and in rooms
         global_data = await self._store.get_global_account_data_for_user(user_id)
@@ -269,6 +272,29 @@ class AdminHandler:
         writer.write_account_data("global", global_data)
         for room_id in by_room_data:
             writer.write_account_data(room_id, by_room_data[room_id])
+        logger.info(
+            "[%s] Written account data for %s rooms", user_id, len(by_room_data)
+        )
+
+        # Get all media ids the user has
+        limit = 100
+        start = 0
+        while True:
+            media_ids, total = await self._store.get_local_media_by_user_paginate(
+                start, limit, user_id
+            )
+            for media in media_ids:
+                writer.write_media_id(media["media_id"], media)
+
+            logger.info(
+                "[%s] Written %d media_ids of %s",
+                user_id,
+                (start + len(media_ids)),
+                total,
+            )
+            if (start + limit) >= total:
+                break
+            start += limit
 
         return writer.finished()
 
@@ -360,6 +386,18 @@ class ExfiltrationWriter(metaclass=abc.ABCMeta):
         raise NotImplementedError()
 
     @abc.abstractmethod
+    def write_media_id(self, media_id: str, media_metadata: JsonDict) -> None:
+        """Write the media's metadata of a user.
+        Exports only the metadata, as this can be fetched from the database via
+        read only. In order to access the files, a connection to the correct
+        media repository would be required.
+
+        Args:
+            media_id: ID of the media.
+            media_metadata: Metadata of one media file.
+        """
+
+    @abc.abstractmethod
     def finished(self) -> Any:
         """Called when all data has successfully been exported and written.