import_realm: Import message.edit_history correctly.

Fixes #26369.

There are two important fixes to make to the dicts in edit_history:
1. Update the user_id so that it points to the imported sender.
2. Apply fix_message_rendered_content to the prev_rendered_content data
to fix up mentions and other such syntax.

zerver/lib/import_realm.py

@@ -324,13 +324,17 @@ def fix_customprofilefield(data: TableData) -> None:
 
 
 def fix_message_rendered_content(
-    realm: Realm, sender_map: Dict[int, Record], messages: List[Record]
+    realm: Realm,
+    sender_map: Dict[int, Record],
+    messages: List[Record],
+    content_key: str = "content",
+    rendered_content_key: str = "rendered_content",
 ) -> None:
     """
     This function sets the rendered_content of the messages we're importing.
     """
     for message in messages:
-        if message["rendered_content"] is not None:
+        if message[rendered_content_key] is not None:
             # For Zulip->Zulip imports, we use the original rendered
             # Markdown; this avoids issues where e.g. a mention can no
             # longer render properly because a user has changed their
@@ -339,7 +343,7 @@ def fix_message_rendered_content(
             # However, we still need to update the data-user-id and
             # similar values stored on mentions, stream mentions, and
             # similar syntax in the rendered HTML.
-            soup = BeautifulSoup(message["rendered_content"], "html.parser")
+            soup = BeautifulSoup(message[rendered_content_key], "html.parser")
 
             user_mentions = soup.findAll("span", {"class": "user-mention"})
             if len(user_mentions) != 0:
@@ -356,7 +360,7 @@ def fix_message_rendered_content(
                     old_user_id = int(mention["data-user-id"])
                     if old_user_id in user_id_map:
                         mention["data-user-id"] = str(user_id_map[old_user_id])
-                message["rendered_content"] = str(soup)
+                message[rendered_content_key] = str(soup)
 
             stream_mentions = soup.findAll("a", {"class": "stream"})
             if len(stream_mentions) != 0:
@@ -365,7 +369,7 @@ def fix_message_rendered_content(
                     old_stream_id = int(mention["data-stream-id"])
                     if old_stream_id in stream_id_map:
                         mention["data-stream-id"] = str(stream_id_map[old_stream_id])
-                message["rendered_content"] = str(soup)
+                message[rendered_content_key] = str(soup)
 
             user_group_mentions = soup.findAll("span", {"class": "user-group-mention"})
             if len(user_group_mentions) != 0:
@@ -374,11 +378,11 @@ def fix_message_rendered_content(
                     old_user_group_id = int(mention["data-user-group-id"])
                     if old_user_group_id in user_group_id_map:
                         mention["data-user-group-id"] = str(user_group_id_map[old_user_group_id])
-                message["rendered_content"] = str(soup)
+                message[rendered_content_key] = str(soup)
             continue
 
         try:
-            content = message["content"]
+            content = message[content_key]
 
             sender_id = message["sender_id"]
             sender = sender_map[sender_id]
@@ -398,7 +402,7 @@ def fix_message_rendered_content(
                 translate_emoticons=translate_emoticons,
             ).rendered_content
 
-            message["rendered_content"] = rendered_content
+            message[rendered_content_key] = rendered_content
             if "scheduled_timestamp" not in message:
                 # This logic runs also for ScheduledMessage, which doesn't use
                 # the rendered_content_version field.
@@ -414,6 +418,30 @@ def fix_message_rendered_content(
             )
 
 
+def fix_message_edit_history(
+    realm: Realm, sender_map: Dict[int, Record], messages: List[Record]
+) -> None:
+    user_id_map = ID_MAP["user_profile"]
+    for message in messages:
+        edit_history_json = message.get("edit_history")
+        if not edit_history_json:
+            continue
+
+        edit_history = orjson.loads(edit_history_json)
+        for edit_history_message_dict in edit_history:
+            edit_history_message_dict["user_id"] = user_id_map[edit_history_message_dict["user_id"]]
+
+        fix_message_rendered_content(
+            realm,
+            sender_map,
+            messages=edit_history,
+            content_key="prev_content",
+            rendered_content_key="prev_rendered_content",
+        )
+
+        message["edit_history"] = orjson.dumps(edit_history).decode()
+
+
 def current_table_ids(data: TableData, table: TableName) -> List[int]:
     """
     Returns the ids present in the current table
@@ -1682,6 +1710,9 @@ def import_message_data(realm: Realm, sender_map: Dict[int, Record], import_dir:
         )
         logging.info("Successfully rendered Markdown for message batch")
 
+        fix_message_edit_history(
+            realm=realm, sender_map=sender_map, messages=data["zerver_message"]
+        )
         # A LOT HAPPENS HERE.
         # This is where we actually import the message data.
         bulk_import_model(data, Message)

zerver/tests/test_import_export.py

@@ -1086,6 +1086,54 @@ class RealmImportExportTest(ExportFile):
             Message.objects.filter(realm=imported_realm).count(),
         )
 
+    def test_import_message_edit_history(self) -> None:
+        realm = get_realm("zulip")
+        iago = self.example_user("iago")
+        hamlet = self.example_user("hamlet")
+        user_mention_message = f"@**King Hamlet|{hamlet.id}** Hello"
+
+        self.login_user(iago)
+        message_id = self.send_stream_message(
+            self.example_user("iago"), "Verona", user_mention_message
+        )
+
+        new_content = "new content"
+        result = self.client_patch(
+            f"/json/messages/{message_id}",
+            {
+                "content": new_content,
+            },
+        )
+        self.assert_json_success(result)
+
+        self.export_realm_and_create_auditlog(realm)
+        with self.settings(BILLING_ENABLED=False), self.assertLogs(level="INFO"):
+            do_import_realm(get_output_dir(), "test-zulip")
+        imported_realm = Realm.objects.get(string_id="test-zulip")
+
+        imported_message = Message.objects.filter(realm=imported_realm).latest("id")
+        imported_hamlet_id = UserProfile.objects.get(
+            delivery_email=hamlet.delivery_email, realm=imported_realm
+        ).id
+        imported_iago_id = UserProfile.objects.get(
+            delivery_email=iago.delivery_email, realm=imported_realm
+        ).id
+
+        edit_history_json = imported_message.edit_history
+        assert edit_history_json is not None
+        edit_history = orjson.loads(edit_history_json)
+        self.assert_length(edit_history, 1)
+
+        prev_version_of_message = edit_history[0]
+        # Ensure the "user_id" (of the sender) was updated correctly
+        # to the imported id in the data.
+        self.assertEqual(prev_version_of_message["user_id"], imported_iago_id)
+
+        # The mention metadata in the rendered content should be updated.
+        self.assertIn(
+            f'data-user-id="{imported_hamlet_id}"', prev_version_of_message["prev_rendered_content"]
+        )
+
     def get_realm_getters(self) -> List[Callable[[Realm], object]]:
         names = set()
         getters: List[Callable[[Realm], object]] = []