Fix timezone parse for non-UTC timezone

Author: kebe7jun

ADDING-A-DATA-SOURCE.md

@@ -238,7 +238,7 @@ def _fetch_channel(self, channel: Channel) -> None:
 ```
 5. Before adding to queue, check whether document is newer than self._last_indexed_at, if not, skip it.
 ```python
-last_modified = datetime.strptime(message["updated_at"], "%Y-%m-%dT%H:%M:%S.%fZ")
+last_modified = datetime.strptime(message["updated_at"], "%Y-%m-%dT%H:%M:%S.%f%z")
 if last_modified < self._last_index_time:
     logger.info(f"Message {message['id']} is too old, skipping")
     continue

app/data_source/sources/bookstack/bookstack.py

@@ -165,7 +165,7 @@ def _feed_book(self, book: Dict):
             self.add_task_to_queue(self._feed_page, raw_page=page)
 
     def _feed_page(self, raw_page: Dict):
-        last_modified = datetime.strptime(raw_page["updated_at"], "%Y-%m-%dT%H:%M:%S.%fZ")
+        last_modified = datetime.strptime(raw_page["updated_at"], "%Y-%m-%dT%H:%M:%S.%f%z")
         if last_modified < self._last_index_time:
             return
 

app/data_source/sources/google_drive/google_drive.py

@@ -77,7 +77,7 @@ def _should_index_file(self, file):
                 f"Skipping file {file['name']} because it's mime type is {file['mimeType']} which is not supported.")
             return False
 
-        last_modified = datetime.strptime(file['modifiedTime'], "%Y-%m-%dT%H:%M:%S.%fZ")
+        last_modified = datetime.strptime(file['modifiedTime'], "%Y-%m-%dT%H:%M:%S.%f%z")
         if last_modified < self._last_index_time:
             return False
 
@@ -170,7 +170,7 @@ def _feed_file(self, file):
 
         parent_name = self._get_parents_string(file)
 
-        last_modified = datetime.strptime(file['modifiedTime'], "%Y-%m-%dT%H:%M:%S.%fZ")
+        last_modified = datetime.strptime(file['modifiedTime'], "%Y-%m-%dT%H:%M:%S.%f%z")
 
         author = file['lastModifyingUser'].get('displayName')
         author_image_url = file['lastModifyingUser'].get('photoLink')

app/data_source/sources/rocketchat/rocketchat.py

@@ -72,7 +72,7 @@ def __init__(self, *args, **kwargs):
         self._authors_cache: Dict[str, RocketchatAuthor] = {}
 
     def _list_rooms(self) -> List[RocketchatRoom]:
-        oldest = self._last_index_time.strftime("%Y-%m-%dT%H:%M:%S.%fZ")
+        oldest = self._last_index_time.strftime("%Y-%m-%dT%H:%M:%S.%f%z")
         r = self._rocket_chat.call_api_get("rooms.get", updatedSince=oldest)
         json = r.json()
         data = json.get("update")
@@ -112,7 +112,7 @@ def _list_threads(self, channel: RocketchatRoom) -> List[RocketchatThread]:
         return [RocketchatThread(id=trds["_id"], name=trds["msg"], channel_id=trds["rid"]) for trds in data]
 
     def _list_messages(self, channel: RocketchatRoom):
-        oldest = self._last_index_time.strftime("%Y-%m-%dT%H:%M:%S.%fZ")
+        oldest = self._last_index_time.strftime("%Y-%m-%dT%H:%M:%S.%f%z")
         data = []
         while oldest:
             r = self._rocket_chat.call_api_get("chat.syncMessages", roomId=channel.id, lastUpdate=oldest)
@@ -126,7 +126,7 @@ def _list_messages(self, channel: RocketchatRoom):
         return data
 
     def _list_thread_messages(self, thread: RocketchatThread):
-        oldest = self._last_index_time.strftime("%Y-%m-%dT%H:%M:%S.%fZ")
+        oldest = self._last_index_time.strftime("%Y-%m-%dT%H:%M:%S.%f%z")
         data = []
         records = 0
         total = 1  # Set 1 to enter the loop
@@ -186,7 +186,7 @@ def _feed_channel(self, channel):
 
             timestamp = message["ts"]
             message_id = message["_id"]
-            readable_timestamp = datetime.datetime.strptime(timestamp, "%Y-%m-%dT%H:%M:%S.%fZ")
+            readable_timestamp = datetime.datetime.strptime(timestamp, "%Y-%m-%dT%H:%M:%S.%f%z")
             message_url = f"{self._raw_config.get('url')}/{channel.id}?msg={message_id}"
             last_msg = BasicDocument(title=channel.name, content=text, author=author.name,
                                      timestamp=readable_timestamp, id=message_id,