danvics/ytdl-sub
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

[BUGFIX] Fix source metadata getting fetched when yt-dlp breaks

Browse source
This commit is contained in:
Jesse Bannon 2024-01-09 17:39:28 -08:00
parent be0f6f2d2d
commit 1ffbb814e8
1 changed files with 24 additions and 18 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

42
src/ytdl_sub/downloaders/ytdlp.py
View file

@ -216,15 +216,12 @@ class YTDLP:
**kwargs **kwargs
arguments passed directory to YoutubeDL extract_info arguments passed directory to YoutubeDL extract_info
""" """
parent_dict: Dict = {}
try: try:
with cls._listen_and_log_downloaded_info_json( with cls._listen_and_log_downloaded_info_json(
working_directory=working_directory, log_prefix=log_prefix_on_info_json_dl working_directory=working_directory, log_prefix=log_prefix_on_info_json_dl
): ):
# TODO: Fetch parent_dict if it breaks # TODO: Fetch parent_dict if it breaks
parent_dict = cls.extract_info( cls.extract_info(ytdl_options_overrides=ytdl_options_overrides, **kwargs)
ytdl_options_overrides=ytdl_options_overrides, **kwargs
)
except RejectedVideoReached: except RejectedVideoReached:
cls.logger.debug( cls.logger.debug(
"RejectedVideoReached, stopping additional downloads " "RejectedVideoReached, stopping additional downloads "
@ -238,23 +235,32 @@ class YTDLP:
except MaxDownloadsReached: except MaxDownloadsReached:
cls.logger.info("MaxDownloadsReached, stopping additional downloads.") cls.logger.info("MaxDownloadsReached, stopping additional downloads.")
# For YouTube playlists in particular, channel metadata is not fetched. Attempt to get parent_dicts: List[Dict] = []
# channel metadata via grabbing uploader_url info json a max of 3 times entry_dicts = cls._get_entry_dicts_from_info_json_files(working_directory=working_directory)
if isinstance(parent_dict, dict): entry_ids = {entry_dict.get("id") for entry_dict in entry_dicts}
current_iter = 0
url = kwargs.get("url") # Try to get additional uploader (source) metadata that yt-dlp does not fetch
uploader_url = parent_dict.get("uploader_url") # in a single request
while current_iter < 3 and uploader_url and url != uploader_url: for entry_dict in entry_dicts:
cls.logger.debug("Attempting to get parent metadata from URL %s", uploader_url) if not (uploader_id := entry_dict.get("uploader_id")):
continue
if uploader_id in entry_ids or not (uploader_url := entry_dict.get("uploader_url")):
continue
cls.logger.debug("Attempting to get parent metadata from URL %s", uploader_url)
try:
parent_dict = cls.extract_info( parent_dict = cls.extract_info(
ytdl_options_overrides=ytdl_options_overrides | {"playlist_items": "0:0"}, ytdl_options_overrides=ytdl_options_overrides | {"playlist_items": "0:0"},
url=uploader_url, url=uploader_url,
) )
except Exception: # pylint: disable=broad-except
# Do not try this uploader_id again
entry_ids.add(uploader_id)
break
if not isinstance(parent_dict, dict): if isinstance(parent_dict, dict):
break parent_dicts.append(parent_dict)
current_iter += 1 entry_ids |= {uploader_id, parent_dict.get("id")}
url = uploader_url
uploader_url = parent_dict.get("uploader_url")
return cls._get_entry_dicts_from_info_json_files(working_directory=working_directory) return entry_dicts + parent_dicts