danvics/soulsync
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 1

#863: parallelize YouTube artist-recovery (5-worker pool) so the whole playlist resolves within the budget, not just the first ~dozen

Browse source
This commit is contained in:
BoulderBadgeDad 2026-06-12 23:36:31 -07:00
parent 0a6325a87b
commit 0093af89d2
1 changed files with 34 additions and 15 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

49
web_server.py
View file

@ -14861,21 +14861,40 @@ def parse_youtube_playlist(url):
cookie_opts = _youtube_cookie_opts() cookie_opts = _youtube_cookie_opts()
deadline = time.time() + 75 # leave headroom under the 120s worker timeout deadline = time.time() + 75 # leave headroom under the 120s worker timeout
recovered = 0 recovered = 0
for t in unresolved: # Per-video lookups are several seconds each, so doing them
if time.time() >= deadline: # sequentially only covered ~a dozen tracks before the budget ran
logger.warning( # out (the back half of a playlist stayed "Unknown Artist"). Run a
f"[YT Parse] Artist-recovery budget hit after {recovered} tracks; " # small pool concurrently to cover the whole playlist within the
f"{len(unresolved) - recovered} left as Unknown for discovery to resolve" # same wall-clock budget, without hammering YouTube.
) from concurrent.futures import (ThreadPoolExecutor, as_completed,
break TimeoutError as _FuturesTimeout)
raw_artist = _fetch_youtube_video_artist(t['id'], cookie_opts) ex = ThreadPoolExecutor(max_workers=5, thread_name_prefix="yt_artist")
if raw_artist: fut_to_track = {ex.submit(_fetch_youtube_video_artist, t['id'], cookie_opts): t
ca = clean_youtube_artist(raw_artist) for t in unresolved}
if ca and ca != 'Unknown Artist': try:
t['artists'] = [ca] for fut in as_completed(fut_to_track, timeout=max(1.0, deadline - time.time())):
t['name'] = clean_youtube_track_title(t.get('name') or '', ca) t = fut_to_track[fut]
t['raw_artist'] = raw_artist try:
recovered += 1 raw_artist = fut.result()
except Exception:
raw_artist = ''
if raw_artist:
ca = clean_youtube_artist(raw_artist)
if ca and ca != 'Unknown Artist':
t['artists'] = [ca]
t['name'] = clean_youtube_track_title(t.get('name') or '', ca)
t['raw_artist'] = raw_artist
recovered += 1
except _FuturesTimeout:
logger.warning(
f"[YT Parse] Artist-recovery budget hit; recovered {recovered}/{len(unresolved)}, "
"rest left for discovery to resolve"
)
finally:
# Don't block the response waiting on in-flight lookups past
# the budget — cancel what hasn't started, let the rest finish
# in the background.
ex.shutdown(wait=False, cancel_futures=True)
if recovered: if recovered:
logger.info(f"[YT Parse] Recovered artist for {recovered}/{len(unresolved)} unresolved tracks") logger.info(f"[YT Parse] Recovered artist for {recovered}/{len(unresolved)} unresolved tracks")