Ретраи и дедупликация в задачах
- элемент плейлиста повторяется до 3 раз с паузой 30 с (разовые 403/429 от YouTube) - yt-dlp: retries/fragment_retries/extractor_retries - уже обработанные URL пропускаются — повторная постановка плейлиста доделывает только упавшее Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
@@ -10,10 +10,15 @@ import asyncio
|
||||
import logging
|
||||
|
||||
from api.db.job import get_job, update_job
|
||||
from api.lib.processing import process_single_video
|
||||
from api.db.video import get_video_by_url
|
||||
from api.lib.processing import PipelineError, process_single_video
|
||||
from api.lib.youtube import expand_url
|
||||
|
||||
|
||||
_ITEM_ATTEMPTS = 3 # попытки на один элемент (YouTube любит разовые 403/429)
|
||||
_RETRY_DELAY = 30.0
|
||||
|
||||
|
||||
logger = logging.getLogger("jobs")
|
||||
|
||||
_queue: asyncio.Queue[str] = asyncio.Queue()
|
||||
@@ -95,21 +100,43 @@ async def _run_job(job_id: str) -> None:
|
||||
def _notify(msg: str, _label: str = label) -> None:
|
||||
_live[job_id] = f"{_label} — {msg}"
|
||||
|
||||
try:
|
||||
await process_single_video(
|
||||
entry["url"],
|
||||
section_id=params.get("section_id"),
|
||||
model_path=params.get("model_path"),
|
||||
summary_max_sentences=params.get("summary_max_sentences"),
|
||||
keep_video=params.get("keep_video"),
|
||||
job_id=job_id,
|
||||
progress=_notify,
|
||||
)
|
||||
# дедупликация: повторная постановка плейлиста не переделывает готовое
|
||||
if await get_video_by_url(entry["url"]) is not None:
|
||||
done += 1
|
||||
except Exception as e:
|
||||
failed += 1
|
||||
errors.append(f"{entry['title']}: {e}")
|
||||
logger.warning("job %s item %d/%d failed: %s", job_id, i, n, e)
|
||||
_live[job_id] = f"{label} — уже в библиотеке, пропуск"
|
||||
logger.info("job %s item %d/%d already processed, skip", job_id, i, n)
|
||||
await update_job(job_id, done_items=done)
|
||||
continue
|
||||
|
||||
for attempt in range(1, _ITEM_ATTEMPTS + 1):
|
||||
try:
|
||||
await process_single_video(
|
||||
entry["url"],
|
||||
section_id=params.get("section_id"),
|
||||
model_path=params.get("model_path"),
|
||||
summary_max_sentences=params.get("summary_max_sentences"),
|
||||
keep_video=params.get("keep_video"),
|
||||
job_id=job_id,
|
||||
progress=_notify,
|
||||
)
|
||||
done += 1
|
||||
break
|
||||
except Exception as e:
|
||||
retryable = not (isinstance(e, PipelineError) and e.status == 400)
|
||||
if retryable and attempt < _ITEM_ATTEMPTS and not is_cancelled(job_id):
|
||||
logger.warning(
|
||||
"job %s item %d/%d attempt %d failed, retrying: %s",
|
||||
job_id, i, n, attempt, e,
|
||||
)
|
||||
_live[job_id] = (
|
||||
f"{label} — ошибка, повтор {attempt}/{_ITEM_ATTEMPTS - 1} через {int(_RETRY_DELAY)} с"
|
||||
)
|
||||
await asyncio.sleep(_RETRY_DELAY)
|
||||
continue
|
||||
failed += 1
|
||||
errors.append(f"{entry['title']}: {e}")
|
||||
logger.warning("job %s item %d/%d failed: %s", job_id, i, n, e)
|
||||
break
|
||||
await update_job(
|
||||
job_id, done_items=done, failed_items=failed, error="\n".join(errors)[:4000]
|
||||
)
|
||||
|
||||
Reference in New Issue
Block a user