Files
immich-photo-flow/apps/trip-cluster/app/ingest.py
T
m038 bd8fe59b49 fix(ingest): don't advance incremental cursor past a failed/sampled asset
A later successful asset could push last_ingest_at past an earlier asset whose
thumbnail download failed, permanently excluding it from later incremental runs
(only --full recovered it). Now the cursor never advances to/past the earliest
failed asset, never below the prior cursor, and not at all on a --subset run.
Regression tests added. (review finding #9)
2026-06-27 18:49:24 +02:00

103 lines
3.9 KiB
Python

import datetime
import logging
import os
from photoflow.core.models import Asset
from photoflow.immich import pipeline
log = logging.getLogger(__name__)
def _now() -> str:
return datetime.datetime.now(datetime.timezone.utc).isoformat()
def run_ingest(client, store, thumbs_dir, *, date_from=None, date_to=None,
tag=None, subset=None, full=False) -> dict:
os.makedirs(thumbs_dir, exist_ok=True)
tag_ids = None
if tag:
tid = client.resolve_tag_id(tag)
if not tid:
raise ValueError(f"Tag {tag!r} not found in Immich")
tag_ids = [tid]
updated_after = None if full else store.get_meta("last_ingest_at")
assets = client.search_assets(taken_after=date_from, taken_before=date_to,
tag_ids=tag_ids, updated_after=updated_after)
if subset is not None:
assets = assets[:subset]
processed_marked = 0
tag_counts: dict = {}
successful_updated: list = [] # updated_at of assets ingested this run
failed_updated: list = [] # updated_at of assets we could NOT ingest
for a in assets:
# Revision (2): sanitize the Immich-provided id before using it as a
# path component, guarding against path traversal (mirror /thumb proxy).
safe_id = os.path.basename(a["id"])
thumb_path = os.path.join(thumbs_dir, f"{safe_id}.jpg")
# Revision (1): treat an existing 0-byte file as missing.
need_thumb = (not os.path.exists(thumb_path)
or os.path.getsize(thumb_path) == 0)
if need_thumb:
try:
data = client.download_thumbnail(a["id"])
except Exception:
log.exception("Thumbnail download failed for asset %s", a["id"])
failed_updated.append(a["updated_at"])
continue
# Write to a temp path and os.replace into place so a failure
# never leaves a 0-byte <id>.jpg behind; clean up the temp file
# if the write itself fails (e.g. disk full).
tmp_path = f"{thumb_path}.tmp"
try:
with open(tmp_path, "wb") as f:
f.write(data)
os.replace(tmp_path, thumb_path)
except Exception:
log.exception("Thumbnail write failed for asset %s", a["id"])
try:
os.unlink(tmp_path)
except OSError:
pass
failed_updated.append(a["updated_at"])
continue
store.upsert_asset(Asset(
immich_id=a["id"], taken_at=a["taken_at"], gps_lat=a["gps_lat"],
gps_lon=a["gps_lon"], place_city=a["place_city"],
place_country=a["place_country"], type=a["type"],
thumb_path=thumb_path, ingested_at=_now(), updated_at=a["updated_at"]))
store.set_asset_tags(a["id"], a["tags"])
for t in a["tags"]:
tag_counts[t] = tag_counts.get(t, 0) + 1
if pipeline.PROCESSED in a["tags"]:
store.mark_processed(a["id"])
processed_marked += 1
successful_updated.append(a["updated_at"])
for name, count in tag_counts.items():
store.upsert_tag(name, count=count)
# Advance the incremental cursor, but never past an asset we failed to ingest
# (a later success must not strand an earlier failure on the next run), never
# below the prior cursor, and not at all on a --subset sampling run.
if subset is None:
new_watermark = updated_after or ""
floor = min((u for u in failed_updated if u), default=None)
for u in successful_updated:
if not u or (floor is not None and u >= floor):
continue
if u > new_watermark:
new_watermark = u
if new_watermark:
store.set_meta("last_ingest_at", new_watermark)
return {"fetched": len(assets), "processed_marked": processed_marked}