feat(scripts): merge_exact_title_duration --playback-only + progress logging

--playback-only restricts to scenes with live playback (app-visible dupes only).
Progress print every 500 merges for long global runs.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
jtrzupek 2026-06-08 11:02:19 +02:00
parent e23e2d1f17
commit a9545a7ab2

View file

@ -26,19 +26,26 @@ from app.db import session_scope
from app.resolve.scene_merge import merge_scenes
def _args() -> tuple[str | None, bool]:
def _args() -> tuple[str | None, bool, bool]:
commit = "--commit" in sys.argv
playback_only = "--playback-only" in sys.argv
pid = None
for a in sys.argv[1:]:
if a != "--commit" and len(a) >= 32:
if not a.startswith("--") and len(a) >= 32:
pid = a
return pid, commit
return pid, commit, playback_only
def _groups(pid: str | None) -> list[list[str]]:
def _groups(pid: str | None, playback_only: bool = False) -> list[list[str]]:
# Grupy scen (per performer) o identycznym lower(trim(title)) + duration_sec.
# member order: refs DESC, srcs DESC, created_at ASC → pierwszy = keeper.
where_perf = "AND sp.performer_id = :pid" if pid else ""
# app-visible: tylko sceny z żywym playbackiem (to co user faktycznie widzi na
# stronach) — pomija canonical stuby bez tube-linków.
where_pb = (
"AND EXISTS (SELECT 1 FROM playback_sources p WHERE p.scene_id=s.id AND p.dead_at IS NULL)"
if playback_only else ""
)
sql = f"""
WITH cand AS (
SELECT s.id,
@ -50,7 +57,7 @@ def _groups(pid: str | None) -> list[list[str]]:
(SELECT count(*) FROM playback_sources p WHERE p.scene_id=s.id) srcs
FROM scenes s
JOIN scene_performers sp ON sp.scene_id=s.id {where_perf}
WHERE s.duration_sec IS NOT NULL AND btrim(s.title) <> ''
WHERE s.duration_sec IS NOT NULL AND btrim(s.title) <> '' {where_pb}
)
SELECT array_agg(id::text ORDER BY refs DESC, srcs DESC, created_at ASC) members
FROM cand
@ -73,11 +80,17 @@ def _groups(pid: str | None) -> list[list[str]]:
def main() -> None:
pid, commit = _args()
groups = _groups(pid)
pid, commit, playback_only = _args()
groups = _groups(pid, playback_only)
pairs = sum(len(g) - 1 for g in groups)
print(f"performer={pid or 'ALL'} groups={len(groups)} merges={pairs} commit={commit}", flush=True)
print(
f"performer={pid or 'ALL'} playback_only={playback_only} "
f"groups={len(groups)} merges={pairs} commit={commit}",
flush=True,
)
import uuid as _u
merged = 0
errors = 0
for g in groups:
keep = g[0]
for drop in g[1:]:
@ -86,12 +99,15 @@ def main() -> None:
continue
try:
with session_scope() as s:
import uuid as _u
merge_scenes(s, keep_id=_u.UUID(keep), drop_id=_u.UUID(drop), resolved_by="merge_exact_title_duration")
merged += 1
if merged % 500 == 0:
print(f" progress merged={merged}/{pairs} errors={errors}", flush=True)
except Exception as e:
print(f" ERR keep {keep[:8]} drop {drop[:8]}: {e}")
print(f"DONE merged={merged}/{pairs}", flush=True)
errors += 1
if errors <= 20:
print(f" ERR keep {keep[:8]} drop {drop[:8]}: {str(e)[:120]}")
print(f"DONE merged={merged}/{pairs} errors={errors}", flush=True)
if __name__ == "__main__":