Fix: metadata cache tables silently missing after DB recovery (stale migration marker)

Nothing was landing in the metadata cache browser because the
metadata_cache_entities / metadata_cache_searches tables did not exist, so every
cache write no-op-ed. Root cause: _add_metadata_cache_tables short-circuited on a
marker-only guard (if the metadata_cache_v1 marker row exists, return). After a
DB corruption-recovery the small metadata table (with the marker) survived but
the large cache tables did not, so the stale marker permanently blocked the
idempotent CREATE TABLE IF NOT EXISTS and the cache was dead forever.

Guard now skips only when the marker is set AND the tables actually exist, so a
stale marker self-heals: the tables are re-created on the next init.

Tests: marker present but tables dropped -> re-created; marker + tables present
-> no-op (idempotent).
This commit is contained in:
BoulderBadgeDad 2026-05-31 23:26:49 -07:00
parent 482d5fbc79
commit efe3895d5d
2 changed files with 64 additions and 2 deletions

View file

@ -4267,9 +4267,17 @@ class MusicDatabase:
def _add_metadata_cache_tables(self, cursor):
"""Create metadata_cache_entities and metadata_cache_searches tables for universal API response caching"""
try:
# Skip only when the marker is set AND the tables actually exist.
# A marker-only guard is fragile: if the `metadata` table survives a
# corruption-recovery but the (large) cache tables don't, the stale
# marker would permanently short-circuit creation and the metadata
# cache would silently never work again (nothing lands in the
# browser). Verifying the table presence makes this self-heal.
cursor.execute("SELECT name FROM sqlite_master WHERE type='table' AND name='metadata_cache_entities'")
tables_present = cursor.fetchone() is not None
cursor.execute("SELECT value FROM metadata WHERE key = 'metadata_cache_v1' LIMIT 1")
if cursor.fetchone():
return # Already migrated
if cursor.fetchone() and tables_present:
return # Already migrated and tables present
logger.info("Creating metadata cache tables...")

View file

@ -0,0 +1,54 @@
"""Regression: metadata cache tables must self-heal when the migration marker
is stale.
After a DB corruption-recovery the `metadata` table (with the
'metadata_cache_v1' marker) can survive while the large
metadata_cache_entities/searches tables do not. A marker-only guard then
permanently skips re-creating them, so the cache silently stops working and the
browser shows nothing. _add_metadata_cache_tables must re-create the tables when
the marker is present but the tables are gone.
"""
from __future__ import annotations
from database.music_database import MusicDatabase
def _tables(cur):
return {r[0] for r in cur.execute(
"SELECT name FROM sqlite_master WHERE type='table' AND name LIKE 'metadata_cache_%'"
).fetchall()}
def test_recreates_cache_tables_when_marker_stale(tmp_path):
db = MusicDatabase(str(tmp_path / "m.db"))
conn = db._get_connection()
cur = conn.cursor()
# Fresh DB: tables + marker present.
assert 'metadata_cache_entities' in _tables(cur)
assert cur.execute("SELECT value FROM metadata WHERE key='metadata_cache_v1'").fetchone()
# Simulate corruption-recovery: marker survives, cache tables don't.
cur.execute("DROP TABLE metadata_cache_entities")
cur.execute("DROP TABLE metadata_cache_searches")
conn.commit()
assert 'metadata_cache_entities' not in _tables(cur)
assert cur.execute("SELECT value FROM metadata WHERE key='metadata_cache_v1'").fetchone() # stale marker
# Re-run the migration — must self-heal despite the stale marker.
db._add_metadata_cache_tables(cur)
conn.commit()
assert 'metadata_cache_entities' in _tables(cur)
assert 'metadata_cache_searches' in _tables(cur)
def test_skips_when_marker_and_tables_both_present(tmp_path):
# Idempotent: a healthy DB shouldn't error on re-run.
db = MusicDatabase(str(tmp_path / "m2.db"))
conn = db._get_connection()
cur = conn.cursor()
before = _tables(cur)
db._add_metadata_cache_tables(cur) # no-op fast path
conn.commit()
assert _tables(cur) == before