fix(stats): review fuzzy AniList duplicates before merging

- Preserve merged title aliases for future episodes
- Fail closed when queued writes cannot drain
This commit is contained in:
2026-08-11 23:11:39 -07:00
parent d82a4fd8c6
commit d8c5fc321a
28 changed files with 1625 additions and 44 deletions
@@ -35,8 +35,85 @@ interface TrackerInternals {
recordWrite: (write: Record<string, unknown>) => void;
mergeAnime: (targetAnimeId: number, sourceAnimeIds: number[]) => Promise<unknown>;
moveVideoToAnime: (videoId: number, targetAnimeId: number) => Promise<unknown>;
rebuildLifetimeSummaries: () => Promise<unknown>;
flushNow: () => void;
}
test('mergeAnime fails closed when queued writes cannot drain', async () => {
const dbPath = makeDbPath();
let tracker: ImmersionTrackerService | null = null;
try {
const Ctor = await loadTrackerCtor();
tracker = new Ctor({ dbPath, policy: { batchSize: 2 } });
const internals = tracker as unknown as TrackerInternals;
seedTwoEntries(internals.db);
queueSubtitleLines(internals, 1);
internals.flushNow = () => {};
await assert.rejects(internals.mergeAnime(1, [2]), /queue did not drain/i);
assert.deepEqual(
internals.db
.prepare('SELECT anime_id AS animeId FROM imm_anime ORDER BY anime_id')
.all()
.map((row) => (row as { animeId: number }).animeId),
[1, 2],
);
} finally {
tracker?.destroy();
cleanupDbPath(dbPath);
}
});
test('moveVideoToAnime fails closed when queued writes cannot drain', async () => {
const dbPath = makeDbPath();
let tracker: ImmersionTrackerService | null = null;
try {
const Ctor = await loadTrackerCtor();
tracker = new Ctor({ dbPath, policy: { batchSize: 2 } });
const internals = tracker as unknown as TrackerInternals;
seedTwoEntries(internals.db);
queueSubtitleLines(internals, 1);
internals.flushNow = () => {};
await assert.rejects(internals.moveVideoToAnime(2, 1), /queue did not drain/i);
assert.equal(
(
internals.db
.prepare('SELECT anime_id AS animeId FROM imm_videos WHERE video_id = 2')
.get() as {
animeId: number;
}
).animeId,
2,
);
} finally {
tracker?.destroy();
cleanupDbPath(dbPath);
}
});
test('rebuildLifetimeSummaries fails closed when queued writes cannot drain', async () => {
const dbPath = makeDbPath();
let tracker: ImmersionTrackerService | null = null;
try {
const Ctor = await loadTrackerCtor();
tracker = new Ctor({ dbPath, policy: { batchSize: 2 } });
const internals = tracker as unknown as TrackerInternals;
seedTwoEntries(internals.db);
queueSubtitleLines(internals, 1);
internals.flushNow = () => {};
await assert.rejects(internals.rebuildLifetimeSummaries(), /queue did not drain/i);
} finally {
tracker?.destroy();
cleanupDbPath(dbPath);
}
});
function seedTwoEntries(db: DatabaseSync): void {
db.exec(`
INSERT INTO imm_anime (anime_id, normalized_title_key, canonical_title, CREATED_DATE, LAST_UPDATE_DATE)
@@ -1053,6 +1053,55 @@ describe('stats server API routes', () => {
assert.equal(body[0].canonicalTitle, 'Little Witch Academia');
});
it('GET /api/stats/anime/merge-recommendations returns pending duplicate pairs', async () => {
const app = createStatsApp(
createMockTracker({
getAnimeMergeRecommendations: async () => [{ recommendationId: 4, animeIds: [1, 2] }],
} as Partial<ImmersionTrackerService>),
);
const res = await app.request('/api/stats/anime/merge-recommendations');
assert.equal(res.status, 200);
assert.deepEqual(await res.json(), {
recommendations: [{ recommendationId: 4, animeIds: [1, 2] }],
});
});
it('DELETE /api/stats/anime/merge-recommendations/:id dismisses a pending pair', async () => {
let dismissedId: number | null = null;
const app = createStatsApp(
createMockTracker({
dismissAnimeMergeRecommendation: async (recommendationId: number) => {
dismissedId = recommendationId;
return true;
},
} as Partial<ImmersionTrackerService>),
);
const res = await app.request('/api/stats/anime/merge-recommendations/4', {
method: 'DELETE',
});
assert.equal(res.status, 200);
assert.equal(dismissedId, 4);
assert.deepEqual(await res.json(), { ok: true });
});
it('DELETE /api/stats/anime/merge-recommendations/:id reports missing recommendations', async () => {
const app = createStatsApp(
createMockTracker({
dismissAnimeMergeRecommendation: async () => false,
} as Partial<ImmersionTrackerService>),
);
const res = await app.request('/api/stats/anime/merge-recommendations/99', {
method: 'DELETE',
});
assert.equal(res.status, 404);
});
it('GET /api/stats/anime/:animeId returns anime detail with episodes', async () => {
const app = createStatsApp(createMockTracker());
const res = await app.request('/api/stats/anime/1');
@@ -327,6 +327,7 @@ export function createCoverArtFetcher(
titleEnglish: selected.title?.english ?? null,
titleNative: selected.title?.native ?? null,
episodesTotal: selected.episodes ?? null,
exactTitleMatch: resolution?.exactTitleMatch ?? false,
});
logger.info(
@@ -156,6 +156,37 @@ test('season 1 resolves to the anchor without relation lookups', async () => {
assert.deepEqual(relationLookups, []);
});
test('reports an exact normalized synonym match as strong evidence', async () => {
const { execute } = createExecutor([
{
id: 1,
episodes: 12,
format: 'TV',
title: { english: 'Hitori Gotoh Story' },
synonyms: ['BOCCHI THE ROCK'],
},
]);
const result = await resolveAnilistSeasonMedia({ title: 'Bocchi the Rock!' }, { execute });
assert.equal(result?.exactTitleMatch, true);
});
test('reports a fuzzy-only search result as weak evidence', async () => {
const { execute } = createExecutor([
{
id: 1,
episodes: 12,
format: 'TV',
title: { english: 'Actual Show' },
},
]);
const result = await resolveAnilistSeasonMedia({ title: 'Unrelated Release' }, { execute });
assert.equal(result?.exactTitleMatch, false);
});
test('strips a season marker already present in the parsed title', async () => {
const { execute, searches } = createExecutor(OREGAIRU_SEARCH, OREGAIRU_RELATIONS);
const result = await resolveAnilistSeasonMedia(
+15 -5
View File
@@ -42,6 +42,8 @@ export interface AnilistSeasonResolution {
seasonResolved: boolean;
requestedSeason: number | null;
via: AnilistSeasonResolutionVia;
/** Exact normalized match against an AniList title or synonym. */
exactTitleMatch: boolean;
}
export interface ResolveAnilistSeasonMediaInput {
@@ -116,7 +118,12 @@ const SEASONAL_FORMAT_PRIORITY = ['TV', 'TV_SHORT', 'ONA'];
const MAX_SEQUEL_HOPS = 12;
function normalizeTitle(value: string): string {
return value.trim().toLowerCase().replace(/\s+/g, ' ');
return value
.normalize('NFKC')
.toLowerCase()
.replace(/[^\p{L}\p{N}]+/gu, ' ')
.trim()
.replace(/\s+/g, ' ');
}
/**
@@ -176,6 +183,7 @@ function toResolution(
season: number | null,
via: AnilistSeasonResolutionVia,
seasonResolved: boolean,
exactTitleMatch: boolean,
): AnilistSeasonResolution {
return {
id: media.id,
@@ -185,6 +193,7 @@ function toResolution(
seasonResolved,
requestedSeason: season,
via,
exactTitleMatch,
};
}
@@ -367,9 +376,10 @@ export async function resolveAnilistSeasonMedia(
episode: season === null || season <= 1 ? input.episode : null,
});
if (!anchor) return null;
const exactTitleMatch = mediaTitles(anchor).includes(normalizeTitle(searchTitle));
if (season === null || season <= 1) {
return toResolution(anchor, searchTitle, season, 'anchor', true);
return toResolution(anchor, searchTitle, season, 'anchor', true, exactTitleMatch);
}
let chainError: unknown = null;
@@ -383,7 +393,7 @@ export async function resolveAnilistSeasonMedia(
deps.logInfo?.(
`[anilist] season ${season} of "${searchTitle}" resolved via sequel chain: ${displayTitle(viaChain, searchTitle)} (${viaChain.id})`,
);
return toResolution(viaChain, searchTitle, season, 'sequel-chain', true);
return toResolution(viaChain, searchTitle, season, 'sequel-chain', true, exactTitleMatch);
}
const viaAirOrder = pickByAirOrder(anchor, season, media);
@@ -391,7 +401,7 @@ export async function resolveAnilistSeasonMedia(
deps.logInfo?.(
`[anilist] season ${season} of "${searchTitle}" resolved via air order: ${displayTitle(viaAirOrder, searchTitle)} (${viaAirOrder.id})`,
);
return toResolution(viaAirOrder, searchTitle, season, 'air-order', true);
return toResolution(viaAirOrder, searchTitle, season, 'air-order', true, exactTitleMatch);
}
// The chain failed for transport reasons rather than because the season is absent;
@@ -403,5 +413,5 @@ export async function resolveAnilistSeasonMedia(
deps.logInfo?.(
`[anilist] could not resolve season ${season} of "${searchTitle}"; falling back to ${displayTitle(anchor, searchTitle)} (${anchor.id})`,
);
return toResolution(anchor, searchTitle, season, 'anchor', false);
return toResolution(anchor, searchTitle, season, 'anchor', false, exactTitleMatch);
}
+24 -5
View File
@@ -93,8 +93,11 @@ import {
} from './immersion-tracker/query-maintenance';
import { repairJellyfinStreamVideoLinks } from './immersion-tracker/jellyfin-link-repair';
import {
dismissAnimeMergeRecommendation,
getAnimeMergeRecommendations,
repairLegacySeasonlessAnimeRows,
resolveAnimeAnilistConflict,
type AnimeMergeRecommendation,
} from './immersion-tracker/anime-season-repair';
import {
mergeAnimeRecords,
@@ -602,7 +605,7 @@ export class ImmersionTrackerService {
}
async rebuildLifetimeSummaries(): Promise<LifetimeRebuildSummary> {
this.drainWriteQueue('rebuilding lifetime summaries');
this.requireWriteQueueDrained('rebuilding lifetime summaries');
return rebuildLifetimeSummaryTables(this.db);
}
@@ -669,6 +672,14 @@ export class ImmersionTrackerService {
return getAnimeLibrary(this.db);
}
async getAnimeMergeRecommendations(): Promise<AnimeMergeRecommendation[]> {
return getAnimeMergeRecommendations(this.db);
}
async dismissAnimeMergeRecommendation(recommendationId: number): Promise<boolean> {
return dismissAnimeMergeRecommendation(this.db, recommendationId);
}
async getAnimeDetail(animeId: number): Promise<AnimeDetailRow | null> {
this.relinkYoutubeAnimeLibrary();
return getAnimeDetail(this.db, animeId);
@@ -774,13 +785,13 @@ export class ImmersionTrackerService {
// This rebuilds the lifetime summaries, which recompute from the database:
// queued writes have to land first or the active session is dropped from
// the merged totals.
this.drainWriteQueue('merging library entries');
this.requireWriteQueueDrained('merging library entries');
return mergeAnimeRecords(this.db, targetAnimeId, sourceAnimeIds);
}
async moveVideoToAnime(videoId: number, targetAnimeId: number): Promise<VideoMoveSummary> {
await this.pendingAnimeMetadataUpdates.get(videoId);
this.drainWriteQueue('moving an episode');
this.requireWriteQueueDrained('moving an episode');
return moveVideoToAnimeQuery(this.db, videoId, targetAnimeId);
}
@@ -794,8 +805,8 @@ export class ImmersionTrackerService {
* soon as a pass makes no progress — a rolled-back batch is pushed back onto
* the queue, and looping on that would spin forever.
*
* Returns false when the queue could not be emptied, in which case the
* rebuild runs against a database still missing those writes.
* Returns false when the queue could not be emptied. Summary-rebuilding
* callers fail closed in that case.
*/
private drainWriteQueue(context: string): boolean {
this.flushTelemetry(true);
@@ -812,6 +823,12 @@ export class ImmersionTrackerService {
return true;
}
private requireWriteQueueDrained(context: string): void {
if (!this.drainWriteQueue(context)) {
throw new Error(`Immersion tracker queue did not drain before ${context}`);
}
}
async reassignAnimeAnilist(
animeId: number,
info: {
@@ -828,7 +845,9 @@ export class ImmersionTrackerService {
// another row already claims the same AniList id.
const repair = resolveAnimeAnilistConflict(this.db, animeId, info.anilistId, {
survivor: 'target',
matchConfidence: 'manual',
});
if (repair.anilistAssignmentBlocked) return;
this.db
.prepare(
`
@@ -5,9 +5,14 @@ import path from 'node:path';
import test from 'node:test';
import { Database } from '../sqlite.js';
import type { DatabaseSync } from '../sqlite.js';
import { applyPragmas, ensureSchema } from '../storage.js';
import { applyPragmas, ensureSchema, getOrCreateAnimeRecord } from '../storage.js';
import { mergeAnimeRecords, moveVideoToAnime } from '../anime-merge.js';
import { resolveAnimeAnilistConflict } from '../anime-season-repair.js';
import {
dismissAnimeMergeRecommendation,
getAnimeMergeRecommendations,
resolveAnimeAnilistConflict,
} from '../anime-season-repair.js';
import { updateAnimeAnilistInfo } from '../query-maintenance.js';
const BASE_MS = 1_700_000_000_000;
@@ -211,6 +216,54 @@ test('mergeAnimeRecords inherits metadata the target is missing without clobberi
});
});
test('mergeAnimeRecords preserves source title identities as aliases of the survivor', () => {
withDb((db) => {
insertAnime(db, { animeId: 1, key: 'show', title: 'Show' });
insertAnime(db, { animeId: 2, key: 'show season 1', title: 'Show Season 1' });
insertEpisode(db, { videoId: 1, animeId: 1 });
insertEpisode(db, { videoId: 2, animeId: 2, season: 1 });
db.prepare(
`INSERT INTO imm_anime_title_aliases(normalized_title_key, anime_id, CREATED_DATE, LAST_UPDATE_DATE)
VALUES ('show s01', 2, ?, ?)`,
).run(BASE_MS, BASE_MS);
mergeAnimeRecords(db, 1, [2]);
const fromSourceTitle = getOrCreateAnimeRecord(db, {
parsedTitle: 'Show Season 1',
canonicalTitle: 'Show Season 1',
seasonScope: 1,
anilistId: null,
titleRomaji: null,
titleEnglish: null,
titleNative: null,
metadataJson: null,
});
const fromTransferredAlias = getOrCreateAnimeRecord(db, {
parsedTitle: 'Show S01',
canonicalTitle: 'Show S01',
anilistId: null,
titleRomaji: null,
titleEnglish: null,
titleNative: null,
metadataJson: null,
});
assert.equal(fromSourceTitle, 1);
assert.equal(fromTransferredAlias, 1);
assert.deepEqual(animeIds(db), [1]);
assert.equal(
(
db.prepare('SELECT canonical_title AS title FROM imm_anime WHERE anime_id = 1').get() as {
title: string;
}
).title,
'Show',
);
});
});
test('mergeAnimeRecords ignores unknown targets and self-merges', () => {
withDb((db) => {
insertAnime(db, { animeId: 1, key: 'show', title: 'Show' });
@@ -311,6 +364,126 @@ test('resolveAnimeAnilistConflict folds a seasonless duplicate into the entry th
});
});
test('resolveAnimeAnilistConflict recommends a weak title collision instead of merging it', () => {
withDb((db) => {
insertAnime(db, {
animeId: 1,
key: 'actual show',
title: 'Actual Show',
anilistId: 163132,
titleRomaji: 'Actual Show',
});
insertAnime(db, { animeId: 2, key: 'unrelated release', title: 'Unrelated Release' });
insertEpisode(db, { videoId: 1, animeId: 1, season: 1 });
insertEpisode(db, { videoId: 2, animeId: 2, season: 1 });
const summary = resolveAnimeAnilistConflict(db, 2, 163132);
assert.equal(summary.repaired, 0);
assert.deepEqual(animeIds(db), [1, 2]);
assert.equal(videoAnimeId(db, 2), 2);
assert.deepEqual(getAnimeMergeRecommendations(db), [{ recommendationId: 1, animeIds: [1, 2] }]);
});
});
test('automatic AniList update leaves a weak collision unassigned for user review', () => {
withDb((db) => {
insertAnime(db, {
animeId: 1,
key: 'actual show',
title: 'Actual Show',
anilistId: 163132,
titleRomaji: 'Actual Show',
});
insertAnime(db, { animeId: 2, key: 'unrelated release', title: 'Unrelated Release' });
insertEpisode(db, { videoId: 1, animeId: 1, season: 1 });
insertEpisode(db, { videoId: 2, animeId: 2, season: 1 });
updateAnimeAnilistInfo(db, 2, {
anilistId: 163132,
titleRomaji: 'Actual Show',
titleEnglish: null,
titleNative: null,
episodesTotal: 12,
exactTitleMatch: false,
});
const target = db
.prepare('SELECT anilist_id AS anilistId FROM imm_anime WHERE anime_id = 2')
.get() as {
anilistId: number | null;
};
assert.equal(target.anilistId, null);
assert.deepEqual(getAnimeMergeRecommendations(db), [{ recommendationId: 1, animeIds: [1, 2] }]);
});
});
test('dismissed weak collision stays dismissed when automatic resolution repeats', () => {
withDb((db) => {
insertAnime(db, {
animeId: 1,
key: 'actual show',
title: 'Actual Show',
anilistId: 163132,
titleRomaji: 'Actual Show',
});
insertAnime(db, { animeId: 2, key: 'unrelated release', title: 'Unrelated Release' });
insertEpisode(db, { videoId: 1, animeId: 1, season: 1 });
insertEpisode(db, { videoId: 2, animeId: 2, season: 1 });
resolveAnimeAnilistConflict(db, 2, 163132);
assert.equal(dismissAnimeMergeRecommendation(db, 1), true);
resolveAnimeAnilistConflict(db, 2, 163132);
assert.deepEqual(getAnimeMergeRecommendations(db), []);
});
});
test('dismissed recommendation prevents a later exact automatic merge of the pair', () => {
withDb((db) => {
insertAnime(db, {
animeId: 1,
key: 'actual show',
title: 'Actual Show',
anilistId: 163132,
titleRomaji: 'Actual Show',
});
insertAnime(db, { animeId: 2, key: 'unrelated release', title: 'Unrelated Release' });
insertEpisode(db, { videoId: 1, animeId: 1, season: 1 });
insertEpisode(db, { videoId: 2, animeId: 2, season: 1 });
resolveAnimeAnilistConflict(db, 2, 163132, { matchConfidence: 'weak' });
assert.equal(dismissAnimeMergeRecommendation(db, 1), true);
const summary = resolveAnimeAnilistConflict(db, 2, 163132, { matchConfidence: 'exact' });
assert.equal(summary.repaired, 0);
assert.deepEqual(animeIds(db), [1, 2]);
assert.equal(videoAnimeId(db, 2), 2);
assert.deepEqual(getAnimeMergeRecommendations(db), []);
});
});
test('manual merge clears recommendations involving the absorbed entry', () => {
withDb((db) => {
insertAnime(db, {
animeId: 1,
key: 'actual show',
title: 'Actual Show',
anilistId: 163132,
titleRomaji: 'Actual Show',
});
insertAnime(db, { animeId: 2, key: 'unrelated release', title: 'Unrelated Release' });
insertEpisode(db, { videoId: 1, animeId: 1, season: 1 });
insertEpisode(db, { videoId: 2, animeId: 2, season: 1 });
resolveAnimeAnilistConflict(db, 2, 163132);
mergeAnimeRecords(db, 1, [2]);
assert.deepEqual(getAnimeMergeRecommendations(db), []);
});
});
test('resolveAnimeAnilistConflict keeps the target entry when the user drove the change', () => {
withDb((db) => {
insertAnime(db, { animeId: 1, key: 'show', title: 'Show', anilistId: 163132 });
@@ -353,6 +526,76 @@ test('resolveAnimeAnilistConflict falls back to season redistribution for multi-
});
});
test('resolveAnimeAnilistConflict leaves explicit incompatible seasons and assignments unchanged', () => {
withDb((db) => {
insertAnime(db, {
animeId: 1,
key: 'show season 1',
title: 'Show Season 1',
anilistId: 163132,
titleRomaji: 'Show',
});
insertAnime(db, { animeId: 2, key: 'show season 2', title: 'Show Season 2' });
insertEpisode(db, { videoId: 1, animeId: 1, season: 1 });
insertEpisode(db, { videoId: 2, animeId: 2, season: 2 });
const summary = resolveAnimeAnilistConflict(db, 2, 163132, { matchConfidence: 'exact' });
assert.equal(summary.repaired, 0);
assert.equal(summary.movedVideos, 0);
assert.equal(summary.deletedAnimeRows, 0);
assert.deepEqual(animeIds(db), [1, 2]);
assert.equal(videoAnimeId(db, 1), 1);
assert.equal(videoAnimeId(db, 2), 2);
const assignments = db
.prepare(
'SELECT anime_id AS animeId, anilist_id AS anilistId FROM imm_anime ORDER BY anime_id',
)
.all() as Array<{ animeId: number; anilistId: number | null }>;
assert.deepEqual(assignments, [
{ animeId: 1, anilistId: 163132 },
{ animeId: 2, anilistId: null },
]);
assert.deepEqual(getAnimeMergeRecommendations(db), []);
});
});
test('automatic AniList update does not transfer an assignment across explicit seasons', () => {
withDb((db) => {
insertAnime(db, {
animeId: 1,
key: 'show season 1',
title: 'Show Season 1',
anilistId: 163132,
titleRomaji: 'Show',
});
insertAnime(db, { animeId: 2, key: 'show season 2', title: 'Show Season 2' });
insertEpisode(db, { videoId: 1, animeId: 1, season: 1 });
insertEpisode(db, { videoId: 2, animeId: 2, season: 2 });
updateAnimeAnilistInfo(db, 2, {
anilistId: 163132,
titleRomaji: 'Show',
titleEnglish: null,
titleNative: null,
episodesTotal: 12,
exactTitleMatch: true,
});
const assignments = db
.prepare(
'SELECT anime_id AS animeId, anilist_id AS anilistId FROM imm_anime ORDER BY anime_id',
)
.all() as Array<{ animeId: number; anilistId: number | null }>;
assert.deepEqual(assignments, [
{ animeId: 1, anilistId: 163132 },
{ animeId: 2, anilistId: null },
]);
assert.equal(videoAnimeId(db, 1), 1);
assert.equal(videoAnimeId(db, 2), 2);
});
});
test('resolveAnimeAnilistConflict leaves an entry that already links elsewhere alone', () => {
withDb((db) => {
insertAnime(db, { animeId: 1, key: 'show', title: 'Show', anilistId: 163132 });
@@ -360,9 +603,20 @@ test('resolveAnimeAnilistConflict leaves an entry that already links elsewhere a
insertEpisode(db, { videoId: 1, animeId: 1, season: 1 });
insertEpisode(db, { videoId: 2, animeId: 2, season: 2 });
resolveAnimeAnilistConflict(db, 2, 163132);
const summary = resolveAnimeAnilistConflict(db, 2, 163132);
assert.equal(videoAnimeId(db, 2), 2);
assert.ok(animeIds(db).includes(2));
assert.equal(
(
db.prepare('SELECT anilist_id AS anilistId FROM imm_anime WHERE anime_id = 2').get() as {
anilistId: number;
}
).anilistId,
999,
);
assert.equal(summary.repaired, 0);
assert.equal(summary.movedVideos, 0);
assert.deepEqual(getAnimeMergeRecommendations(db), []);
});
});
@@ -0,0 +1,169 @@
import type { DatabaseSync } from './sqlite';
import { animeSeasonsAreMergeCompatible, getParsedSeasonsForAnime } from './anime-merge';
import { toDbTimestamp } from './query-shared';
import { normalizeAnimeIdentityKey } from './storage';
import { nowMs } from './time';
export interface AnimeMergeRecommendation {
recommendationId: number;
animeIds: [number, number];
}
export interface AnimeConflictRecommendationOptions {
survivor?: 'target' | 'existing';
/** Automatic matches must be exact; manual assignment is authoritative. */
matchConfidence?: 'exact' | 'weak' | 'manual';
}
interface AnimeTitleRow {
canonical_title: string;
title_romaji: string | null;
title_english: string | null;
title_native: string | null;
}
function getAnimeTitles(db: DatabaseSync, animeId: number): AnimeTitleRow | null {
return db
.prepare(
`SELECT canonical_title, title_romaji, title_english, title_native
FROM imm_anime
WHERE anime_id = ?`,
)
.get(animeId) as AnimeTitleRow | null;
}
function getParsedTitles(db: DatabaseSync, animeId: number): Array<string | null> {
return (
db.prepare('SELECT parsed_title FROM imm_videos WHERE anime_id = ?').all(animeId) as Array<{
parsed_title: string | null;
}>
).map((row) => row.parsed_title);
}
function stripSeasonIdentitySuffix(title: string): string {
return title
.replace(/\bseason\s*\d{1,2}\b/gi, ' ')
.replace(/\b\d{1,2}(?:st|nd|rd|th)\s+season\b/gi, ' ')
.replace(/\bs\d{1,2}\b/gi, ' ');
}
export function hasExactStoredTitleMatch(
db: DatabaseSync,
targetAnimeId: number,
conflictAnimeId: number,
): boolean {
const target = getAnimeTitles(db, targetAnimeId);
const conflict = getAnimeTitles(db, conflictAnimeId);
if (!target || !conflict) return false;
const targetKeys = [target.canonical_title, ...getParsedTitles(db, targetAnimeId)]
.filter((title): title is string => Boolean(title?.trim()))
.map((title) => normalizeAnimeIdentityKey(stripSeasonIdentitySuffix(title)))
.filter(Boolean);
const anilistTitleKeys = [
conflict.title_romaji,
conflict.title_english,
conflict.title_native,
conflict.canonical_title,
]
.filter((title): title is string => Boolean(title?.trim()))
.map(normalizeAnimeIdentityKey)
.filter(Boolean);
return targetKeys.some((key) => anilistTitleKeys.includes(key));
}
export function shouldRecommendAnilistConflict(
db: DatabaseSync,
targetAnimeId: number,
conflictAnimeId: number,
options: AnimeConflictRecommendationOptions,
): boolean {
if (options.survivor === 'target' || options.matchConfidence === 'manual') return false;
if (
!animeSeasonsAreMergeCompatible(
getParsedSeasonsForAnime(db, targetAnimeId),
getParsedSeasonsForAnime(db, conflictAnimeId),
)
) {
return false;
}
return (
options.matchConfidence === 'weak' ||
(options.matchConfidence === undefined &&
!hasExactStoredTitleMatch(db, targetAnimeId, conflictAnimeId))
);
}
export function recordAnimeMergeRecommendation(
db: DatabaseSync,
firstCandidateAnimeId: number,
secondCandidateAnimeId: number,
anilistId: number,
): void {
const firstAnimeId = Math.min(firstCandidateAnimeId, secondCandidateAnimeId);
const secondAnimeId = Math.max(firstCandidateAnimeId, secondCandidateAnimeId);
const timestamp = toDbTimestamp(nowMs());
db.prepare(
`INSERT INTO imm_anime_merge_recommendations(
first_anime_id, second_anime_id, anilist_id, status, CREATED_DATE, LAST_UPDATE_DATE
) VALUES (?, ?, ?, 'pending', ?, ?)
ON CONFLICT(first_anime_id, second_anime_id, anilist_id) DO UPDATE SET
LAST_UPDATE_DATE = excluded.LAST_UPDATE_DATE`,
).run(firstAnimeId, secondAnimeId, anilistId, timestamp, timestamp);
}
export function hasDismissedAnimeMergeRecommendation(
db: DatabaseSync,
firstCandidateAnimeId: number,
secondCandidateAnimeId: number,
): boolean {
const firstAnimeId = Math.min(firstCandidateAnimeId, secondCandidateAnimeId);
const secondAnimeId = Math.max(firstCandidateAnimeId, secondCandidateAnimeId);
return Boolean(
db
.prepare(
`SELECT 1
FROM imm_anime_merge_recommendations
WHERE first_anime_id = ?
AND second_anime_id = ?
AND status = 'dismissed'
LIMIT 1`,
)
.get(firstAnimeId, secondAnimeId),
);
}
export function getAnimeMergeRecommendations(db: DatabaseSync): AnimeMergeRecommendation[] {
return (
db
.prepare(
`SELECT recommendation_id AS recommendationId,
first_anime_id AS firstAnimeId,
second_anime_id AS secondAnimeId
FROM imm_anime_merge_recommendations
WHERE status = 'pending'
ORDER BY recommendation_id ASC`,
)
.all() as Array<{
recommendationId: number;
firstAnimeId: number;
secondAnimeId: number;
}>
).map((row) => ({
recommendationId: row.recommendationId,
animeIds: [row.firstAnimeId, row.secondAnimeId],
}));
}
export function dismissAnimeMergeRecommendation(
db: DatabaseSync,
recommendationId: number,
): boolean {
const result = db
.prepare(
`UPDATE imm_anime_merge_recommendations
SET status = 'dismissed', LAST_UPDATE_DATE = ?
WHERE recommendation_id = ? AND status = 'pending'`,
)
.run(toDbTimestamp(nowMs()), recommendationId) as { changes: number };
return result.changes > 0;
}
@@ -23,6 +23,7 @@ export interface VideoMoveSummary {
}
interface AnimeMetadataRow {
normalized_title_key: string;
anilist_id: number | null;
title_romaji: string | null;
title_english: string | null;
@@ -51,7 +52,7 @@ function readAnimeMetadata(db: DatabaseSync, animeId: number): AnimeMetadataRow
return (db
.prepare(
`
SELECT anilist_id, title_romaji, title_english, title_native, episodes_total, description
SELECT normalized_title_key, anilist_id, title_romaji, title_english, title_native, episodes_total, description
FROM imm_anime
WHERE anime_id = ?
`,
@@ -177,6 +178,19 @@ export function mergeAnimeRecordsInTransaction(
'UPDATE imm_subtitle_lines SET anime_id = ?, LAST_UPDATE_DATE = ? WHERE video_id = ?',
);
const dropLifetimeStmt = db.prepare('DELETE FROM imm_lifetime_anime WHERE anime_id = ?');
const sourceAliasesStmt = db.prepare(
'SELECT normalized_title_key AS normalizedTitleKey FROM imm_anime_title_aliases WHERE anime_id = ?',
);
const upsertAliasStmt = db.prepare(
`INSERT INTO imm_anime_title_aliases(normalized_title_key, anime_id, CREATED_DATE, LAST_UPDATE_DATE)
VALUES (?, ?, ?, ?)
ON CONFLICT(normalized_title_key) DO UPDATE SET
anime_id = excluded.anime_id,
LAST_UPDATE_DATE = excluded.LAST_UPDATE_DATE`,
);
const dropSourceAliasesStmt = db.prepare(
'DELETE FROM imm_anime_title_aliases WHERE anime_id = ?',
);
const dropAnimeStmt = db.prepare('DELETE FROM imm_anime WHERE anime_id = ?');
for (const sourceAnimeId of new Set(sourceAnimeIds)) {
@@ -185,6 +199,9 @@ export function mergeAnimeRecordsInTransaction(
}
const sourceMetadata = readAnimeMetadata(db, sourceAnimeId);
const sourceAliases = sourceAliasesStmt.all(sourceAnimeId) as Array<{
normalizedTitleKey: string;
}>;
const sourceVideoIds = (sourceVideosStmt.all(sourceAnimeId) as Array<{ videoId: number }>).map(
(row) => row.videoId,
);
@@ -194,6 +211,13 @@ export function mergeAnimeRecordsInTransaction(
for (const videoId of sourceVideoIds) {
moveLinesStmt.run(targetAnimeId, updatedAt, videoId);
}
dropSourceAliasesStmt.run(sourceAnimeId);
for (const alias of [
...(sourceMetadata ? [sourceMetadata.normalized_title_key] : []),
...sourceAliases.map((row) => row.normalizedTitleKey),
]) {
upsertAliasStmt.run(alias, targetAnimeId, updatedAt, updatedAt);
}
dropLifetimeStmt.run(sourceAnimeId);
dropAnimeStmt.run(sourceAnimeId);
absorbAnimeMetadata(db, targetAnimeId, sourceMetadata, updatedAt);
@@ -4,6 +4,13 @@ import {
getParsedSeasonsForAnime,
mergeAnimeRecordsInTransaction,
} from './anime-merge';
import {
hasExactStoredTitleMatch,
hasDismissedAnimeMergeRecommendation,
recordAnimeMergeRecommendation,
shouldRecommendAnilistConflict,
type AnimeConflictRecommendationOptions,
} from './anime-merge-recommendations';
import { getOrCreateAnimeRecord } from './storage';
import { toDbTimestamp } from './query-shared';
import { nowMs } from './time';
@@ -18,9 +25,13 @@ export interface AnimeSeasonRepairSummary {
* so callers can keep pointing at a row that still exists.
*/
survivingAnimeId: number | null;
/** True when an ambiguous AniList collision was saved for user review. */
mergeRecommended: boolean;
/** True when automatic metadata must not assign the colliding AniList id. */
anilistAssignmentBlocked: boolean;
}
export interface AnimeAnilistConflictOptions {
export interface AnimeAnilistConflictOptions extends AnimeConflictRecommendationOptions {
/**
* Which row keeps its identity when two entries claim the same AniList id.
* `existing` (the default) keeps the row that already held the id, so
@@ -30,6 +41,12 @@ export interface AnimeAnilistConflictOptions {
survivor?: 'target' | 'existing';
}
export {
dismissAnimeMergeRecommendation,
getAnimeMergeRecommendations,
type AnimeMergeRecommendation,
} from './anime-merge-recommendations';
interface AnimeRow {
anime_id: number;
anilist_id: number | null;
@@ -59,6 +76,8 @@ function emptySummary(scanned = 0): AnimeSeasonRepairSummary {
movedVideos: 0,
deletedAnimeRows: 0,
survivingAnimeId: null,
mergeRecommended: false,
anilistAssignmentBlocked: false,
};
}
@@ -71,6 +90,8 @@ function mergeSummary(
target.movedVideos += source.movedVideos;
target.deletedAnimeRows += source.deletedAnimeRows;
target.survivingAnimeId = source.survivingAnimeId ?? target.survivingAnimeId;
target.mergeRecommended ||= source.mergeRecommended;
target.anilistAssignmentBlocked ||= source.anilistAssignmentBlocked;
return target;
}
@@ -324,12 +345,11 @@ export function repairLegacySeasonlessAnimeRows(db: DatabaseSync): AnimeSeasonRe
}
/**
* Two library entries cannot both hold the same AniList id (imm_anime.anilist_id
* is UNIQUE), and two entries resolving to the same id are the same show split
* by a title or season-suffix mismatch. Fold them together when their parsed
* seasons agree; fall back to the legacy season redistribution when the
* conflicting row actually spans several seasons, since merging there would
* pile unrelated seasons onto one card.
* Two library entries cannot both hold the same AniList id
* (`imm_anime.anilist_id` is UNIQUE). Fold an automatic collision only when
* exact title evidence and compatible parsed seasons make it safe. Persist a
* review recommendation for compatible weak matches. Fall back to legacy
* season redistribution when the conflicting row spans several seasons.
*/
export function resolveAnimeAnilistConflict(
db: DatabaseSync,
@@ -353,6 +373,33 @@ export function resolveAnimeAnilistConflict(
}
return runInTransaction(db, () => {
const targetRow = getAnimeRow(db, targetAnimeId);
if (
options.survivor !== 'target' &&
targetRow?.anilist_id != null &&
targetRow.anilist_id !== anilistId
) {
// An automatic lookup disagreeing with an existing explicit link is a
// mis-resolution, not evidence that either row should move or merge.
return emptySummary(1);
}
const isManual = options.survivor === 'target' || options.matchConfidence === 'manual';
if (!isManual && hasDismissedAnimeMergeRecommendation(db, targetAnimeId, conflict.animeId)) {
const summary = emptySummary(1);
summary.anilistAssignmentBlocked = true;
return summary;
}
const targetSeasons = getParsedSeasonsForAnime(db, targetAnimeId);
const conflictSeasons = getParsedSeasonsForAnime(db, conflict.animeId);
if (
targetSeasons.size === 1 &&
conflictSeasons.size === 1 &&
[...targetSeasons][0] !== [...conflictSeasons][0]
) {
const summary = emptySummary(1);
summary.anilistAssignmentBlocked = true;
return summary;
}
if (canMergeAnilistConflict(db, targetAnimeId, conflict.animeId, anilistId, options)) {
const survivingAnimeId = options.survivor === 'target' ? targetAnimeId : conflict.animeId;
const absorbedAnimeId = survivingAnimeId === targetAnimeId ? conflict.animeId : targetAnimeId;
@@ -371,6 +418,13 @@ export function resolveAnimeAnilistConflict(
return summary;
}
if (shouldRecommendAnilistConflict(db, targetAnimeId, conflict.animeId, options)) {
recordAnimeMergeRecommendation(db, targetAnimeId, conflict.animeId, anilistId);
const summary = emptySummary(1);
summary.mergeRecommended = true;
return summary;
}
return redistributeAnimeRowByParsedSeasonsInTransaction(db, conflict.animeId, {
transferAnilistToAnimeId: targetAnimeId,
overwriteTargetAnilist: true,
@@ -398,6 +452,13 @@ function canMergeAnilistConflict(
return false;
}
}
if (
options.matchConfidence === 'weak' ||
(options.matchConfidence === undefined &&
!hasExactStoredTitleMatch(db, targetAnimeId, conflictAnimeId))
) {
return false;
}
return animeSeasonsAreMergeCompatible(
getParsedSeasonsForAnime(db, targetAnimeId),
getParsedSeasonsForAnime(db, conflictAnimeId),
@@ -418,6 +418,7 @@ export function updateAnimeAnilistInfo(
titleEnglish: string | null;
titleNative: string | null;
episodesTotal: number | null;
exactTitleMatch?: boolean;
},
): void {
const row = db.prepare('SELECT anime_id FROM imm_videos WHERE video_id = ?').get(videoId) as {
@@ -425,7 +426,10 @@ export function updateAnimeAnilistInfo(
} | null;
if (!row?.anime_id) return;
const repair = resolveAnimeAnilistConflict(db, row.anime_id, info.anilistId);
const repair = resolveAnimeAnilistConflict(db, row.anime_id, info.anilistId, {
matchConfidence: info.exactTitleMatch === false ? 'weak' : 'exact',
});
if (repair.mergeRecommended || repair.anilistAssignmentBlocked) return;
const targetRow = db
.prepare('SELECT anime_id FROM imm_videos WHERE video_id = ?')
.get(videoId) as {
+40 -2
View File
@@ -530,6 +530,36 @@ function ensureStatsExcludedWordsTable(db: DatabaseSync): void {
`);
}
function ensureAnimeMergeTables(db: DatabaseSync): void {
db.exec(`
CREATE TABLE IF NOT EXISTS imm_anime_title_aliases(
normalized_title_key TEXT PRIMARY KEY,
anime_id INTEGER NOT NULL,
CREATED_DATE TEXT,
LAST_UPDATE_DATE TEXT,
FOREIGN KEY(anime_id) REFERENCES imm_anime(anime_id) ON DELETE CASCADE
);
CREATE INDEX IF NOT EXISTS idx_anime_title_aliases_anime_id
ON imm_anime_title_aliases(anime_id);
CREATE TABLE IF NOT EXISTS imm_anime_merge_recommendations(
recommendation_id INTEGER PRIMARY KEY AUTOINCREMENT,
first_anime_id INTEGER NOT NULL,
second_anime_id INTEGER NOT NULL,
anilist_id INTEGER NOT NULL,
status TEXT NOT NULL DEFAULT 'pending' CHECK(status IN ('pending', 'dismissed')),
CREATED_DATE TEXT,
LAST_UPDATE_DATE TEXT,
CHECK(first_anime_id < second_anime_id),
UNIQUE(first_anime_id, second_anime_id, anilist_id),
FOREIGN KEY(first_anime_id) REFERENCES imm_anime(anime_id) ON DELETE CASCADE,
FOREIGN KEY(second_anime_id) REFERENCES imm_anime(anime_id) ON DELETE CASCADE
);
CREATE INDEX IF NOT EXISTS idx_anime_merge_recommendations_status
ON imm_anime_merge_recommendations(status, recommendation_id);
`);
}
export function getOrCreateAnimeRecord(db: DatabaseSync, input: AnimeRecordInput): number {
const seasonScope = normalizeSeasonScope(input.seasonScope);
const identityTitle = buildSeasonScopedAnimeTitle(input.parsedTitle, seasonScope);
@@ -550,8 +580,14 @@ export function getOrCreateAnimeRecord(db: DatabaseSync, input: AnimeRecordInput
const byNormalizedTitle = db
.prepare('SELECT anime_id FROM imm_anime WHERE normalized_title_key = ?')
.get(normalizedTitleKey) as { anime_id: number } | null;
const existing = byAnilistId ?? byNormalizedTitle;
const byTitleAlias = db
.prepare('SELECT anime_id FROM imm_anime_title_aliases WHERE normalized_title_key = ?')
.get(normalizedTitleKey) as { anime_id: number } | null;
const existing = byAnilistId ?? byNormalizedTitle ?? byTitleAlias;
if (existing?.anime_id) {
// An alias remembers an intentionally merged-away spelling. Reusing it
// must not rename the survivor back to that discarded display title.
const canonicalTitleUpdate = byAnilistId || byNormalizedTitle ? canonicalTitle : null;
db.prepare(
`
UPDATE imm_anime
@@ -566,7 +602,7 @@ export function getOrCreateAnimeRecord(db: DatabaseSync, input: AnimeRecordInput
WHERE anime_id = ?
`,
).run(
canonicalTitle,
canonicalTitleUpdate,
input.anilistId,
input.titleRomaji,
input.titleEnglish,
@@ -751,6 +787,7 @@ export function ensureSchema(db: DatabaseSync): void {
if (currentVersion?.schema_version === SCHEMA_VERSION) {
ensureLifetimeSummaryTables(db);
ensureStatsExcludedWordsTable(db);
ensureAnimeMergeTables(db);
return;
}
@@ -799,6 +836,7 @@ export function ensureSchema(db: DatabaseSync): void {
FOREIGN KEY(anime_id) REFERENCES imm_anime(anime_id) ON DELETE SET NULL
);
`);
ensureAnimeMergeTables(db);
db.exec(`
CREATE TABLE IF NOT EXISTS imm_sessions(
session_id INTEGER PRIMARY KEY AUTOINCREMENT,
+1 -1
View File
@@ -1,4 +1,4 @@
export const SCHEMA_VERSION = 19;
export const SCHEMA_VERSION = 20;
export const DEFAULT_QUEUE_CAP = 1_000;
export const DEFAULT_BATCH_SIZE = 25;
export const DEFAULT_FLUSH_INTERVAL_MS = 500;
@@ -132,6 +132,19 @@ export function registerStatsLibraryRoutes(
return c.json(statsJson('animeLibrary', rows));
});
app.get('/api/stats/anime/merge-recommendations', async (c) => {
const recommendations = await tracker.getAnimeMergeRecommendations();
return c.json(statsJson('animeMergeRecommendations', { recommendations }));
});
app.delete('/api/stats/anime/merge-recommendations/:recommendationId', async (c) => {
const recommendationId = parseIntQuery(c.req.param('recommendationId'), 0);
if (recommendationId <= 0) return c.body(null, 400);
const dismissed = await tracker.dismissAnimeMergeRecommendation(recommendationId);
if (!dismissed) return c.body(null, 404);
return c.json(statsJson('dismissAnimeMergeRecommendation', { ok: true }));
});
app.get('/api/stats/anime/:animeId', async (c) => {
const animeId = parseIntQuery(c.req.param('animeId'), 0);
if (animeId <= 0) return c.body(null, 400);