Added tag reassociation and dedupe migration.

migrations/20260222055254_unique_origin_tags.js

@@ -0,0 +1,31 @@
+exports.up = async (knex) => {
+	// dedupe
+	await knex.raw(`
+		DELETE
+		FROM releases_tags
+		WHERE ctid IN
+		(
+			SELECT ctid
+			FROM(
+				SELECT
+					*,
+					ctid,
+					row_number() OVER (PARTITION BY release_id, original_tag ORDER BY ctid)
+				FROM releases_tags
+			)s
+			WHERE row_number >= 2
+		)
+	`);
+
+	await knex.schema.alterTable('releases_tags', (table) => {
+		table.increments('id');
+		table.unique(['release_id', 'original_tag']);
+	});
+};
+
+exports.down = async (knex) => {
+	await knex.schema.alterTable('releases_tags', (table) => {
+		table.dropColumn('id');
+		table.dropUnique(['release_id', 'original_tag']);
+	});
+};

seeds/00_tags.js

@@ -838,6 +838,11 @@ const tags = [
 		slug: 'natural-boobs',
 		group: 'body',
 	},
+	{
+		name: 'natural butt',
+		slug: 'natural-butt',
+		group: 'body',
+	},
 	{
 		name: 'nipple clamps',
 		slug: 'nipple-clamps',
@@ -1616,6 +1621,10 @@ const aliases = [
 		name: 'big tits d-dd cup',
 		for: 'big-boobs',
 	},
+	{
+		name: 'busty',
+		for: 'big-boobs',
+	},
 	{
 		name: 'busty - big boobs',
 		for: 'big-boobs',
@@ -2159,6 +2168,18 @@ const aliases = [
 		name: 'natural tits',
 		for: 'natural-boobs',
 	},
+	{
+		name: 'natural butt',
+		for: 'natural-butt',
+	},
+	{
+		name: 'natural ass',
+		for: 'natural-butt',
+	},
+	{
+		name: 'real ass',
+		for: 'natural-butt',
+	},
 	{
 		name: 'oiled',
 		for: 'oil',

src/app.js

@@ -25,6 +25,7 @@ const { scrapeActors, deleteActors, flushActors, flushProfiles, interpolateProfi
 const { flushEntities } = require('./entities');
 const { deleteScenes, deleteMovies, flushScenes, flushMovies, flushBatches } = require('./releases');
 const { flushOrphanedMedia } = require('./media');
+const { reassociateEntityReleaseTags, reassociateReleaseTags, reassociateOriginalTags } = require('./tags');
 const getFileEntries = require('./utils/file-entries');
 
 const inspector = new Inspector();
@@ -186,6 +187,18 @@ async function init() {
 			await deleteMovies(argv.deleteMovies);
 		}
 
+		if (argv.originalTags) {
+			await reassociateOriginalTags(argv.originalTags, argv.rematchTags);
+		}
+
+		if (argv.releaseTags) {
+			await reassociateReleaseTags(argv.releaseTags, argv.rematchTags);
+		}
+
+		if (argv.networkReleaseTags || argv.channelReleaseTags) {
+			await reassociateEntityReleaseTags(argv.networkReleaseTags, argv.channelReleaseTags, argv.rematchTags);
+		}
+
 		if (argv.flushOrphanedMedia) {
 			await flushOrphanedMedia();
 		}

src/argv.js

@@ -407,6 +407,32 @@ const { argv } = yargs
 		type: 'array',
 		alias: ['delete-movie', 'remove-movies', 'remove-movies'],
 	})
+	.option('original-tags', {
+		describe: 'Reassociate original tag names',
+		type: 'array',
+		alias: ['tags'],
+	})
+	.option('release-tags', {
+		describe: 'Reassociate tags for scene IDs',
+		type: 'array',
+		alias: ['scene-tags'],
+	})
+	.option('channel-release-tags', {
+		describe: 'Reassociate tags for all channel releases',
+		type: 'array',
+		alias: ['channel-scene-tags', 'channel-tags'],
+	})
+	.option('network-release-tags', {
+		describe: 'Reassociate tags for all network releases',
+		type: 'array',
+		alias: ['network-scene-tags', 'network-tags'],
+	})
+	.option('rematch-tags', {
+		describe: 'Reassociate tags that are already associated',
+		type: 'boolean',
+		alias: 'rematch',
+		default: false,
+	})
 	.option('request', {
 		describe: 'Make an arbitrary HTTP request',
 		type: 'string',

src/entities.js

@@ -368,9 +368,7 @@ async function searchEntities(query, type, limit) {
 	return curateEntities(entities);
 }
 
-async function flushEntities(networkSlugs = [], channelSlugs = []) {
+async function fetchEntityReleaseIds(networkSlugs = [], channelSlugs = []) {
-	const entitySlugs = networkSlugs.concat(channelSlugs).join(', ');
-
 	const entityQuery = knex
 		.withRecursive('selected_entities', knex.raw(`
 			SELECT entities.*
@@ -443,6 +441,17 @@ async function flushEntities(networkSlugs = [], channelSlugs = []) {
 		})
 		.pluck('series.id');
 
+	return {
+		sceneIds,
+		movieIds,
+		serieIds,
+	};
+}
+
+async function flushEntities(networkSlugs = [], channelSlugs = []) {
+	const { sceneIds, movieIds, serieIds } = await fetchEntityReleaseIds(networkSlugs, channelSlugs);
+	const entitySlugs = networkSlugs.concat(channelSlugs).join(', ');
+
 	if (sceneIds.length === 0 && movieIds.length === 0 && serieIds.length === 0) {
 		logger.info(`No scenes, movies or series found to remove for ${entitySlugs}`);
 		return;
@@ -479,6 +488,7 @@ module.exports = {
 	fetchIncludedEntities,
 	fetchReleaseEntities,
 	fetchEntitiesBySlug,
+	fetchEntityReleaseIds,
 	fetchEntity,
 	fetchEntities,
 	getRecursiveParent,

src/tags.js

@@ -1,6 +1,10 @@
 'use strict';
 
+const logger = require('./logger')(__filename);
 const knex = require('./knex');
+
+const { fetchEntityReleaseIds } = require('./entities');
+
 const slugify = require('./utils/slugify');
 const bulkInsert = require('./utils/bulk-insert');
 
@@ -73,9 +77,8 @@ function withRelations(queryBuilder, withMedia) {
 	}
 }
 
-async function matchReleaseTags(releases) {
+async function matchTags(rawTags) {
-	const tags = releases
+	const tags = rawTags
-		.map((release) => release.tags).flat()
 		.map((tag) => tag?.trim().match(/[a-z0-9()]+/ig)?.join(' ').toLowerCase())
 		.filter(Boolean);
 
@@ -153,7 +156,7 @@ async function associateReleaseTags(releases, type = 'release') {
 		return;
 	}
 
-	const tagIdsBySlug = await matchReleaseTags(releases);
+	const tagIdsBySlug = await matchTags(releases.flatMap((release) => release.tags));
 	const entityTagIdsByEntityId = await getEntityTags(releases);
 
 	const tagAssociations = buildReleaseTagAssociations(releases, tagIdsBySlug, entityTagIdsByEntityId, type);
@@ -187,8 +190,53 @@ async function fetchTags(limit = 100) {
 	return tags.map((tag) => curateTag(tag));
 }
 
+async function reassociateTagEntries(tagEntries, rematch) {
+	const updatableTagEntries = tagEntries.filter((tagEntry) => (!tagEntry.tag_id || rematch) && tagEntry.original_tag);
+	const matchedTags = await matchTags(Array.from(new Set(updatableTagEntries.map((tagEntry) => tagEntry.original_tag))));
+
+	const updatedTagEntries = updatableTagEntries.map((tagEntry) => ({
+		...tagEntry,
+		tag_id: matchedTags[slugify(tagEntry.original_tag)],
+	})).filter((tagEntry) => tagEntry.tag_id);
+
+	if (updatedTagEntries.length > 0) {
+		await knex('releases_tags')
+			.whereIn('id', updatedTagEntries.map((tagEntry) => tagEntry.id))
+			.delete();
+
+		await knex('releases_tags').insert(updatedTagEntries.map((tagEntry) => ({
+			...tagEntry,
+			id: undefined,
+		})));
+	}
+
+	logger.info(`Updated ${updatedTagEntries.length} tags in ${new Set(updatedTagEntries.map((tagEntry) => tagEntry.release_id)).size} scenes`);
+}
+
+async function reassociateReleaseTags(rawSceneIds, rematch) {
+	const sceneIds = rawSceneIds.map((sceneId) => Number(sceneId)).filter(Boolean);
+	const tagEntries = await knex('releases_tags').whereIn('release_id', sceneIds);
+
+	await reassociateTagEntries(tagEntries, rematch);
+}
+
+async function reassociateEntityReleaseTags(networkSlugs = [], channelSlugs = [], rematch = false) {
+	const { sceneIds } = await fetchEntityReleaseIds(networkSlugs, channelSlugs);
+
+	await reassociateReleaseTags(sceneIds, rematch);
+}
+
+async function reassociateOriginalTags(originalTags, rematch) {
+	const tagEntries = await knex('releases_tags').whereIn(knex.raw('lower(releases_tags.original_tag)'), originalTags.map((originalTag) => originalTag.toLowerCase()));
+
+	await reassociateTagEntries(tagEntries, rematch);
+}
+
 module.exports = {
 	associateReleaseTags,
 	fetchTag,
 	fetchTags,
+	reassociateEntityReleaseTags,
+	reassociateReleaseTags,
+	reassociateOriginalTags,
 };