import config from 'config'; import util from 'util'; /* eslint-disable-line no-unused-vars */ import { knexOwner as knex, knexManticore } from './knex.js'; import { searchApi, utilsApi } from './manticore.js'; import { HttpError } from './errors.js'; import { fetchActorsById, curateActor, sortActorsByGender } from './actors.js'; import { fetchTagsById } from './tags.js'; import { fetchEntitiesById } from './entities.js'; import { curateStash } from './stashes.js'; function curateMedia(media) { if (!media) { return null; } return { id: media.id, path: media.path, thumbnail: media.thumbnail, lazy: media.lazy, isS3: media.is_s3, width: media.width, height: media.height, }; } function curateScene(rawScene, assets) { if (!rawScene) { return null; } return { id: rawScene.id, title: rawScene.title, slug: rawScene.slug, url: rawScene.url, date: rawScene.date, createdAt: rawScene.created_at, effectiveDate: rawScene.effective_date, description: rawScene.description, duration: rawScene.duration, channel: { id: assets.channel.id, slug: assets.channel.slug, name: assets.channel.name, type: assets.channel.type, isIndependent: assets.channel.independent, hasLogo: assets.channel.has_logo, }, network: assets.channel.network_id ? { id: assets.channel.network_id, slug: assets.channel.network_slug, name: assets.channel.network_name, type: assets.channel.network_type, hasLogo: assets.channel.has_logo, } : null, actors: sortActorsByGender(assets.actors.map((actor) => curateActor(actor, { sceneDate: rawScene.effective_date }))), directors: assets.directors.map((director) => ({ id: director.id, slug: director.slug, name: director.name, })), tags: assets.tags.map((tag) => ({ id: tag.id, slug: tag.slug, name: tag.name, })), poster: curateMedia(assets.poster), photos: assets.photos.map((photo) => curateMedia(photo)), stashes: assets.stashes?.map((stash) => curateStash(stash)) || [], createdBatchId: rawScene.created_batch_id, updatedBatchId: rawScene.updated_batch_id, }; } export async function fetchScenesById(sceneIds, reqUser) { const [scenes, channels, actors, directors, tags, posters, photos, stashes] = await Promise.all([ knex('releases').whereIn('releases.id', sceneIds), knex('releases') .select('channels.*', 'networks.id as network_id', 'networks.slug as network_slug', 'networks.name as network_name', 'networks.type as network_type') .whereIn('releases.id', sceneIds) .leftJoin('entities as channels', 'channels.id', 'releases.entity_id') .leftJoin('entities as networks', 'networks.id', 'channels.parent_id') .groupBy('channels.id', 'networks.id'), knex('releases_actors') .select( 'actors.*', 'actors_meta.*', 'releases_actors.release_id', /* why would we need this for scenes? 'birth_countries.alpha2 as birth_country_alpha2', knex.raw('COALESCE(birth_countries.alias, birth_countries.name) as birth_country_name'), 'residence_countries.alpha2 as residence_country_alpha2', knex.raw('COALESCE(residence_countries.alias, residence_countries.name) as residence_country_name'), */ ) .leftJoin('actors', 'actors.id', 'releases_actors.actor_id') .leftJoin('actors_meta', 'actors_meta.actor_id', 'actors.id') .whereIn('release_id', sceneIds), /* .leftJoin('countries as birth_countries', 'birth_countries.alpha2', 'actors_meta.birth_country_alpha2') .leftJoin('countries as residence_countries', 'residence_countries.alpha2', 'actors_meta.residence_country_alpha2'), */ knex('releases_directors') .whereIn('release_id', sceneIds) .leftJoin('actors as directors', 'directors.id', 'releases_directors.director_id'), knex('releases_tags') .select('id', 'slug', 'name', 'release_id') .leftJoin('tags', 'tags.id', 'releases_tags.tag_id') .whereNotNull('tags.id') .whereIn('release_id', sceneIds) .orderBy('priority', 'desc'), knex('releases_posters') .whereIn('release_id', sceneIds) .leftJoin('media', 'media.id', 'releases_posters.media_id'), knex('releases_photos') .whereIn('release_id', sceneIds) .leftJoin('media', 'media.id', 'releases_photos.media_id'), reqUser ? knex('stashes_scenes') .leftJoin('stashes', 'stashes.id', 'stashes_scenes.stash_id') .where('stashes.user_id', reqUser.id) .whereIn('stashes_scenes.scene_id', sceneIds) : [], ]); return sceneIds.map((sceneId) => { const scene = scenes.find((sceneEntry) => sceneEntry.id === sceneId); if (!scene) { return null; } const sceneChannel = channels.find((entity) => entity.id === scene.entity_id); const sceneActors = actors.filter((actor) => actor.release_id === sceneId); const sceneDirectors = directors.filter((director) => director.release_id === sceneId); const sceneTags = tags.filter((tag) => tag.release_id === sceneId); const scenePoster = posters.find((poster) => poster.release_id === sceneId); const scenePhotos = photos.filter((photo) => photo.release_id === sceneId); const sceneStashes = stashes.filter((stash) => stash.scene_id === sceneId); return curateScene(scene, { channel: sceneChannel, actors: sceneActors, directors: sceneDirectors, tags: sceneTags, poster: scenePoster, photos: scenePhotos, stashes: sceneStashes, }); }).filter(Boolean); } const sqlImplied = ['scenes_stashed']; function curateOptions(options) { if (options?.limit > 100) { throw new HttpError('Limit must be <= 100', 400); } return { limit: options?.limit || 30, page: Number(options?.page) || 1, aggregate: options.aggregate ?? true, aggregateActors: (options.aggregate ?? true) && (options.aggregateActors ?? true), aggregateTags: (options.aggregate ?? true) && (options.aggregateTags ?? true), aggregateChannels: (options.aggregate ?? true) && (options.aggregateChannels ?? true), index: options.index || 'scenes', useSql: options.useSql || (typeof options.useSql === 'undefined' && sqlImplied.includes(options.index)) || false, }; } function buildQuery(filters = {}, options) { const query = { bool: { must: [], }, }; let sort = [{ effective_date: 'desc' }]; if (!filters.scope || filters.scope === 'latest') { query.bool.must.push({ range: { effective_date: { lte: Math.round(Date.now() / 1000), }, }, }); } if (filters.scope === 'upcoming') { query.bool.must.push({ range: { effective_date: { gt: Math.round(Date.now() / 1000), }, }, }); sort = [{ effective_date: 'asc' }]; } if (filters.scope === 'new') { sort = [{ created_at: 'desc' }, { effective_date: 'asc' }]; } if (filters.scope === 'likes') { sort = [{ stashed: 'desc' }, { effective_date: 'desc' }]; } if (filters.scope === 'results') { sort = [{ _score: 'desc' }, { effective_date: 'desc' }]; } if (filters.query) { /* query.bool.must.push({ bool: { should: [ { match: { title_filtered: filters.query } }, { match: { actors: filters.query } }, { match: { tags: filters.query } }, { match: { channel_name: filters.query } }, { match: { network_name: filters.query } }, { match: { channel_slug: filters.query } }, { match: { network_slug: filters.query } }, { match: { meta: filters.query } }, // date ], }, }); */ query.bool.must.push({ match: { '!title': filters.query } }); // title_filtered is matched instead of title } if (filters.tagIds) { filters.tagIds.forEach((tagId) => { query.bool.must.push({ equals: { 'any(tag_ids)': tagId } }); }); } if (filters.entityId) { query.bool.must.push({ bool: { should: [ { equals: { channel_id: filters.entityId } }, { equals: { network_id: filters.entityId } }, ], }, }); } if (filters.actorIds) { filters.actorIds.forEach((actorId) => { query.bool.must.push({ equals: { 'any(actor_ids)': actorId } }); }); } if (filters.stashId && options.index === 'scenes_stashed') { query.bool.must.push({ equals: { stash_id: filters.stashId } }); } /* tag filter must_not: [ { in: { 'any(tag_ids)': [101, 180, 32], }, }, ], */ return { query, sort }; } function buildAggregates(options) { const aggregates = {}; if (options.aggregateActors) { aggregates.actorIds = { terms: { field: 'actor_ids', size: config.database.manticore.maxAggregateSize, }, sort: [{ 'count(*)': { order: 'desc' } }], }; } if (options.aggregateTags) { aggregates.tagIds = { terms: { field: 'tag_ids', size: config.database.manticore.maxAggregateSize, }, sort: [{ 'count(*)': { order: 'desc' } }], }; } if (options.aggregateChannels) { aggregates.channelIds = { terms: { field: 'channel_id', size: config.database.manticore.maxAggregateSize, }, sort: [{ 'count(*)': { order: 'desc' } }], }; } return aggregates; } function countAggregations(buckets) { if (!buckets) { return null; } return Object.fromEntries(buckets.map((bucket) => [bucket.key, { count: bucket.doc_count }])); } async function queryManticoreJson(filters, options, _reqUser) { const { query, sort } = buildQuery(filters, options); const result = await searchApi.search({ index: options.index, query, limit: options.limit, offset: (options.page - 1) * options.limit, sort, aggs: buildAggregates(options), options: { max_matches: config.database.manticore.maxMatches, max_query_time: config.database.manticore.maxQueryTime, field_weights: { title_filtered: 7, actors: 10, tags: 9, meta: 6, channel_name: 2, channel_slug: 3, network_name: 1, network_slug: 1, }, }, }); const scenes = result.hits.hits.map((hit) => ({ id: hit._id, ...hit._source, _score: hit._score, })); return { scenes, total: result.hits.total, aggregations: result.aggregations && Object.fromEntries(Object.entries(result.aggregations).map(([key, { buckets }]) => [key, buckets])), }; } async function queryManticoreSql(filters, options, _reqUser) { const aggSize = 10 || config.database.manticore.maxAggregateSize; const sqlQuery = knexManticore.raw(` :query: OPTION field_weights=( title_filtered=7, actors=10, tags=9, meta=6, channel_name=2, channel_slug=3, network_name=1, network_slug=1 ), max_matches=:maxMatches:, max_query_time=:maxQueryTime: :actorsFacet: :tagsFacet: :channelsFacet: `, { query: knexManticore('scenes') .select(knex.raw('*, weight() as _score')) .modify((builder) => { if (filters.stashId) { builder .innerJoin('scenes_stashed', 'scenes.id', 'scenes_stashed.scene_id') .where('scenes_stashed.stash_id', filters.stashId); } if (filters.query) { builder.whereRaw('match(\'@!title :query:\', scenes)', { query: filters.query }); } if (filters.tagIds?.length > 0) { builder.whereIn('any(tag_ids)', filters.tagIds); } if (filters.entityId) { builder.where((whereBuilder) => { whereBuilder .where('channel_id', filters.entityId) .orWhere('network_id', filters.entityId); }); } if (filters.actorIds?.length > 0) { builder.whereIn('any(actor_ids)', filters.actorIds); } if (!filters.scope || filters.scope === 'latest') { builder .where('effective_date', '<=', Math.round(Date.now() / 1000)) .orderBy('effective_date', 'desc'); } else if (filters.scope === 'upcoming') { builder .where('effective_date', '>', Math.round(Date.now() / 1000)) .orderBy('effective_date', 'asc'); } else if (filters.scope === 'new') { builder.orderBy([ { column: 'created_at', order: 'desc' }, { column: 'effective_date', order: 'asc' }, ]); } else if (filters.scope === 'likes') { builder.orderBy([ { column: 'stashed', order: 'desc' }, { column: 'effective_date', order: 'desc' }, ]); } else if (filters.scope === 'results') { builder.orderBy([ { column: '_score', order: 'desc' }, { column: 'effective_date', order: 'desc' }, ]); } else { builder.orderBy('effective_date', 'desc'); } }) .limit(options.limit) .toString(), // option threads=1 fixes actors, but drastically slows down performance, wait for fix actorsFacet: options.aggregateActors ? knex.raw('facet actor_ids order by count(*) desc limit ?', [aggSize]) : null, tagsFacet: options.aggregateTags ? knex.raw('facet tag_ids order by count(*) desc limit ?', [aggSize]) : null, channelsFacet: options.aggregateChannels ? knex.raw('facet channel_id order by count(*) desc limit ?', [aggSize]) : null, maxMatches: config.database.manticore.maxMatches, maxQueryTime: config.database.manticore.maxQueryTime, }).toString(); console.log(sqlQuery); const results = await utilsApi.sql(sqlQuery); const actorIds = results .find((result) => result.columns[0].actor_ids && result.columns[1]['count(*)']) ?.data.map((row) => ({ key: row.actor_ids, doc_count: row['count(*)'] })) || []; const tagIds = results .find((result) => result.columns[0].tag_ids && result.columns[1]['count(*)']) ?.data.map((row) => ({ key: row.tag_ids, doc_count: row['count(*)'] })) || []; const channelIds = results .find((result) => result.columns[0].channel_id && result.columns[1]['count(*)']) ?.data.map((row) => ({ key: row.channel_id, doc_count: row['count(*)'] })) || []; return { scenes: results[0].data, total: results[0].total, aggregations: { actorIds, tagIds, channelIds, }, }; } export async function fetchScenes(filters, rawOptions, reqUser) { const options = curateOptions(rawOptions); console.log('filters', filters); console.log('options', options); /* const result = config.database.manticore.forceSql || filters.stashId ? await queryManticoreSql(filters, options, reqUser) : await queryManticoreJson(filters, options, reqUser); */ console.time('manticore sql'); const result = await queryManticoreSql(filters, options, reqUser); console.timeEnd('manticore sql'); console.time('manticore json'); await queryManticoreJson(filters, options, reqUser); console.timeEnd('manticore json'); const actorCounts = options.aggregateActors && countAggregations(result.aggregations?.actorIds); const tagCounts = options.aggregateTags && countAggregations(result.aggregations?.tagIds); const channelCounts = options.aggregateChannels && countAggregations(result.aggregations?.channelIds); console.time('fetch aggregations'); const [aggActors, aggTags, aggChannels] = await Promise.all([ options.aggregateActors ? fetchActorsById(result.aggregations.actorIds.map((bucket) => bucket.key), { order: ['name', 'asc'], append: actorCounts }) : [], options.aggregateTags ? fetchTagsById(result.aggregations.tagIds.map((bucket) => bucket.key), { order: ['name', 'asc'], append: tagCounts }) : [], options.aggregateChannels ? fetchEntitiesById(result.aggregations.channelIds.map((bucket) => bucket.key), { order: ['name', 'asc'], append: channelCounts }) : [], ]); console.timeEnd('fetch aggregations'); console.time('fetch full'); const sceneIds = result.scenes.map((scene) => Number(scene.id)); const scenes = await fetchScenesById(sceneIds, reqUser); console.timeEnd('fetch full'); return { scenes, aggActors, aggTags, aggChannels, total: result.total, limit: options.limit, }; }