traxxx/src/update-search.js

'use strict';

const manticore = require('manticoresearch');
const { format } = require('date-fns');

const knex = require('./knex');
const logger = require('./logger')(__filename);
const bulkInsert = require('./utils/bulk-insert');
const chunk = require('./utils/chunk');

const mantiClient = new manticore.ApiClient();
const indexApi = new manticore.IndexApi(mantiClient);

async function updateManticoreStashedScenes(docs) {
	await chunk(docs, 1000).reduce(async (chain, docsChunk) => {
		await chain;

		const sceneIds = docsChunk.map((doc) => doc.replace.id);

		const stashes = await knex('stashes_scenes')
			.select('stashes_scenes.id as stashed_id', 'stashes_scenes.scene_id', 'stashes_scenes.created_at', 'stashes.id as stash_id', 'stashes.user_id as user_id')
			.leftJoin('stashes', 'stashes.id', 'stashes_scenes.stash_id')
			.whereIn('scene_id', sceneIds);

		const stashDocs = docsChunk.flatMap((doc) => {
			const sceneStashes = stashes.filter((stash) => stash.scene_id === doc.replace.id);

			if (sceneStashes.length === 0) {
				return [];
			}

			const stashDoc = sceneStashes.map((stash) => ({
				replace: {
					index: 'scenes_stashed',
					id: stash.stashed_id,
					doc: {
						// ...doc.replace.doc,
						scene_id: doc.replace.id,
						user_id: stash.user_id,
						stash_id: stash.stash_id,
						created_at: Math.round(stash.created_at.getTime() / 1000),
					},
				},
			}));

			return stashDoc;
		});

		if (stashDocs.length > 0) {
			await indexApi.bulk(stashDocs.map((doc) => JSON.stringify(doc)).join('\n'));
		}
	}, Promise.resolve());
}

async function updateManticoreSceneSearch(releaseIds) {
	logger.info(`Updating Manticore search documents for ${releaseIds ? releaseIds.length : 'all' } releases`);

	const scenes = await knex.raw(`
        SELECT
            releases.id AS id,
			releases.title,
			releases.created_at,
			releases.date,
			releases.shoot_id,
			scenes_meta.stashed,
			entities.id as channel_id,
			entities.slug as channel_slug,
			entities.name as channel_name,
			parents.id as network_id,
			parents.slug as network_slug,
			parents.name as network_name,
			grandparents.id as parent_network_id,
			COALESCE(JSON_AGG(DISTINCT (actors.id, actors.name)) FILTER (WHERE actors.id IS NOT NULL), '[]') as actors,
			COALESCE(JSON_AGG(DISTINCT (tags.id, tags.name, tags.priority, tags_aliases.name)) FILTER (WHERE tags.id IS NOT NULL), '[]') as tags,
			COALESCE(JSON_AGG(DISTINCT (movies_scenes.movie_id)) FILTER (WHERE movies_scenes.movie_id IS NOT NULL), '[]') as movies,
			COALESCE(JSON_AGG(DISTINCT (series_scenes.serie_id)) FILTER (WHERE series_scenes.serie_id IS NOT NULL), '[]') as series,
			studios.showcased IS NOT false
			AND (entities.showcased IS NOT false OR COALESCE(studios.showcased, false) = true)
			AND (parents.showcased IS NOT false OR COALESCE(entities.showcased, false) = true OR COALESCE(studios.showcased, false) = true)
			AND (releases_summaries.batch_showcased IS NOT false)
			AS showcased
        FROM releases
		LEFT JOIN releases_summaries ON releases_summaries.release_id = releases.id
		LEFT JOIN scenes_meta ON scenes_meta.scene_id = releases.id
        LEFT JOIN entities ON releases.entity_id = entities.id
        LEFT JOIN entities AS parents ON parents.id = entities.parent_id
        LEFT JOIN entities AS grandparents ON grandparents.id = parents.parent_id
        LEFT JOIN entities AS studios ON studios.id = releases.studio_id
        LEFT JOIN releases_actors AS local_actors ON local_actors.release_id = releases.id
        LEFT JOIN releases_directors AS local_directors ON local_directors.release_id = releases.id
        LEFT JOIN releases_tags AS local_tags ON local_tags.release_id = releases.id
        LEFT JOIN actors ON local_actors.actor_id = actors.id
        LEFT JOIN actors AS directors ON local_directors.director_id = directors.id
        LEFT JOIN tags ON local_tags.tag_id = tags.id
        LEFT JOIN tags as tags_aliases ON local_tags.tag_id = tags_aliases.alias_for AND tags_aliases.secondary = true
        LEFT JOIN movies_scenes ON movies_scenes.scene_id = releases.id
        LEFT JOIN series_scenes ON series_scenes.scene_id = releases.id
        ${releaseIds ? 'WHERE releases.id = ANY(?)' : ''}
        GROUP BY
            releases.id,
			releases.title,
			releases.created_at,
			releases.date,
			releases.shoot_id,
			scenes_meta.stashed,
			releases_summaries.batch_showcased,
			entities.id,
			entities.name,
			entities.slug,
			entities.alias,
			entities.showcased,
			parents.id,
			parents.name,
			parents.slug,
			parents.alias,
			grandparents.id,
			parents.showcased,
			studios.showcased
    `, releaseIds && [releaseIds]);

	// console.log(scenes.rows);

	const docs = scenes.rows.map((scene) => {
		const flatActors = scene.actors.flatMap((actor) => actor.f2.split(' '));
		const flatTags = scene.tags.filter((tag) => tag.f3 > 6).flatMap((tag) => [tag.f2].concat(tag.f4)).filter(Boolean); // only make top tags searchable to minimize cluttered results
		const filteredTitle = scene.title && [...flatActors, ...flatTags].reduce((accTitle, tag) => accTitle.replace(new RegExp(`\\b${tag.replace(/[^\w\s]+/g, '')}\\b`, 'gi'), ''), scene.title).trim().replace(/\s{2,}/, ' ');

		return {
			replace: {
				index: 'scenes',
				id: scene.id,
				doc: {
					title: scene.title || undefined,
					title_filtered: filteredTitle || undefined,
					date: scene.date ? Math.round(scene.date.getTime() / 1000) : undefined,
					created_at: Math.round(scene.created_at.getTime() / 1000),
					effective_date: Math.round((scene.date || scene.created_at).getTime() / 1000),
					is_showcased: scene.showcased,
					shoot_id: scene.shoot_id || undefined,
					channel_id: scene.channel_id,
					channel_slug: scene.channel_slug,
					channel_name: scene.channel_name,
					network_id: scene.network_id || undefined,
					network_slug: scene.network_slug || undefined,
					network_name: scene.network_name || undefined,
					entity_ids: [scene.channel_id, scene.network_id, scene.parent_network_id].filter(Boolean), // manticore does not support OR, this allows IN
					actor_ids: scene.actors.map((actor) => actor.f1),
					actors: scene.actors.map((actor) => actor.f2).join(),
					tag_ids: scene.tags.map((tag) => tag.f1),
					tags: flatTags.join(' '), // only make top tags searchable to minimize cluttered results
					movie_ids: scene.movies,
					serie_ids: scene.series,
					meta: scene.date ? format(scene.date, 'y yy M MMM MMMM d') : undefined,
					stashed: scene.stashed || 0,
				},
			},
		};
	});

	if (docs.length === 0) {
		return;
	}

	await Promise.all([
		indexApi.bulk(docs.map((doc) => JSON.stringify(doc)).join('\n')),
		updateManticoreStashedScenes(docs),
	]);
}

async function updateSqlSceneSearch(releaseIds) {
	logger.info(`Updating SQL search documents for ${releaseIds ? releaseIds.length : 'all' } releases`);

	const documents = await knex.raw(`
        SELECT
            releases.id AS release_id,
            TO_TSVECTOR(
                'english',
                COALESCE(releases.title, '') || ' ' ||
                releases.entry_id || ' ' ||
                entities.name || ' ' ||
                entities.slug || ' ' ||
                COALESCE(array_to_string(entities.alias, ' '), '') || ' ' ||
                COALESCE(parents.name, '') || ' ' ||
                COALESCE(parents.slug, '') || ' ' ||
                COALESCE(array_to_string(parents.alias, ' '), '') || ' ' ||
                COALESCE(releases.shoot_id, '') || ' ' ||
                COALESCE(TO_CHAR(releases.date, 'YYYY YY MM FMMM FMMonth mon DD FMDD'), '') || ' ' ||
                STRING_AGG(COALESCE(actors.name, ''), ' ') || ' ' ||
                STRING_AGG(COALESCE(directors.name, ''), ' ') || ' ' ||
                STRING_AGG(COALESCE(tags.name, ''), ' ') || ' ' ||
                STRING_AGG(COALESCE(tags_aliases.name, ''), ' ')
            ) as document
        FROM releases
        LEFT JOIN entities ON releases.entity_id = entities.id
        LEFT JOIN entities AS parents ON parents.id = entities.parent_id
        LEFT JOIN releases_actors AS local_actors ON local_actors.release_id = releases.id
        LEFT JOIN releases_directors AS local_directors ON local_directors.release_id = releases.id
        LEFT JOIN releases_tags AS local_tags ON local_tags.release_id = releases.id
        LEFT JOIN actors ON local_actors.actor_id = actors.id
        LEFT JOIN actors AS directors ON local_directors.director_id = directors.id
        LEFT JOIN tags ON local_tags.tag_id = tags.id AND tags.priority >= 6
        LEFT JOIN tags as tags_aliases ON local_tags.tag_id = tags_aliases.alias_for AND tags_aliases.secondary = true
        ${releaseIds ? 'WHERE releases.id = ANY(?)' : ''}
        GROUP BY releases.id, entities.name, entities.slug, entities.alias, parents.name, parents.slug, parents.alias;
    `, releaseIds && [releaseIds]);

	if (documents.rows?.length > 0) {
		await bulkInsert('releases_search', documents.rows, ['release_id']);
	}

	await knex.raw('REFRESH MATERIALIZED VIEW releases_summaries;');
}

async function updateSceneSearch(releaseIds) {
	await knex.raw('REFRESH MATERIALIZED VIEW scenes_meta;');

	await updateSqlSceneSearch(releaseIds);
	await updateManticoreSceneSearch(releaseIds);
}

async function updateManticoreMovieSearch(movieIds) {
	const movies = await knex.raw(`
        SELECT
            movies.id AS id,
			movies.title,
			movies.created_at,
			movies.date,
			movies_meta.stashed,
			entities.id as channel_id,
			entities.slug as channel_slug,
			entities.name as channel_name,
			parents.id as network_id,
			parents.slug as network_slug,
			parents.name as network_name,
			movies_covers IS NOT NULL as has_cover,
			COALESCE(JSON_AGG(DISTINCT (actors.id, actors.name)) FILTER (WHERE actors.id IS NOT NULL), '[]') as actors,
			COALESCE(JSON_AGG(DISTINCT (tags.id, tags.name, tags.priority, tags_aliases.name)) FILTER (WHERE tags.id IS NOT NULL), '[]') as tags
        FROM movies
		LEFT JOIN movies_meta ON movies_meta.movie_id = movies.id
		LEFT JOIN movies_scenes ON movies_scenes.movie_id = movies.id
        LEFT JOIN entities ON movies.entity_id = entities.id
        LEFT JOIN entities AS parents ON parents.id = entities.parent_id
        LEFT JOIN releases_actors AS local_actors ON local_actors.release_id = movies_scenes.scene_id
        LEFT JOIN releases_directors AS local_directors ON local_directors.release_id = movies_scenes.scene_id
        LEFT JOIN releases_tags AS local_tags ON local_tags.release_id = movies_scenes.scene_id
        LEFT JOIN actors ON local_actors.actor_id = actors.id
        LEFT JOIN actors AS directors ON local_directors.director_id = directors.id
        LEFT JOIN tags ON local_tags.tag_id = tags.id
        LEFT JOIN tags as tags_aliases ON local_tags.tag_id = tags_aliases.alias_for AND tags_aliases.secondary = true
        LEFT JOIN movies_covers ON movies_covers.movie_id = movies.id
        ${movieIds ? 'WHERE movies.id = ANY(?)' : ''}
        GROUP BY
            movies.id,
			movies.title,
			movies.created_at,
			movies.date,
			movies_meta.stashed,
			movies_meta.stashed_scenes,
			movies_meta.stashed_total,
			entities.id,
			entities.name,
			entities.slug,
			entities.alias,
			parents.id,
			parents.name,
			parents.slug,
			parents.alias,
			movies_covers.*
    `, movieIds && [movieIds]);

	const docs = movies.rows.map((movie) => {
		const flatActors = movie.actors.flatMap((actor) => actor.f2.match(/[\w']+/g)); // match word characters to filter out brackets etc.
		const flatTags = movie.tags.filter((tag) => tag.f3 > 6).flatMap((tag) => (tag.f4 ? `${tag.f2} ${tag.f4}` : tag.f2).match(/[\w']+/g)); // only make top tags searchable to minimize cluttered results
		const filteredTitle = movie.title && [...flatActors, ...flatTags].reduce((accTitle, tag) => accTitle.replace(new RegExp(tag.replace(/[^\w\s]+/g, ''), 'gi'), ''), movie.title).trim().replace(/\s{2,}/g, ' ');

		return {
			replace: {
				index: 'movies',
				id: movie.id,
				doc: {
					title: movie.title || undefined,
					title_filtered: filteredTitle || undefined,
					date: movie.date ? Math.round(movie.date.getTime() / 1000) : undefined,
					created_at: Math.round(movie.created_at.getTime() / 1000),
					effective_date: Math.round((movie.date || movie.created_at).getTime() / 1000),
					channel_id: movie.channel_id,
					channel_slug: movie.channel_slug,
					channel_name: movie.channel_name,
					network_id: movie.network_id || undefined,
					network_slug: movie.network_slug || undefined,
					network_name: movie.network_name || undefined,
					entity_ids: [movie.channel_id, movie.network_id].filter(Boolean), // manticore does not support OR, this allows IN
					actor_ids: movie.actors.map((actor) => actor.f1),
					actors: movie.actors.map((actor) => actor.f2).join(),
					tag_ids: movie.tags.map((tag) => tag.f1),
					tags: flatTags.join(' '),
					has_cover: movie.has_cover,
					meta: movie.date ? format(movie.date, 'y yy M MMM MMMM d') : undefined,
					stashed: movie.stashed || 0,
					stashed_scenes: movie.stashed_scenes || 0,
					stashed_total: movie.stashed_total || 0,
				},
			},
		};
	});

	if (docs.length === 0) {
		return;
	}

	await indexApi.bulk(docs.map((doc) => JSON.stringify(doc)).join('\n'));
}

async function updateSqlMovieSearch(movieIds, target = 'movie') {
	logger.info(`Updating search documents for ${movieIds ? movieIds.length : 'all' } ${target}s`);

	const documents = await knex.raw(`
		SELECT
			${target}s.id AS ${target}_id,
				TO_TSVECTOR(
				'english',
					COALESCE(${target}s.title, '') || ' ' ||
					entities.name || ' ' ||
					entities.slug || ' ' ||
					COALESCE(array_to_string(entities.alias, ' '), '') || ' ' ||
					COALESCE(parents.name, '') || ' ' ||
					COALESCE(parents.slug, '') || ' ' ||
					COALESCE(array_to_string(parents.alias, ' '), '') || ' ' ||
					COALESCE(TO_CHAR(${target}s.date, 'YYYY YY MM FMMM FMMonth mon DD FMDD'), '') || ' ' ||
					STRING_AGG(COALESCE(releases.title, ''), ' ') || ' ' ||
					STRING_AGG(COALESCE(actors.name, ''), ' ') || ' ' ||
					STRING_AGG(COALESCE(tags.name, ''), ' ')
			) as document
		FROM ${target}s
		LEFT JOIN entities ON ${target}s.entity_id = entities.id
		LEFT JOIN entities AS parents ON parents.id = entities.parent_id
		LEFT JOIN ${target}s_scenes ON ${target}s_scenes.${target}_id = ${target}s.id
		LEFT JOIN releases ON releases.id = ${target}s_scenes.scene_id
		LEFT JOIN releases_actors ON releases_actors.release_id = ${target}s_scenes.scene_id
		LEFT JOIN releases_tags ON releases_tags.release_id = releases.id
		LEFT JOIN actors ON actors.id = releases_actors.actor_id
		LEFT JOIN tags ON tags.id = releases_tags.tag_id
        ${movieIds ? `WHERE ${target}s.id = ANY(?)` : ''}
		GROUP BY ${target}s.id, entities.name, entities.slug, entities.alias, parents.name, parents.slug, parents.alias;
    `, movieIds && [movieIds]);

	if (documents.rows?.length > 0) {
		await bulkInsert(`${target}s_search`, documents.rows, [`${target}_id`]);
	}
}

async function updateMovieSearch(releaseIds) {
	await knex.raw('REFRESH MATERIALIZED VIEW movies_meta;');

	await updateSqlMovieSearch(releaseIds);
	await updateManticoreMovieSearch(releaseIds);
}

module.exports = {
	updateSceneSearch,
	updateMovieSearch,
};