Added profile flush.

src/actors.js

@@ -7,6 +7,7 @@ const moment = require('moment');
 const blake2 = require('blake2');
 const DOMPurify = require('dompurify');
 const { JSDOM } = require('jsdom');
+const omit = require('object.omit');
 
 const { window } = new JSDOM('');
 const domPurify = DOMPurify(window);
@@ -21,7 +22,7 @@ const bulkInsert = require('./utils/bulk-insert');
 const logger = require('./logger')(__filename);
 
 const { toBaseReleases } = require('./deep');
-const { associateAvatars } = require('./media');
+const { associateAvatars, flushOrphanedMedia } = require('./media');
 
 const slugify = require('./utils/slugify');
 const capitalize = require('./utils/capitalize');
@@ -436,18 +437,26 @@ async function curateProfile(profile, actor) {
 	}
 }
 
-async function interpolateProfiles(actorIdsOrNames) {
-	const profiles = await knex('actors_profiles')
-		.select('actors_profiles.*', knex.raw('row_to_json(media) as avatar'))
+async function fetchProfiles(actorIdsOrNames) {
+	return knex('actors_profiles')
+		.select(knex.raw('actors_profiles.*, row_to_json(actors) as actor, row_to_json(media) as avatar'))
 		.leftJoin('actors', 'actors.id', 'actors_profiles.actor_id')
 		.modify((query) => {
-			if (actorIdsOrNames?.length > 0) {
+			if (actorIdsOrNames) {
 				query
 					.whereIn('actor_id', actorIdsOrNames.filter(idOrName => typeof idOrName === 'number'))
-					.orWhereIn('actors.name', actorIdsOrNames.filter(idOrName => typeof idOrName === 'string'));
+					.orWhere((builder) => {
+						builder
+							.whereIn('actors.name', actorIdsOrNames.filter(idOrName => typeof idOrName === 'string'))
+							.whereNull('actors.entity_id');
+					});
 			}
 		})
 		.leftJoin('media', 'actors_profiles.avatar_media_id', 'media.id');
+}
+
+async function interpolateProfiles(actorIdsOrNames) {
+	const profiles = await fetchProfiles(actorIdsOrNames);
 
 	const profilesByActorId = profiles.reduce((acc, profile) => ({
 		...acc,
@@ -545,6 +554,27 @@ async function interpolateProfiles(actorIdsOrNames) {
 
 	const transaction = await knex.transaction();
 
+	// clear existing interpolated data
+	const emptyProfile = Object
+		.keys(omit(curateProfileEntry({ id: 1 }), ['id', 'actor_id', 'entity_id', 'url', 'description_hash']))
+		.reduce((acc, key) => ({ ...acc, [key]: null }), {});
+
+	await knex('actors')
+		.modify((modifyBuilder) => {
+			if (actorIdsOrNames) {
+				modifyBuilder
+					.whereIn('id', actorIdsOrNames.filter(idOrName => typeof idOrName === 'number'))
+					.orWhere((whereBuilder) => {
+						whereBuilder
+							.whereIn('name', actorIdsOrNames.filter(idOrName => typeof idOrName === 'string'))
+							.whereNull('entity_id');
+					});
+			}
+		})
+		.update(emptyProfile)
+		.transacting(transaction);
+
+	// insert new interpolated data
 	const queries = interpolatedProfiles.map(profile => knex('actors')
 		.where('id', profile.id)
 		.update(profile)
@@ -689,6 +719,30 @@ async function storeProfiles(profiles) {
 	await interpolateProfiles(actorIds);
 }
 
+async function flushProfiles(actorIdsOrNames) {
+	const profiles = await fetchProfiles(actorIdsOrNames);
+	const actorNames = Array.from(new Set(profiles.map(profile => profile.actor.name)));
+
+	const deleteCount = await knex('actors_profiles')
+		.whereIn('id', profiles.map(profile => profile.id))
+		.delete();
+
+	await interpolateProfiles(actorIdsOrNames);
+	await flushOrphanedMedia(); // don't flush until main avatar is detached by re-interpolating
+
+	if (actorNames.length > 20) {
+		logger.info(`Removed ${deleteCount} profiles for ${actorNames.length} actors`);
+		return;
+	}
+
+	if (deleteCount > 0) {
+		logger.info(`Removed ${deleteCount} profiles for ${actorNames.join(', ')}`);
+		return;
+	}
+
+	logger.info(`Removed ${deleteCount} profiles`);
+}
+
 async function scrapeActors(argNames) {
 	const actorNames = await getActorNames(argNames);
 	const baseActors = toBaseActors(actorNames);
@@ -905,8 +959,9 @@ async function searchActors(query) {
 module.exports = {
 	associateActors,
 	fetchActor,
+	flushProfiles,
+	interpolateProfiles,
 	scrapeActors,
 	searchActors,
 	toBaseActors,
-	interpolateProfiles,
 };

src/app.js

@@ -9,7 +9,7 @@ const knex = require('./knex');
 const fetchUpdates = require('./updates');
 const { fetchScenes, fetchMovies } = require('./deep');
 const { storeScenes, storeMovies, updateReleasesSearch } = require('./store-releases');
-const { scrapeActors, interpolateProfiles } = require('./actors');
+const { scrapeActors, flushProfiles, interpolateProfiles } = require('./actors');
 const { flushEntities } = require('./entities');
 const { deleteScenes, deleteMovies, flushBatches } = require('./releases');
 const { flushOrphanedMedia } = require('./media');
@@ -26,7 +26,11 @@ async function init() {
 	}
 
 	if (argv.interpolateProfiles) {
-		await interpolateProfiles(argv.interpolateProfiles);
+		await interpolateProfiles(argv.flushProfiles.length > 0 ? argv.flushProfiles : null);
+	}
+
+	if (argv.flushProfiles) {
+		await flushProfiles(argv.flushProfiles.length > 0 ? argv.flushProfiles : null);
 	}
 
 	if (argv.flushNetworks || argv.flushChannels) {

src/argv.js

@@ -268,6 +268,11 @@ const { argv } = yargs
 		type: 'array',
 		alias: 'flush-network',
 	})
+	.option('flush-profiles', {
+		describe: 'Delete all profiles for an actor.',
+		type: 'array',
+		alias: ['flush-profile', 'flush-actors', 'flush-actor'],
+	})
 	.option('flush-batches', {
 		describe: 'Delete all scenes and movies from batch by ID.',
 		type: 'array',

src/media.js

@@ -779,6 +779,7 @@ async function flushOrphanedMedia() {
 							knex('releases_teasers').select('media_id'),
 							knex('movies_covers').select('media_id'),
 							knex('movies_trailers').select('media_id'),
+							knex('actors').select(knex.raw('avatar_media_id as media_id')),
 							knex('actors_profiles').select(knex.raw('avatar_media_id as media_id')),
 							knex('actors_photos').select('media_id'),
 							knex('clips_photos').select('media_id'),