Improved and documented actor profile scraping.

src/actors.js

@@ -582,9 +582,31 @@ async function scrapeProfiles(actor, sources, entitiesBySlug, existingProfilesBy
 	return profiles.filter(Boolean);
 }
 
-async function scrapeActors(actorNames) {
+async function getActorNames(actorNames) {
+	if (actorNames.length > 0) {
+		return actorNames;
+	}
+
+	const actorsWithoutProfiles = await knex.raw(`
+		SELECT actors.name
+		FROM actors
+		WHERE NOT EXISTS (
+			SELECT *
+			FROM actors_profiles
+			WHERE actors_profiles.actor_id = actors.id
+			AND actors_profiles.updated_at <= (?)
+		)
+	`, [argv.actorsUpdate || new Date()]);
+
+	return actorsWithoutProfiles.rows.map(actor => actor.name);
+}
+
+async function scrapeActors(argNames) {
+	const actorNames = await getActorNames(argNames);
 	const baseActors = toBaseActors(actorNames);
 
+	logger.info(`Scraping profiles for ${actorNames.length} actors`);
+
 	const sources = argv.sources || config.profiles || Object.keys(scrapers.actors);
 	const entitySlugs = sources.flat();
 
@@ -596,11 +618,7 @@ async function scrapeActors(actorNames) {
 			.orderBy('entities.type'),
 		knex('actors')
 			.select(['id', 'name', 'slug', 'entry_id'])
-			.modify((queryBuilder) => {
-				if (actorNames.length > 0) {
-					queryBuilder.whereIn('slug', baseActors.map(baseActor => baseActor.slug));
-				}
-			})
+			.whereIn('slug', baseActors.map(baseActor => baseActor.slug))
 			.whereNull('alias_for'),
 	]);