Scraping and storing actor profiles.

src/utils/posters.js

@@ -8,10 +8,10 @@ const moment = require('moment');
 const argv = require('../argv');
 const knex = require('../knex');
 
-async function init() {
+async function actorPosters(actorNames) {
 	const posters = await knex('actors')
 		.select('actors.name as actor_name', 'releases.title', 'releases.date', 'media.path', 'media.index', 'sites.name as site_name', 'networks.name as network_name')
-		.whereIn('actors.name', (argv.actors || []).concat(argv._))
+		.whereIn('actors.name', actorNames)
 		.join('releases_actors', 'releases_actors.actor_id', 'actors.id')
 		.join('releases', 'releases_actors.release_id', 'releases.id')
 		.join('sites', 'sites.id', 'releases.site_id')
@@ -37,4 +37,41 @@ async function init() {
 	knex.destroy();
 }
 
+async function sitePosters(siteSlugs) {
+	const posters = await knex('sites')
+		.select('sites.name as site_name', 'releases.title', 'releases.date', 'media.path')
+		.whereIn('sites.slug', siteSlugs)
+		.join('releases', 'releases.site_id', 'sites.id')
+		.join('releases_posters', 'releases_posters.release_id', 'releases.id')
+		.join('media', 'releases_posters.media_id', 'media.id');
+		// .where('releases.date', '<', '2020-01-01');
+
+	const files = await Promise.all(posters.map(async (poster) => {
+		const directory = path.join(config.media.path, 'extracted', poster.site_name);
+
+		const source = path.join(config.media.path, poster.path);
+		const target = path.join(directory, `${poster.site_name} - ${moment.utc(poster.date).format('YYYY-MM-DD')} - ${poster.title.replace(/[/.]/g, '_')}.jpeg`);
+
+		await fs.mkdir(directory, { recursive: true });
+		await fs.copyFile(source, target);
+
+		return target;
+	}));
+
+	console.log(files);
+
+	knex.destroy();
+}
+
+async function init() {
+	if (argv.actors) {
+		await actorPosters(argv.actors);
+		return;
+	}
+
+	if (argv.sites) {
+		await sitePosters(argv.sites);
+	}
+}
+
 init();