Added Teen Mega World scraper.

src/actors.js

@@ -278,8 +278,8 @@ function curateActorEntry(baseActor, batchId) {
 	return {
 		name: baseActor.name,
 		slug: baseActor.slug,
-		entity_id: collisionLikely ? baseActor.entity.id : null,
-		entry_id: collisionLikely ? baseActor.entryId : null,
+		entity_id: (collisionLikely && baseActor.entity?.id) || null,
+		entry_id: (collisionLikely && baseActor.entryId) || null,
 		batch_id: batchId,
 	};
 }

src/scrapers/scrapers.js

@@ -54,6 +54,7 @@ const score = require('./score');
 const spizoo = require('./spizoo');
 const teamskeet = require('./teamskeet');
 const teencoreclub = require('./teencoreclub');
+const teenmegaworld = require('./teenmegaworld');
 const topwebmodels = require('./topwebmodels');
 const traxxx = require('./traxxx');
 const vivid = require('./vivid');
@@ -138,6 +139,7 @@ const scrapers = {
 		spizoo,
 		swallowsalon: julesjordan,
 		teencoreclub,
+		teenmegaworld,
 		teamskeet,
 		topwebmodels,
 		transbella: porndoe,
@@ -262,6 +264,7 @@ const scrapers = {
 		swallowed: mikeadriano,
 		teamskeet,
 		teencoreclub,
+		teenmegaworld,
 		thatsitcomshow: nubiles,
 		topwebmodels,
 		transangels: mindgeek,

src/scrapers/teenmegaworld.js

@@ -0,0 +1,118 @@
+'use strict';
+
+const qu = require('../utils/q');
+const slugify = require('../utils/slugify');
+
+function scrapeAll(scenes, channel) {
+	return scenes.map(({ query }) => {
+		const release = {};
+		const network = channel.type === 'network' ? channel : channel.parent;
+
+		release.url = query.url('a.title');
+		release.entryId = new URL(release.url).pathname.match(/\/trailers\/(.+).html/)?.[1].toLowerCase();
+
+		release.title = query.cnt('.title');
+
+		release.date = query.date('time', 'MMMM D, YYYY');
+		release.actors = query.all('.actors a').map(el => ({
+			name: query.cnt(el),
+			url: query.url(el, null),
+		}));
+
+		[release.poster, ...release.photos] = Object
+			.entries(query.el('.images img').dataset)
+			.filter(([key]) => /src/.test(key))
+			.map(([, value]) => qu.prefixUrl(value, network.url));
+
+		release.likes = query.number('.rating-on-thumb');
+
+		const siteId = query.url('.site a', 'href', { origin: network.url, object: true })?.searchParams.get('site[]');
+
+		if (siteId) {
+			release.channel = network.children.find(child => child.parameters.siteId.toString() === siteId)?.slug;
+		}
+
+		return release;
+	});
+}
+
+function scrapeScene({ query }, url, channel) {
+	const release = {};
+	const network = channel.type === 'network' ? channel : channel.parent;
+
+	release.entryId = new URL(url).pathname.match(/\/trailers\/(.+).html/)[1].toLowerCase();
+
+	release.title = query.cnt('.title-line h1');
+	release.description = query.cnt('.shadow .text');
+
+	release.date = query.date('.title-line .date', 'MMMM D, YYYY');
+	release.duration = query.number('.dur') * 60;
+
+	release.actors = query.all('.site a[href*="/models"]').map(el => ({
+		name: query.cnt(el),
+		url: query.url(el, null),
+	}));
+
+	release.tags = query.cnts('.tag-list a');
+
+	release.poster = query.poster();
+	release.trailer = query.video();
+
+	release.likes = query.number('.rating-data .value');
+
+	const siteId = query.url('.site a[href*="site[]"]', 'href', { origin: network.url, object: true })?.searchParams.get('site[]');
+
+	if (siteId) {
+		release.channel = network.children.find(child => child.parameters.siteId.toString() === siteId)?.slug;
+	}
+
+	return release;
+}
+
+function scrapeProfile({ query, el }, url, entity) {
+	const profile = {};
+
+	const bio = query.all('.stats li').reduce((acc, bioEl) => ({
+		...acc,
+		[slugify(query.cnt(bioEl, '.title'))]: query.cnt(bioEl, '.value'),
+	}), {});
+
+	profile.url = url;
+	profile.description = query.cnt('.bio .text');
+
+	profile.hairColor = bio.hair;
+	profile.eyes = bio.eyes;
+
+	profile.avatar = query.img('.model-info .photo img', 'data-src', { origin: entity.url });
+	profile.scenes = scrapeAll(qu.initAll(el, '.video-list .video[data-type="vids"]'), entity);
+
+	return profile;
+}
+
+async function fetchLatest(channel, page = 1) {
+	const url = `${channel.type === 'network' ? channel.url : channel.parent.url}/search.php?site[]=${channel.parameters.siteId}&page=${page}`;
+	const res = await qu.getAll(url, '.video-list .video');
+
+	if (res.ok) {
+		return scrapeAll(res.items, channel);
+	}
+
+	return res.status;
+}
+
+async function fetchProfile({ name: actorName }, entity, include) {
+	const url = `${entity.url}/models/${slugify(actorName)}.html`;
+	const res = await qu.get(url);
+
+	if (res.ok) {
+		return scrapeProfile(res.item, url, entity, include);
+	}
+
+	return res.status;
+}
+
+module.exports = {
+	fetchLatest,
+	fetchProfile,
+	scrapeScene,
+};

src/utils/qu.js

@@ -275,10 +275,15 @@ function images(context, selector = 'img', attr, { origin, protocol = 'https' }
 	return imageEls.map(imageEl => prefixUrl(imageEl, origin, protocol));
 }
 
-function url(context, selector = 'a', attr = 'href', { origin, protocol = 'https' } = {}) {
+function url(context, selector = 'a', attr = 'href', { origin, protocol = 'https', object = false } = {}) {
 	const urlEl = q(context, selector, attr);
+	const prefixedUrl = prefixUrl(urlEl, origin, protocol);
 
-	return prefixUrl(urlEl, origin, protocol);
+	if (prefixedUrl && object) {
+		return new URL(prefixedUrl);
+	}
+
+	return prefixedUrl;
 }
 
 function urls(context, selector = 'a', attr = 'href', { origin, protocol = 'https' } = {}) {
@@ -349,7 +354,10 @@ function videos(context, selector = 'source', attr = 'src', { origin, protocol =
 function duration(context, selector, match, attr = 'textContent') {
 	const durationString = q(context, selector, attr);
 
-	if (!durationString) return null;
+	if (!durationString) {
+		return null;
+	}
+
 	const durationMatch = durationToSeconds(durationString, match);
 
 	if (durationMatch) {