Added Bang Bros members page scraper (not for full scenes).

src/scrapers/bangbros.js

@@ -8,6 +8,7 @@ const logger = require('../logger')(__filename);
 const slugify = require('../utils/slugify');
 const http = require('../utils/http');
 const qu = require('../utils/qu');
+const args = require('../argv');
 
 function scrape(html, site) {
 	const $ = cheerio.load(html, { normalizeWhitespace: true });
@@ -43,7 +44,7 @@ function scrape(html, site) {
 	});
 }
 
-function scrapeLegacy(scenes, site) {
+function scrapeAllLegacy(scenes, site) {
 	return scenes.map(({ query }) => {
 		const release = {};
 
@@ -63,6 +64,38 @@ function scrapeLegacy(scenes, site) {
 	});
 }
 
+function scrapeAllMembers(scenes, _channel) {
+	return scenes.map(({ query, el }) => {
+		const release = {};
+		const data = JSON.parse(query.q(el, null, 'data-shoot'));
+
+		release.entryId = data?.id || query.url('a.etLnk')?.match(/\d+$/)?.[0];
+		release.shootId = data?.code;
+		release.url = data.url ? qu.prefixUrl(data.url, 'https://members.bangbros.com') : query.url('a.etLnk');
+
+		release.title = data?.title || query.cnt('.etl-hdd');
+		release.description = data?.description || query.cnt('.etl-desc');
+
+		release.date = query.date('.etl-dt', 'MMM DD, YYYY', /\w{3} \d{1,2}, \d{4}/);
+		release.actors = data?.model.map((actor) => ({
+			name: actor.name,
+			url: qu.prefixUrl(actor.url, 'https://members.bangbros.com'),
+		}));
+
+		const rolloverUrl = query.q('.rollover-image', 'data-rollover-url');
+		release.poster = data?.image || query.img('.rollover-image', 'data-initial-image-url');
+
+		if (rolloverUrl) {
+			release.photos = Array.from({ length: 15 }, (value, index) => `${rolloverUrl}${index + 1}.jpg`);
+		}
+
+		release.trailer = data?.trailer;
+		release.tags = data?.tag.map((tag) => tag.name);
+
+		return release;
+	});
+}
+
 /* no dates available, breaks database
 function scrapeUpcoming(html, site) {
     const { document } = ex(html);
@@ -147,6 +180,30 @@ function scrapeSceneLegacy({ query }, url) {
 	return release;
 }
 
+function scrapeSceneMembers({ query }, url) {
+	const release = {};
+
+	release.entryId = new URL(url).pathname.match(/(\d+)\/?$/)[1];
+	release.shootId = query.img('.player img')?.match(/\/shoots\/(\w+)\//)?.[1];
+
+	release.title = query.cnt('.vdo-hdd1');
+	release.description = query.cnt('.ndcp');
+
+	release.actors = query.all('.vdsc a[href*="/model"]').map((actorEl) => ({
+		name: query.cnt(actorEl, 'span'),
+		url: query.url(actorEl, null, 'href', { origin: 'https://members.bangbros.com' }),
+		avatar: query.img(actorEl, 'img'),
+	}));
+
+	release.date = query.date('.ran:nth-child(2)', 'MMM DD, YYYY', /\w{3} \d{1,2}, \d{4}/);
+	release.duration = query.duration('.ran:nth-child(3)');
+
+	release.tags = query.cnts('.tag a[href*="/tags"]');
+	release.channel = slugify(query.cnt('.tag a[href*="/site"]'), '');
+
+	return release;
+}
+
 function scrapeProfile(html, scope) {
 	const { query } = qu.ex(html);
 	const profile = {};
@@ -167,17 +224,6 @@ function scrapeProfileSearch(html, actorName) {
 }
 
 async function fetchLatest(site, page = 1) {
-	if (site.parameters?.legacy) {
-		const url = `${site.parameters?.latest || site.url}/videos/${page}`;
-		const res = await qu.getAll(url, '.videoList');
-
-		if (res.ok) {
-			return scrapeLegacy(res.items, site);
-		}
-
-		return res.status;
-	}
-
 	const res = await qu.get(`${site.parameters?.latest || site.url}/${page}`);
 
 	if (res.ok) {
@@ -187,6 +233,39 @@ async function fetchLatest(site, page = 1) {
 	return res.status;
 }
 
+async function fetchLatestMembers(channel, page = 1, { parameters }) {
+	if (!parameters.product) {
+		throw new Error(`No member area product ID known for '${channel.name}'`);
+	}
+
+	if (!args.cookie) {
+		throw new Error(`Please specifiy --cookie "PHPSESSID=xxx" to access the '${channel.name}' members area.`);
+	}
+
+	const url = `https://members.bangbros.com/product/${parameters.product}/videos/latest/${page}`;
+
+	const res = await qu.getAll(url, '.thumbHolder .echThumb', {
+		cookie: args.cookie,
+	});
+
+	if (res.ok) {
+		return scrapeAllMembers(res.items, channel);
+	}
+
+	return res.status;
+}
+
+async function fetchLatestLegacy(site, page = 1) {
+	const url = `${site.parameters?.latest || site.url}/videos/${page}`;
+	const res = await qu.getAll(url, '.videoList');
+
+	if (res.ok) {
+		return scrapeAllLegacy(res.items, site);
+	}
+
+	return res.status;
+}
+
 /*
 async function fetchUpcoming(site) {
     const res = await http.get('https://www.bangbros.com');
@@ -218,6 +297,26 @@ async function fetchScene(url, site, release) {
 	return scrapeScene(res.item.html, url, site);
 }
 
+async function fetchSceneMembers(url, baseRelease, channel, { parameters }) {
+	if (!parameters.product) {
+		throw new Error(`No member area product ID known for '${channel.name}'`);
+	}
+
+	if (!args.cookie) {
+		throw new Error(`Please specifiy --cookie "PHPSESSID=xxx" to access the '${channel.name}' members area.`);
+	}
+
+	const res = await qu.get(url, null, {
+		cookie: args.cookie,
+	});
+
+	if (res.ok) {
+		return scrapeSceneMembers(res.item, url, channel);
+	}
+
+	return res.status;
+}
+
 async function fetchProfile({ name: actorName }, scope) {
 	const actorSlug = slugify(actorName);
 	const url = `https://bangbros.com/search/${actorSlug}`;
@@ -242,5 +341,12 @@ module.exports = {
 	fetchLatest,
 	fetchScene,
 	fetchProfile,
+	legacy: {
+		fetchLatest: fetchLatestLegacy,
+	},
+	members: {
+		fetchLatest: fetchLatestMembers,
+		fetchScene: fetchSceneMembers,
+	},
 	// fetchUpcoming, no dates available
 };