Added Men network and Icon Male to MindGeek. Added entropy filter to media module to help filter out generic avatars. Added Pure Taboo. Various logo updates.

src/scrapers/puretaboo.js

@@ -0,0 +1,103 @@
+'use strict';
+
+const bhttp = require('bhttp');
+const moment = require('moment');
+
+// const { fetchLatest } = require('./gamma');
+const { ex, ctxa } = require('../utils/q');
+const { getPhotos } = require('./gamma');
+
+function scrape(html) {
+    const { document } = ex(html);
+
+    return ctxa(document, '.sceneList .sceneContainer').map(({ q, qa }) => {
+        const release = {};
+
+        const linkEl = q('a.imgLink');
+
+        release.entryId = linkEl.dataset.id;
+        release.title = linkEl.title;
+        release.url = `https://www.puretaboo.com${linkEl.href}`;
+        release.actors = qa('.sceneActors a', true);
+
+        release.poster = q('.imgLink img').dataset.original;
+        release.trailer = `https://videothumb.gammacdn.com/307x224/${release.entryId}.mp4`;
+
+        return release;
+    });
+}
+
+async function scrapeScene(html, url, site) {
+    const release = {};
+
+    const { q, qa, qd, qm } = ex(html);
+    const { pathname } = new URL(url);
+
+    const dataPrefix = 'dataLayer = [';
+    const dataStart = html.indexOf(dataPrefix);
+    const dataString = html.slice(dataStart + dataPrefix.length, html.indexOf('];', dataStart));
+    const { sceneDetails: data } = JSON.parse(dataString);
+
+    const dataString2 = q('script[type="application/ld+json"]', true);
+    const data2 = JSON.parse(dataString2)[1];
+
+    const videoPrefix = 'window.ScenePlayerOptions = ';
+    const videoStart = html.indexOf(videoPrefix);
+    const videoString = html.slice(videoStart + videoPrefix.length, html.indexOf('};', videoStart) + 1);
+    const videoData = JSON.parse(videoString);
+
+    release.entryId = data.sceneId || videoData.playerOptions.sceneInfos.sceneId || pathname.slice(pathname.lastIndexOf('/') + 1);
+    release.url = url;
+    release.title = data.sceneTitle || videoData.playerOptions.sceneInfos.sceneTitle || qm('meta[name="twitter:title"]') || q('.sceneTitle', true);
+    release.description = data.sceneDescription || data2.description || q('.sceneDesc', true).replace('Video Description:', '');
+
+    release.duration = moment.duration(data2.duration.slice(2)).asSeconds();
+
+    // data2.dateCreated is the date the scene was first added to 'upcoming'
+    release.date = videoData.playerOptions.sceneInfos.sceneReleaseDate
+        ? moment.utc(videoData.playerOptions.sceneInfos.sceneReleaseDate, 'YYYY-MM-DD').toDate()
+        : qd('.updatedDate', 'MM-DD-YYYY');
+
+    release.actors = data.sceneActors?.map(actor => actor.actorName)
+        || data2.actor?.map(actor => actor.name)
+        || qa('.sceneColActors a', true);
+
+    release.tags = data2.keywords.split(', ') || qa('.sceneColCategories a', 'title');
+
+    release.poster = videoData.picPreview;
+    release.trailer = {
+        src: `${videoData.playerOptions.host}${videoData.url}`,
+    };
+
+    release.photos = await getPhotos(q('.picturesItem a').href, 'puretaboo.com', site);
+
+    return release;
+}
+
+async function fetchLatest(site, page = 1) {
+    const res = await bhttp.get(`${site.url}/en/videos/AllCategories/0/${page}`);
+
+    if (res.statusCode === 200) {
+        return scrape(res.body.toString(), site);
+    }
+
+    return null;
+}
+
+async function fetchUpcoming(site) {
+    const res = await bhttp.get(`${site.url}/en/videos/AllCategories/0/1/upcoming`);
+
+    return scrape(res.body.toString(), site);
+}
+
+async function fetchScene(url, site) {
+    const res = await bhttp.get(url);
+
+    return scrapeScene(res.body.toString(), url, site);
+}
+
+module.exports = {
+    fetchLatest,
+    fetchScene,
+    fetchUpcoming,
+};