Added Bang Bros scraper.

src/scrapers/bangbros.js

@@ -0,0 +1,100 @@
+'use strict';
+
+/* eslint-disable newline-per-chained-call */
+const bhttp = require('bhttp');
+const cheerio = require('cheerio');
+const moment = require('moment');
+
+const knex = require('../knex');
+const { matchTags } = require('../tags');
+
+function scrapeLatest(html, site) {
+    const $ = cheerio.load(html, { normalizeWhitespace: true });
+    const sceneElements = $('.echThumb').toArray();
+
+    return sceneElements.map((element) => {
+        const sceneLinkElement = $(element).find('.thmb_lnk');
+        const title = sceneLinkElement.attr('title');
+        const url = `https://bangbros.com${sceneLinkElement.attr('href')}`;
+        const shootId = sceneLinkElement.attr('id') && sceneLinkElement.attr('id').split('-')[1];
+        const entryId = url.split('/')[3].slice(5);
+
+        const date = moment.utc($(element).find('.thmb_mr_2 span.faTxt').text(), 'MMM D, YYYY').toDate();
+        const actors = $(element).find('.cast-wrapper a.cast').map((actorIndex, actorElement) => $(actorElement).text().trim()).toArray();
+
+        const duration = moment.duration(`0:${$(element).find('.thmb_pic b.tTm').text()}`).asSeconds();
+
+        return {
+            url,
+            entryId,
+            shootId,
+            title,
+            actors,
+            date,
+            duration,
+            rating: null,
+            site,
+        };
+    });
+}
+
+async function scrapeScene(html, url, site) {
+    const $ = cheerio.load(html, { normalizeWhitespace: true });
+    const sceneElement = $('.playerSection');
+
+    const shootId = sceneElement.find('.vdoCast:contains("Release")').text().replace('Release: ', '');
+    const entryId = url.split('/')[3].slice(5);
+    const title = sceneElement.find('.ps-vdoHdd h1').text();
+    const description = sceneElement.find('.vdoDesc').text().trim();
+
+    const [siteName, ...actors] = sceneElement.find('.vdoCast a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
+    const siteId = siteName.replace(/[\s']+/g, '').toLowerCase();
+
+    const rawTags = $('.vdoTags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
+
+    const [channelSite, tags] = await Promise.all([
+        knex('sites')
+            .where({ id: siteId })
+            .orWhere({ name: siteName })
+            .first(),
+        matchTags(rawTags),
+    ]);
+
+    const stars = Number(sceneElement.find('.bVdPl_it_like .bVdPl_txt').text().replace('% like', '')) / 20;
+
+    return {
+        url,
+        shootId,
+        entryId,
+        title,
+        description,
+        actors,
+        tags,
+        rating: {
+            stars,
+        },
+        site: channelSite || site,
+    };
+}
+
+async function fetchLatest(site, page = 1) {
+    const res = await bhttp.get(`https://bangbros.com/websites/${site.id}/${page}`);
+
+    return scrapeLatest(res.body.toString(), site);
+}
+
+async function fetchScene(url, site) {
+    const { origin } = new URL(url);
+    const res = await bhttp.get(url);
+
+    if (origin !== 'https://bangbros.com') {
+        throw new Error('Cannot fetch from this URL. Please find the scene on https://bangbros.com and try again.');
+    }
+
+    return scrapeScene(res.body.toString(), url, site);
+}
+
+module.exports = {
+    fetchLatest,
+    fetchScene,
+};

src/scrapers/index.js

@@ -1,6 +1,7 @@
 'use strict';
 
 const twentyonesextury = require('./21sextury');
+const bangbros = require('./bangbros');
 const blowpass = require('./blowpass');
 const brazzers = require('./brazzers');
 const ddfnetwork = require('./ddfnetwork');
@@ -17,6 +18,7 @@ const xempire = require('./xempire');
 
 module.exports = {
     '21sextury': twentyonesextury,
+    bangbros,
     blowpass,
     brazzers,
     ddfnetwork,

src/utils/try-links.js

@@ -1,51 +0,0 @@
-'use strict';
-
-const Promise = require('bluebird');
-const bhttp = require('bhttp');
-const fs = Promise.promisifyAll(require('fs'));
-
-const knex = require('../knex');
-const argv = require('../argv');
-
-const options = {
-    responseTimeout: 30000,
-};
-
-async function tryLinks() {
-    const sites = await knex('sites').whereIn('network_id', argv.network);
-
-    const results = await Promise.all(sites.map(async (site) => {
-        console.log(`Trying ${site.name} URLs`);
-
-        const [resHttp, resHttpWww, resHttps, resHttpsWww] = await Promise.all([
-            bhttp.get(`http://${site.id}.com/`, options).catch(error => ({ statusCode: error.message })),
-            bhttp.get(`http://www.${site.id}.com/`, options).catch(error => ({ statusCode: error.message })),
-            bhttp.get(`https://${site.id}.com/`, options).catch(error => ({ statusCode: error.message })),
-            bhttp.get(`https://www.${site.id}.com/`, options).catch(error => ({ statusCode: error.message })),
-        ]);
-
-        console.log(`Got results for ${site.name}`);
-
-        return {
-            ...site,
-            url: (resHttp.statusCode === 200 && `http://${site.id}.com`)
-            || (resHttpWww.statusCode === 200 && `http://www.${site.id}.com`)
-            || (resHttps.statusCode === 200 && `https://${site.id}.com`)
-            || (resHttpsWww.statusCode === 200 && `https://www.${site.id}.com`)
-            || site.url,
-            network_id: site.network_id,
-        };
-    }));
-
-    const sortedResults = results.sort((siteA, siteB) => {
-        if (siteA.id > siteB.id) return 1;
-        if (siteA.id < siteB.id) return -1;
-
-        return 0;
-    });
-
-    console.log(sortedResults);
-    await fs.writeFileAsync('./src/utils/link-results.json', JSON.stringify(sortedResults, null, 4));
-}
-
-tryLinks();