Added Mike Adriano scraper (True Anal, All Anal, Swallowed, Nympho). Updated README.

src/scrapers/realitykings.js

@@ -4,38 +4,71 @@
 const Promise = require('bluebird');
 const bhttp = require('bhttp');
 const { CookieJar } = Promise.promisifyAll(require('tough-cookie'));
-const cheerio = require('cheerio');
-const moment = require('moment');
+const { JSDOM } = require('jsdom');
 
 const { matchTags } = require('../tags');
 
-function scrapeLatest(html, site) {
-    const $ = cheerio.load(html, { normalizeWhitespace: true });
-    const sceneElements = $('.card.card--release').toArray();
+async function scrapeLatest(html, site) {
+    const { document } = new JSDOM(html).window;
 
-    console.log(sceneElements);
+    const scriptString = document.querySelector('script').textContent;
+    const prefix = 'window.__JUAN.initialState = {';
 
-    return sceneElements.map((element) => {
-        const sceneLinkElement = $(element).find('.card-info__title a');
-        const title = sceneLinkElement.attr('title');
-        const url = `${site.url}${sceneLinkElement.attr('href')}`;
-        const entryId = url.split('/').slice(-3)[0];
+    const dataStart = scriptString.slice(scriptString.indexOf(prefix) + prefix.length - 1);
+    const dataString = dataStart.slice(0, dataStart.indexOf('};') + 1);
+    const data = JSON.parse(dataString);
 
-        const date = moment.utc($(element).find('.card-info__meta-date').text(), 'MMMM DD, YYYY').toDate();
-        const actors = $(element).find('.card-info__cast a').map((actorIndex, actorElement) => $(actorElement).text().trim()).toArray();
+    const actorsMap = data.entities.actors;
+    const tagsMap = data.entities.tags;
+    const scenes = Object.values(data.entities.releases);
 
-        console.log(date, actors, title);
+    return Promise.all(scenes.map(async (scene) => {
+        const {
+            id: entryId,
+            title,
+            description,
+        } = scene;
+
+        const url = `https://www.realitykings.com/scene/${entryId}`;
+        const date = new Date(scene.dateReleased);
+        const actors = scene.actors.map(actorId => actorsMap[actorId].name);
+
+        const rawTags = scene.tags.map(tagId => tagsMap[tagId].name);
+        const tags = await matchTags(rawTags);
+
+        if (!scene.images.poster) {
+            console.log(site.name, site.id);
+            console.log(scene);
+            console.log(title, url, scene.images);
+        }
+
+        const [poster, ...photos] = scene.images.poster.map(image => image.xl.url);
+
+        const duration = scene.videos.mediabook.length;
+        const trailer720p = scene.videos.mediabook.files['720p'] && scene.videos.mediabook.files['720p'].urls.view;
+        const trailer360p = scene.videos.mediabook.files['360p'] && scene.videos.mediabook.files['360p'].urls.view;
+
+        const { likes, dislikes } = scene.stats;
 
         return {
             url,
             entryId,
             title,
+            description,
             actors,
             date,
-            rating: null,
+            tags,
+            duration,
+            poster,
+            photos,
+            trailer: {
+                src: trailer720p || trailer360p,
+                quality: trailer720p ? 720 : 360,
+            },
+            rating: { likes, dislikes },
             site,
         };
-    });
+    }));
 }
 
 async function scrapeScene(data, url, site) {
@@ -46,23 +79,19 @@ async function scrapeScene(data, url, site) {
     } = data;
 
     const date = new Date(data.dateReleased);
-    const actors = data.actors
-        .sort(({ gender: genderA }, { gender: genderB }) => {
-            if (genderA === 'female' && genderB === 'male') return -1;
-            if (genderA === 'male' && genderB === 'female') return 1;
-
-            return 0;
-        })
-        .map(actor => actor.name);
+    const actors = data.actors.map(actor => actor.name);
 
     const { likes, dislikes } = data.stats;
-    const duration = data.videos.mediabook.length;
-
-    console.log(data);
 
     const rawTags = data.tags.map(tag => tag.name);
     const tags = await matchTags(rawTags);
 
+    const [poster, ...photos] = data.images.poster.map(image => image.xl.url);
+
+    const duration = data.videos.mediabook.length;
+    const trailer720p = data.videos.mediabook.files['720p'] && data.videos.mediabook.files['720p'].urls.view;
+    const trailer360p = data.videos.mediabook.files['360p'] && data.videos.mediabook.files['360p'].urls.view;
+
     return {
         url,
         entryId,
@@ -72,6 +101,12 @@ async function scrapeScene(data, url, site) {
         date,
         duration,
         tags,
+        poster,
+        photos,
+        trailer: {
+            src: trailer720p || trailer360p,
+            quality: trailer720p ? 720 : 360,
+        },
         rating: {
             likes,
             dislikes,
@@ -81,13 +116,25 @@ async function scrapeScene(data, url, site) {
 }
 
 async function fetchLatest(site, page = 1) {
-    const res = await bhttp.get(`https://www.realitykings.com/tour/videos/${site.name.replace(/\s+/g, '-').toLowerCase()}/all-categories/all-time/recent/${page}`);
+    const { hostname, search } = new URL(site.url);
 
-    return scrapeLatest(res.body.toString(), site);
+    if (hostname.match(/(www\.)?realitykings\.com/) && search.match(/\?site=\d+/)) {
+        const res = await bhttp.get(`${site.url}&page=${page}`);
+
+        return scrapeLatest(res.body.toString(), site);
+    }
+
+    if (site.parameters && site.parameters.siteId) {
+        const res = await bhttp.get(`https://www.realitykings.com/scenes?site=${site.parameters.siteId}&page=${page}`);
+
+        return scrapeLatest(res.body.toString(), site);
+    }
+
+    throw new Error(`Reality Kings site '${site.name}' (${site.url}) not supported`);
 }
 
 async function fetchScene(url, site) {
-    if (site.isFallback || (site.parameters && site.parameters.altLayout)) {
+    if (site.isFallback) {
         throw new Error('Cannot fetch scene details from this resource');
     }