Added DDFNetwork scraper. Modified tag matching query to be case insensitive.

src/fetch-releases.js

@@ -121,7 +121,6 @@ async function fetchNewReleases(scraper, site, afterDate, accReleases = [], page
 
 async function fetchReleases() {
     const sites = await accumulateIncludedSites();
-    // const releases = await getExistingReleases();
 
     const scenesPerSite = await Promise.all(sites.map(async (site) => {
         const scraper = scrapers[site.id] || scrapers[site.networkId];

src/fetch-scene.js

@@ -13,19 +13,19 @@ async function findSite(url) {
         .where({ url: `${protocol}//www.${hostname}` })
         .orWhere({ url: `${protocol}//${hostname}` })
         .first()
-        // scenes might be listed on network site, let network scraper find channel site
+        // scene might use generic network URL, let network scraper determine channel site
         || await knex('networks')
             .where({ url: `${protocol}//www.${hostname}` })
             .orWhere({ url: `${protocol}//${hostname}` })
             .first();
 
-
     return {
         id: site.id,
         name: site.name,
         description: site.description,
         url: site.url,
         networkId: site.network_id || site.id,
+        isFallback: site.network_id === undefined,
     };
 }
 

src/scrapers/ddfnetwork.js

@@ -0,0 +1,91 @@
+'use strict';
+
+/* eslint-disable */
+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 = $('.card.m-1').toArray();
+
+    return sceneElements.map((element) => {
+        const sceneLinkElement = $(element).find('a').first();
+        const title = sceneLinkElement.attr('title');
+        const url = `${site.url}${sceneLinkElement.attr('href')}`;
+        const entryId = url.split('/').slice(-1)[0];
+
+        const date = moment.utc($(element).find('.card-footer .text-muted').text(), 'MMMM DD, YYYY').toDate();
+        const actors = $(element).find('.card-subtitle a').map((actorIndex, actorElement) => $(actorElement).text().trim()).toArray().filter(actor => actor);
+
+        const duration = Number($(element).find('.card-info div:nth-child(2) .card-text').text().slice(0, -4)) * 60;
+
+        return {
+            url,
+            entryId,
+            title,
+            actors,
+            date,
+            duration,
+            rating: null,
+            site,
+        };
+    });
+}
+
+async function scrapeScene(html, url, site) {
+    const $ = cheerio.load(html, { normalizeWhitespace: true });
+
+    const entryId = url.split('/').slice(-1)[0];
+    const title = $('.video-title h1').text();
+    const description = $('.description-box .box-container').text();
+
+    const date = moment.utc($('.video-title .remain time').text(), 'MMMM DD, YYYY').toDate();
+    const actors = $('.pornstars-box .pornstar-card .card-title a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
+
+    const likes = Number($('.info-panel.likes .likes').text());
+    const duration = Number($('.info-panel.duration .duration').text().slice(0, -4)) * 60;
+
+    const { origin } = new URL($('.pornstar-card meta[itemprop="url"]').first().attr('content'));
+    const rawTags = $('#tagsBox .tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
+
+    const [channelSite, tags] = await Promise.all([
+        // don't find site if original is already specific
+        site.isFallback ? knex('sites').where({ url: origin }).first() : site,
+        matchTags(rawTags),
+    ]);
+
+    return {
+        url: channelSite ? `${channelSite.url}${new URL(url).pathname}` : url,
+        entryId,
+        title,
+        actors,
+        date,
+        duration,
+        tags,
+        rating: {
+            likes,
+        },
+        site: channelSite || site,
+    };
+}
+
+async function fetchLatest(site, page = 1) {
+    const res = await bhttp.get(`https://ddfnetwork.com/videos/search/latest/ever/${new URL(site.url).hostname}/-/${page}`);
+
+    return scrapeLatest(res.body.toString(), site);
+}
+
+async function fetchScene(url, site) {
+    const res = await bhttp.get(`https://ddfnetwork.com${new URL(url).pathname}`);
+
+    return scrapeScene(res.body.toString(), url, site);
+}
+
+module.exports = {
+    fetchLatest,
+    fetchScene,
+};

src/scrapers/index.js

@@ -2,6 +2,7 @@
 
 const blowpass = require('./blowpass');
 const brazzers = require('./brazzers');
+const ddfnetwork = require('./ddfnetwork');
 const julesjordan = require('./julesjordan');
 const kink = require('./kink');
 const legalporno = require('./legalporno');
@@ -14,6 +15,7 @@ const xempire = require('./xempire');
 module.exports = {
     blowpass,
     brazzers,
+    ddfnetwork,
     julesjordan,
     kink,
     legalporno,

src/scrapers/kink.js

@@ -49,8 +49,6 @@ async function scrapeScene(html, url, shootId, ratingRes, site) {
     const title = $('h1.shoot-title span.favorite-button').attr('data-title');
     const actorsRaw = $('.shoot-info p.starring');
 
-    const sitename = $('.shoot-logo a').attr('href').split('/')[2];
-
     const date = moment.utc($(actorsRaw)
         .prev()
         .text()
@@ -64,6 +62,7 @@ async function scrapeScene(html, url, shootId, ratingRes, site) {
 
     const { average: stars } = ratingRes.body;
 
+    const sitename = $('.shoot-logo a').attr('href').split('/')[2];
     const rawTags = $('.tag-list > a[href*="/tag"]').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
 
     const [channelSite, tags] = await Promise.all([

src/scrapers/template.js

@@ -12,6 +12,8 @@ function scrapeLatest(html, site) {
     const sceneElements = $('.scenes-latest').toArray();
 
     return sceneElements.map((element) => {
+        const actors = $('.actors a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
+
         return {
             url,
             shootId,
@@ -33,6 +35,8 @@ function scrapeUpcoming(html, site) {
     const sceneElements = $('.scenes-upcoming').toArray();
 
     return sceneElements.map((element) => {
+        const actors = $('.actors a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
+
         return {
             url,
             shootId,
@@ -55,6 +59,11 @@ async function scrapeScene(html, url, site) {
     const rawTags = [];
     const tags = await matchTags(rawTags);
 
+    const actors = $('.actors a').map((actorIndex, actorElement) => $(actorElement).text()).toArray();
+
+    const rawTags = $('.tags a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
+    const tags = await matchTags(rawTags);
+
     return {
         url,
         shootId,

src/tags.js

@@ -3,11 +3,14 @@
 const knex = require('./knex');
 
 async function matchTags(rawTags) {
-    const tagEntries = await knex('tags')
+    const tagQuery = knex('tags')
         .select(knex.raw('ifnull(original.tag, tags.tag) as tag'), knex.raw('ifnull(original.tag, tags.tag) as tag'))
         .whereIn('tags.tag', rawTags)
-        .orWhereIn('tags.tag', rawTags.map(tag => tag.toLowerCase()))
-        .leftJoin('tags as original', 'tags.alias_for', 'original.tag');
+        .leftJoin('tags as original', 'tags.alias_for', 'original.tag')
+        .toString()
+        .replace('where `tags`.`tag` in', 'where `tags`.`tag` collate NOCASE in');
+
+    const tagEntries = await knex.raw(tagQuery);
 
     return Array.from(new Set(tagEntries.map(({ tag }) => tag))).sort(); // reduce to tag name and filter duplicates
 }