[scraper] add tpb fake torrent removal

scraper/lib/repository.js

@@ -170,7 +170,11 @@ function getTorrent(torrent) {
 }
 
 function getTorrentsBasedOnTitle(titleQuery, type) {
-  return Torrent.findAll({ where: { title: { [Op.regexp]: `${titleQuery}` }, type: type } });
+  return getTorrentsBasedOnQuery({ title: { [Op.regexp]: `${titleQuery}` }, type: type });
+}
+
+function getTorrentsBasedOnQuery(where) {
+  return Torrent.findAll({ where: where });
 }
 
 function getTorrentsWithoutSize() {
@@ -219,6 +223,10 @@ function setTorrentSeeders(torrent, seeders) {
   );
 }
 
+function deleteTorrent(torrent) {
+  return Torrent.destroy({ where: { infoHash: torrent.infoHash } })
+}
+
 function createFile(file) {
   if (file.id) {
     return (file.dataValues ? file.save() : File.upsert(file))
@@ -304,6 +312,8 @@ module.exports = {
   setTorrentSeeders,
   getTorrent,
   getTorrentsBasedOnTitle,
+  getTorrentsBasedOnQuery,
+  deleteTorrent,
   getUpdateSeedersTorrents,
   getNoContentsTorrents,
   createFile,

scraper/scheduler/scrapers.js

@@ -1,4 +1,5 @@
 const thepiratebayScraper = require('../scrapers/thepiratebay/thepiratebay_scraper');
+const thepiratebayFakeRemoval = require('../scrapers/thepiratebay/thepiratebay_fakes_removal');
 const ytsScraper = require('../scrapers/yts/yts_scraper');
 const eztvScraper = require('../scrapers/eztv/eztv_scraper');
 const leetxScraper = require('../scrapers/1337x/1337x_scraper');
@@ -17,6 +18,7 @@ module.exports = [
   { scraper: rarbgScraper, name: rarbgScraper.NAME, cron: '0 0 */1 ? * *' },
   { scraper: rutorScraper, name: rutorScraper.NAME, cron: '0 0 */4 ? * *' },
   { scraper: thepiratebayScraper, name: thepiratebayScraper.NAME, cron: '0 0 */2 ? * *' },
+  { scraper: thepiratebayFakeRemoval, name: thepiratebayFakeRemoval.NAME, cron: '0 30 1 ? * * *' },
   { scraper: torrentGalaxyScraper, name: torrentGalaxyScraper.NAME, cron: '0 0 */4 ? * *' },
   { scraper: leetxScraper, name: leetxScraper.NAME, cron: '0 0 */4 ? * *' },
   // { scraper: kickassScraper, name: kickassScraper.NAME, cron: '0 0 */4 ? * *' },

scraper/scrapers/thepiratebay/thepiratebay_fakes_removal.js

@@ -0,0 +1,42 @@
+const moment = require('moment');
+const { Sequelize } = require('sequelize');
+const thepiratebay = require('./thepiratebay_api.js');
+const { Type } = require('../../lib/types');
+const repository = require('../../lib/repository');
+const Promises = require('../../lib/promises');
+
+const NAME = 'ThePirateBay';
+const EMPTY_HASH = '0000000000000000000000000000000000000000';
+
+const Op = Sequelize.Op;
+
+async function scrape() {
+  console.log(`Starting ${NAME} fake removal...`);
+  const startCreatedAt = moment().subtract(14, 'day');
+  const endCreatedAt = moment().subtract(1, 'day');
+  const whereQuery = {
+    provider: NAME,
+    type: Type.MOVIE,
+    createdAt: { [Op.between]: [startCreatedAt, endCreatedAt] }
+  };
+  return repository.getTorrentsBasedOnQuery(whereQuery)
+      .then(torrents => {
+        console.log(`Checking for ${NAME} fake entries in ${torrents.length} torrents`);
+        return Promises.sequence(torrents.map(torrent => () => removeIfFake(torrent)))
+      })
+      .then(results => {
+        const removed = results.filter(result => result);
+        console.log(`Finished ${NAME} fake removal with ${removed.length} removals in ${results.length} torrents`);
+      });
+}
+
+async function removeIfFake(torrent) {
+  const tpbTorrentInfo = await thepiratebay.torrent(torrent.torrentId).catch(() => null);
+  if (tpbTorrentInfo && tpbTorrentInfo.infoHash === EMPTY_HASH) {
+    console.log(`Removing ${NAME} fake torrent [${torrent.torrentId}][${torrent.infoHash}] ${torrent.title}`);
+    return repository.deleteTorrent(torrent).catch(() => null);
+  }
+  return Promise.resolve(null);
+}
+
+module.exports = { scrape, NAME };