adds horriblesubs scrapper

This commit is contained in:
TheBeastLT
2019-03-15 17:17:28 +01:00
parent fa80d34883
commit 88bddcff54
8 changed files with 280 additions and 38 deletions

View File

@@ -2,17 +2,36 @@ const torrentStream = require('torrent-stream');
const cheerio = require('cheerio');
const needle = require('needle');
const parseTorrent = require('parse-torrent');
const cloudscraper = require('cloudscraper');
const Tracker = require("peer-search/tracker");
const MAX_PEER_CONNECTIONS = process.env.MAX_PEER_CONNECTIONS || 20;
const EXTENSIONS = ["3g2", "3gp", "avi", "flv", "mkv", "mov", "mp2", "mp4", "mpe", "mpeg", "mpg", "mpv", "webm", "wmv"];
let TRACKERS = [];
module.exports.init = async function() {
TRACKERS = await getTrackerList();
};
module.exports.torrentFiles = function(torrent) {
return filesFromKat(torrent.infoHash)
return filesFromTorrentFile(torrent)
.catch(() => filesFromKat(torrent.infoHash))
.catch(() => filesFromTorrentStream(torrent))
.then((files) => files
.filter((file) => isVideo(file))
.map((file) => `${file.fileIndex}@@${file.path}`));
.then((files) => files.filter((file) => isVideo(file)));
};
module.exports.currentSeeders = function (torrent) {
if (!torrent.infoHash) {
return Promise.reject(new Error("no infoHash"));
}
return new Promise((resolve) =>
Promise.all(TRACKERS.map((tracker) => new Promise((resolve) => {
const t = new Tracker(tracker, { }, torrent.infoHash);
t.run();
t.on("info", (inf) => resolve([inf.seeders, inf.leechers]));
setTimeout(() => resolve([0, 0]), 1000);
}))
).then((results) => resolve(results.reduce((seeders, next) => seeders + next[0], 0)))
);
};
// async function filesFromBtSeeds(infoHash) {
@@ -24,10 +43,13 @@ module.exports.torrentFiles = function(torrent) {
// }
function filesFromKat(infoHash) {
if (!infoHash) {
return Promise.reject(new Error("no infoHash"));
}
const url = `http://kat.rip/torrent/${infoHash}.html`;
return needle('get', url, { open_timeout: 2000 })
.then((response) => {
if (!response.body) {
if (!response.body || response.statusCode !== 200) {
throw new Error('torrent not found in kat')
}
return response.body
@@ -43,13 +65,41 @@ function filesFromKat(infoHash) {
size: convertToBytes($(row).find('td[class=\'torFileSize\']').text())
});
});
if (!files[files.length - 1].size) {
throw new Error('not full file list')
}
return files;
})
}
async function filesFromTorrentFile(torrent) {
if (!torrent.torrentLink) {
return Promise.reject(new Error("no torrentLink"));
}
needle('get', torrent.torrentLink, { open_timeout: 2000 })
.then((response) => {
if (!response.body || response.statusCode !== 200) {
throw new Error('torrent not found')
}
return response.body
})
.then((body) => parseTorrent(body))
.then((info) => info.files.map((file, fileId) => ({
fileIndex: fileId,
name: file.name,
path: file.path.replace(/^[^\/]+\//, ''),
size: file.length
})));
}
async function filesFromTorrentStream(torrent) {
if (!torrent.infoHash && !torrent.magnetLink) {
return Promise.reject(new Error("no infoHash or magnetLink"));
}
return new Promise((resolve, rejected) => {
const engine = new torrentStream(torrent.infoHash, { connections: MAX_PEER_CONNECTIONS });
const engine = new torrentStream(torrent.magnetLink || torrent.infoHash, { connections: MAX_PEER_CONNECTIONS, trackers: TRACKERS });
engine.ready(() => {
const files = engine.files
@@ -74,7 +124,6 @@ function isVideo(title) {
return EXTENSIONS.includes(title.path.match(/\.(\w{2,4})$/)[1]);
}
function convertToBytes(sizeString) {
if (!sizeString) {
return;
@@ -88,22 +137,34 @@ function convertToBytes(sizeString) {
return Math.floor(parseFloat(sizeString) * multiplier);
}
function dynamicTimeout(torrent) {
if (torrent.seeders < 5) {
return 2000;
} else if (torrent.seeders < 10) {
return 3000;
} else if (torrent.seeders < 20) {
} else if (torrent.seeders < 10) {
return 4000;
} else if (torrent.seeders < 30) {
} else if (torrent.seeders < 20) {
return 5000;
} else if (torrent.seeders < 50) {
} else if (torrent.seeders < 30) {
return 7000;
} else if (torrent.seeders < 50) {
return 9000;
} else if (torrent.seeders < 100) {
return 10000;
return 12000;
} else {
return 15000;
}
}
function getTrackerList() {
return needle('get', 'https://torrents.me/tracker-list/', { open_timeout: 2000 })
.then((response) => {
if (!response.body || response.statusCode !== 200) {
throw new Error('tracker list not found')
}
return response.body
})
.then((body) => cheerio.load(body))
.then(($) => $('div[class="small-12 columns"] pre').text())
.then((text) => text.replace(/"/g, '').trim().split('\n'))
}