[scraper] adds update seeders cron

This commit is contained in:
TheBeastLT
2020-04-08 15:53:02 +02:00
parent d3cc8c6b57
commit ce0e330172
10 changed files with 118 additions and 52 deletions

View File

@@ -8,12 +8,14 @@ const METADATA_PREFIX = `${GLOBAL_KEY_PREFIX}|metadata`;
const RESOLVED_URL_KEY_PREFIX = `${GLOBAL_KEY_PREFIX}|moch`;
const PROXY_KEY_PREFIX = `${GLOBAL_KEY_PREFIX}|proxy`;
const USER_AGENT_KEY_PREFIX = `${GLOBAL_KEY_PREFIX}|agent`;
const TRACKERS_KEY_PREFIX = `${GLOBAL_KEY_PREFIX}|trackers`;
const GLOBAL_TTL = process.env.METADATA_TTL || 7 * 24 * 60 * 60; // 7 days
const MEMORY_TTL = process.env.METADATA_TTL || 2 * 60 * 60; // 2 hours
const RESOLVED_URL_TTL = 2 * 60; // 2 minutes
const PROXY_TTL = 60 * 60; // 60 minutes
const USER_AGENT_TTL = 2 * 24 * 60 * 60; // 2 days
const TRACKERS_TTL = 2 * 24 * 60 * 60; // 2 days
const MONGO_URI = process.env.MONGODB_URI;
@@ -75,12 +77,17 @@ function cacheUserAgent(id, method) {
return cacheWrap(memoryCache, `${USER_AGENT_KEY_PREFIX}:${id}`, method, { ttl: { USER_AGENT_TTL } });
}
function cacheTrackers(method) {
return cacheWrap(memoryCache, `${TRACKERS_KEY_PREFIX}`, method, { ttl: { TRACKERS_TTL } });
}
module.exports = {
cacheWrapImdbId,
cacheWrapKitsuId,
cacheWrapMetadata,
cacheWrapResolvedUrl,
cacheWrapProxy,
cacheUserAgent
cacheUserAgent,
cacheTrackers
};

View File

@@ -1,3 +1,4 @@
const moment = require('moment');
const { Sequelize, fn, col, literal } = require('sequelize');
const Op = Sequelize.Op;
@@ -119,11 +120,18 @@ function getTorrentsBasedOnTitle(titleQuery, type) {
function getTorrentsWithoutSize() {
return Torrent.findAll({
where: literal(
'exists (select 1 from files where files."infoHash" = torrent."infoHash" and files.size = 300000000) and random() < 0.01'),
'exists (select 1 from files where files."infoHash" = torrent."infoHash" and files.size = 300000000)'),
order: [
['seeders', 'DESC']
],
limit: 1000
]
});
}
function getUpdateSeedersTorrents() {
const until = moment().subtract(7, 'days').format('YYYY-MM-DD');
return Torrent.findAll({
where: literal(`torrent."updatedAt" < \'${until}\' and random() < 0.001`),
limit: 100
});
}
@@ -186,6 +194,7 @@ module.exports = {
createTorrent,
getTorrent,
getTorrentsBasedOnTitle,
getUpdateSeedersTorrents,
createFile,
getFiles,
getFilesBasedOnTitle,

View File

@@ -4,20 +4,22 @@ const parseTorrent = require('parse-torrent');
const async = require('async');
const decode = require('magnet-uri');
const isVideo = require('./video');
const { cacheTrackers } = require('./cache');
const TRACKERS_URL = 'https://ngosang.github.io/trackerslist/trackers_best.txt';
const MAX_PEER_CONNECTIONS = process.env.MAX_PEER_CONNECTIONS || 20;
const SEEDS_CHECK_TIMEOUT = process.env.SEEDS_CHECK_TIMEOUT || 10 * 1000; // 10 secs
module.exports.updateCurrentSeeders = function (torrent) {
return new Promise((resolve) => {
if (!torrent.magnetLink) {
return new Promise(async (resolve) => {
if (!torrent.magnetLink && !torrent.infoHash) {
return resolve(0);
}
const Tracker = require("peer-search/tracker");
const seeders = {};
const decodedMagnetLink = decode(torrent.magnetLink);
const trackers = decodedMagnetLink && decodedMagnetLink.tr;
const decodedMagnetLink = torrent.magnetLink && decode(torrent.magnetLink);
const trackers = decodedMagnetLink && decodedMagnetLink.tr || torrent.trackers || await getDefaultTrackers();
const callback = () => resolve(Math.max(...Object.values(seeders).map(values => values[0]).concat(0)));
setTimeout(callback, SEEDS_CHECK_TIMEOUT);
@@ -30,12 +32,19 @@ module.exports.updateCurrentSeeders = function (torrent) {
ready();
});
}, callback);
}).then(seeders => ({ ...torrent, seeders: torrent.seeders || seeders }));
}).then(seeders => {
torrent.seeders = seeders;
return torrent;
});
};
module.exports.updateTorrentSize = function (torrent) {
return filesAndSizeFromTorrentStream(torrent, SEEDS_CHECK_TIMEOUT)
.then(result => ({ ...torrent, size: result.size, files: result.files }));
.then(result => {
torrent.size = result.size;
torrent.files = result.files;
return torrent;
});
};
module.exports.sizeAndFiles = torrent => filesAndSizeFromTorrentStream(torrent, 30000);
@@ -124,19 +133,9 @@ function filterExtras(files) {
return files.filter(file => !isExtra(file));
}
function dynamicTimeout(torrent) {
if (torrent.seeders < 5) {
return 5000;
} else if (torrent.seeders < 10) {
return 7000;
} else if (torrent.seeders < 20) {
return 10000;
} else if (torrent.seeders < 30) {
return 15000;
} else if (torrent.seeders < 50) {
return 20000;
} else {
return 30000;
}
async function getDefaultTrackers() {
return cacheTrackers(() => needle('get', TRACKERS_URL, { open_timeout: SEEDS_CHECK_TIMEOUT })
.then(response => response.body && response.body.trim())
.then(body => body && body.split('\n\n') || []));
}

View File

@@ -63,6 +63,8 @@ async function updateTorrentSeeders(torrent) {
.catch(() => undefined)
.then(stored => {
if (stored) {
console.log(
`Updated seeder for ${torrent.provider} [${torrent.infoHash}] ${torrent.title} from ${stored.seeders} to ${torrent.seeders}`);
stored.seeders = torrent.seeders;
stored.changed('updatedAt', true);
return stored.save()