const urllib = require('url'); const querystring = require('querystring'); const sax = require('sax'); const request = require('miniget'); const util = require('./util'); const extras = require('./info-extras'); const sig = require('./sig'); const FORMATS = require('./formats'); const VIDEO_URL = 'https://www.youtube.com/watch?v='; const EMBED_URL = 'https://www.youtube.com/embed/'; const VIDEO_EURL = 'https://youtube.googleapis.com/v/'; const INFO_HOST = 'www.youtube.com'; const INFO_PATH = '/get_video_info'; const KEYS_TO_SPLIT = [ 'keywords', 'fmt_list', 'fexp', 'watermark' ]; /** * Gets info from a video without getting additional formats. * * @param {String} id * @param {Object} options * @param {Function(Error, Object)} callback */ exports.getBasicInfo = (id, options, callback) => { // Try getting config from the video page first. const params = 'hl=' + (options.lang || 'en'); let url = VIDEO_URL + id + '&' + params + '&bpctr=' + Math.ceil(Date.now() / 1000); // Remove header from watch page request. // Otherwise, it'll use a different framework for rendering content. const reqOptions = Object.assign({}, options.requestOptions); reqOptions.headers = Object.assign({}, reqOptions.headers, { 'User-Agent': '' }); request(url, reqOptions, (err, res, body) => { if (err) return callback(err); // Check if there are any errors with this video page. const unavailableMsg = util.between(body, '
'); if (unavailableMsg && !/\bhid\b/.test(util.between(unavailableMsg, 'class="', '"'))) { // Ignore error about age restriction. if (!body.includes('
', '').trim())); } } // Parse out additional metadata from this page. const additional = { // Get the author/uploader. author: extras.getAuthor(body), // Get the day the vid was published. published: extras.getPublished(body), // Get description. description: extras.getVideoDescription(body), // Get media info. media: extras.getVideoMedia(body), // Get related videos. related_videos: extras.getRelatedVideos(body), // Give the standard link to the video. video_url: VIDEO_URL + id, }; const jsonStr = util.between(body, 'ytplayer.config = ', ''); let config; if (jsonStr) { config = jsonStr.slice(0, jsonStr.lastIndexOf(';ytplayer.load')); gotConfig(id, options, additional, config, false, callback); } else { // If the video page doesn't work, maybe because it has mature content. // and requires an account logged in to view, try the embed page. url = EMBED_URL + id + '?' + params; request(url, options.requestOptions, (err, res, body) => { if (err) return callback(err); config = util.between(body, 't.setConfig({\'PLAYER_CONFIG\': ', '},\''); gotConfig(id, options, additional, config, true, callback); }); } }); }; /** * @param {Object} id * @param {Object} options * @param {Object} additional * @param {Object} config * @param {Boolean} fromEmbed * @param {Function(Error, Object)} callback */ function gotConfig(id, options, additional, config, fromEmbed, callback) { if (!config) { return callback(new Error('Could not find player config')); } try { config = JSON.parse(config + (fromEmbed ? '}' : '')); } catch (err) { return callback(new Error('Error parsing config: ' + err.message)); } const url = urllib.format({ protocol: 'https', host: INFO_HOST, pathname: INFO_PATH, query: { video_id: id, eurl: VIDEO_EURL + id, ps: 'default', gl: 'US', hl: (options.lang || 'en'), sts: config.sts, }, }); request(url, options.requestOptions, (err, res, body) => { if (err) return callback(err); let info = querystring.parse(body); if (info.status === 'fail') { if (info.errorcode === '150' && config.args) { info = config.args; info.no_embed_allowed = true; } else { return callback( new Error(`Code ${info.errorcode}: ${util.stripHTML(info.reason)}`)); } } const player_response = config.args.player_response || info.player_response; if (player_response) { try { info.player_response = JSON.parse(player_response); } catch (err) { return callback( new Error('Error parsing `player_response`: ' + err.message)); } let playability = info.player_response.playabilityStatus; if (playability && playability.status === 'UNPLAYABLE') { return callback(new Error(playability.reason)); } } // Split some keys by commas. KEYS_TO_SPLIT.forEach((key) => { if (!info[key]) return; info[key] = info[key] .split(',') .filter((v) => v !== ''); }); info.fmt_list = info.fmt_list ? info.fmt_list.map((format) => format.split('/')) : []; info.formats = util.parseFormats(info); // Add additional properties to info. Object.assign(info, additional); info.age_restricted = fromEmbed; info.html5player = config.assets.js; if (config.args.dashmpd && info.dashmpd !== config.args.dashmpd) { info.dashmpd2 = config.args.dashmpd; } callback(null, info); }); } /** * Gets info from a video additional formats and deciphered URLs. * * @param {String} id * @param {Object} options * @param {Function(Error, Object)} callback */ exports.getFullInfo = (id, options, callback) => { return exports.getBasicInfo(id, options, (err, info) => { if (err) return callback(err); if (info.formats.length || info.dashmpd || info.dashmpd2 || info.hlsvp) { const html5playerfile = urllib.resolve(VIDEO_URL, info.html5player); sig.getTokens(html5playerfile, options, (err, tokens) => { if (err) return callback(err); sig.decipherFormats(info.formats, tokens, options.debug); let funcs = []; if (info.dashmpd) { info.dashmpd = decipherURL(info.dashmpd, tokens); funcs.push(getDashManifest.bind(null, info.dashmpd, options)); } if (info.dashmpd2) { info.dashmpd2 = decipherURL(info.dashmpd2, tokens); funcs.push(getDashManifest.bind(null, info.dashmpd2, options)); } if (info.hlsvp) { info.hlsvp = decipherURL(info.hlsvp, tokens); funcs.push(getM3U8.bind(null, info.hlsvp, options)); } util.parallel(funcs, (err, results) => { if (err) return callback(err); if (results[0]) { mergeFormats(info, results[0]); } if (results[1]) { mergeFormats(info, results[1]); } if (results[2]) { mergeFormats(info, results[2]); } if (!info.formats.length) { callback(new Error('No formats found')); return; } if (options.debug) { info.formats.forEach((format) => { const itag = format.itag; if (!FORMATS[itag]) { console.warn(`No format metadata for itag ${itag} found`); } }); } info.formats.forEach(util.addFormatMeta); info.formats.sort(util.sortFormats); info.full = true; callback(null, info); }); }); } else { callback(new Error('This video is unavailable')); } }); }; /** * @param {String} url * @param {Array.} tokens */ function decipherURL(url, tokens) { return url.replace(/\/s\/([a-fA-F0-9.]+)/, (_, s) => { return '/signature/' + sig.decipher(tokens, s); }); } /** * Merges formats from DASH or M3U8 with formats from video info page. * * @param {Object} info * @param {Object} formatsMap */ function mergeFormats(info, formatsMap) { info.formats.forEach((f) => { if (!formatsMap[f.itag]) { formatsMap[f.itag] = f; } }); info.formats = []; for (let itag in formatsMap) { info.formats.push(formatsMap[itag]); } } /** * Gets additional DASH formats. * * @param {String} url * @param {Object} options * @param {Function(!Error, Array.)} callback */ function getDashManifest(url, options, callback) { let formats = {}; const parser = sax.parser(false); parser.onerror = callback; parser.onopentag = (node) => { if (node.name === 'REPRESENTATION') { const itag = node.attributes.ID; formats[itag] = { itag, url }; } }; parser.onend = () => { callback(null, formats); }; const req = request(urllib.resolve(VIDEO_URL, url), options.requestOptions); req.setEncoding('utf8'); req.on('error', callback); req.on('data', (chunk) => { parser.write(chunk); }); req.on('end', parser.close.bind(parser)); } /** * Gets additional formats. * * @param {String} url * @param {Object} options * @param {Function(!Error, Array.)} callback */ function getM3U8(url, options, callback) { url = urllib.resolve(VIDEO_URL, url); request(url, options.requestOptions, (err, res, body) => { if (err) return callback(err); let formats = {}; body .split('\n') .filter((line) => /https?:\/\//.test(line)) .forEach((line) => { const itag = line.match(/\/itag\/(\d+)\//)[1]; formats[itag] = { itag: itag, url: line }; }); callback(null, formats); }); } // Cached for getting basic/full info. exports.cache = new Map(); exports.cache.timeout = 1000; // Cache get info functions. // In case a user wants to get a video's info before downloading. for (let fnName of ['getBasicInfo', 'getFullInfo']) { /** * @param {String} link * @param {Object} options * @param {Function(Error, Object)} callback */ const fn = exports[fnName]; exports[fnName] = (link, options, callback) => { if (typeof options === 'function') { callback = options; options = {}; } else if (!options) { options = {}; } if (!callback) { return new Promise((resolve, reject) => { exports[fnName](link, options, (err, info) => { if (err) return reject(err); resolve(info); }); }); } const id = util.getVideoID(link); if (id instanceof Error) return callback(id); const key = [fnName, id, options.lang].join('-'); if (exports.cache.has(key)) { callback(null, exports.cache.get(key)); } else { fn(id, options, (err, info) => { if (err) return callback(err); exports.cache.set(key, info); setTimeout(() => { exports.cache.delete(key); }, exports.cache.timeout); callback(null, info); }); } }; }