X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=server%2Fhelpers%2Fyoutube-dl.ts;h=3c80e7d41ccc10954ee6566e985a916a883ace30;hb=0bae66632a8767f3d5aca080af314bcab56e1889;hp=748c67e6c6b65b97123c38b1c3099b8d2010ffb2;hpb=cc68049424c2a7fc7fb919bc39a43f169e128780;p=github%2FChocobozzz%2FPeerTube.git diff --git a/server/helpers/youtube-dl.ts b/server/helpers/youtube-dl.ts index 748c67e6c..3c80e7d41 100644 --- a/server/helpers/youtube-dl.ts +++ b/server/helpers/youtube-dl.ts @@ -1,226 +1,392 @@ -import { truncate } from 'lodash' -import { CONSTRAINTS_FIELDS, VIDEO_CATEGORIES } from '../initializers' -import { logger } from './logger' -import { generateVideoTmpPath } from './utils' -import { join } from 'path' -import { root } from './core-utils' -import { ensureDir, writeFile } from 'fs-extra' -import * as request from 'request' import { createWriteStream } from 'fs' +import { ensureDir, move, pathExists, remove, writeFile } from 'fs-extra' +import got from 'got' +import { join } from 'path' +import { CONFIG } from '@server/initializers/config' +import { HttpStatusCode } from '../../shared/models/http/http-error-codes' +import { VideoResolution } from '../../shared/models/videos' +import { CONSTRAINTS_FIELDS, VIDEO_CATEGORIES, VIDEO_LANGUAGES, VIDEO_LICENCES } from '../initializers/constants' +import { peertubeTruncate, pipelinePromise, root } from './core-utils' +import { isVideoFileExtnameValid } from './custom-validators/videos' +import { logger } from './logger' +import { generateVideoImportTmpPath } from './utils' export type YoutubeDLInfo = { name?: string description?: string category?: number + language?: string licence?: number nsfw?: boolean tags?: string[] thumbnailUrl?: string + ext?: string + originallyPublishedAt?: Date } +export type YoutubeDLSubs = { + language: string + filename: string + path: string +}[] + const processOptions = { maxBuffer: 1024 * 1024 * 10 // 10MB } -function getYoutubeDLInfo (url: string, opts?: string[]): Promise { - return new Promise(async (res, rej) => { - const options = opts || [ '-j', '--flat-playlist' ] +class YoutubeDL { + + constructor (private readonly url: string = '', private readonly enabledResolutions: number[] = []) { + + } + + getYoutubeDLInfo (opts?: string[]): Promise { + return new Promise((res, rej) => { + let args = opts || [] + + if (CONFIG.IMPORT.VIDEOS.HTTP.FORCE_IPV4) { + args.push('--force-ipv4') + } + + args = this.wrapWithProxyOptions(args) + args = [ '-f', this.getYoutubeDLVideoFormat() ].concat(args) - const youtubeDL = await safeGetYoutubeDL() - youtubeDL.getInfo(url, options, (err, info) => { - if (err) return rej(err) - if (info.is_live === true) return rej(new Error('Cannot download a live streaming.')) + YoutubeDL.safeGetYoutubeDL() + .then(youtubeDL => { + youtubeDL.getInfo(this.url, args, processOptions, (err, info) => { + if (err) return rej(err) + if (info.is_live === true) return rej(new Error('Cannot download a live streaming.')) - const obj = buildVideoInfo(normalizeObject(info)) - if (obj.name && obj.name.length < CONSTRAINTS_FIELDS.VIDEOS.NAME.min) obj.name += ' video' + const obj = this.buildVideoInfo(this.normalizeObject(info)) + if (obj.name && obj.name.length < CONSTRAINTS_FIELDS.VIDEOS.NAME.min) obj.name += ' video' - return res(obj) + return res(obj) + }) + }) + .catch(err => rej(err)) }) - }) -} + } -function downloadYoutubeDLVideo (url: string) { - const path = generateVideoTmpPath(url) + getYoutubeDLSubs (opts?: object): Promise { + return new Promise((res, rej) => { + const cwd = CONFIG.STORAGE.TMP_DIR + const options = opts || { all: true, format: 'vtt', cwd } + + YoutubeDL.safeGetYoutubeDL() + .then(youtubeDL => { + youtubeDL.getSubs(this.url, options, (err, files) => { + if (err) return rej(err) + if (!files) return [] + + logger.debug('Get subtitles from youtube dl.', { url: this.url, files }) + + const subtitles = files.reduce((acc, filename) => { + const matched = filename.match(/\.([a-z]{2})(-[a-z]+)?\.(vtt|ttml)/i) + if (!matched || !matched[1]) return acc + + return [ + ...acc, + { + language: matched[1], + path: join(cwd, filename), + filename + } + ] + }, []) + + return res(subtitles) + }) + }) + .catch(err => rej(err)) + }) + } + + getYoutubeDLVideoFormat () { + /** + * list of format selectors in order or preference + * see https://github.com/ytdl-org/youtube-dl#format-selection + * + * case #1 asks for a mp4 using h264 (avc1) and the exact resolution in the hope + * of being able to do a "quick-transcode" + * case #2 is the first fallback. No "quick-transcode" means we can get anything else (like vp9) + * case #3 is the resolution-degraded equivalent of #1, and already a pretty safe fallback + * + * in any case we avoid AV1, see https://github.com/Chocobozzz/PeerTube/issues/3499 + **/ + const resolution = this.enabledResolutions.length === 0 + ? VideoResolution.H_720P + : Math.max(...this.enabledResolutions) + + return [ + `bestvideo[vcodec^=avc1][height=${resolution}]+bestaudio[ext=m4a]`, // case #1 + `bestvideo[vcodec!*=av01][vcodec!*=vp9.2][height=${resolution}]+bestaudio`, // case #2 + `bestvideo[vcodec^=avc1][height<=${resolution}]+bestaudio[ext=m4a]`, // case #3 + `bestvideo[vcodec!*=av01][vcodec!*=vp9.2]+bestaudio`, + 'best[vcodec!*=av01][vcodec!*=vp9.2]', // case fallback for known formats + 'best' // Ultimate fallback + ].join('/') + } - logger.info('Importing youtubeDL video %s', url) + downloadYoutubeDLVideo (fileExt: string, timeout: number) { + // Leave empty the extension, youtube-dl will add it + const pathWithoutExtension = generateVideoImportTmpPath(this.url, '') - const options = [ '-f', 'bestvideo[ext=mp4]+bestaudio[ext=m4a]/best', '-o', path ] + let timer - return new Promise(async (res, rej) => { - const youtubeDL = await safeGetYoutubeDL() - youtubeDL.exec(url, options, processOptions, err => { - if (err) return rej(err) + logger.info('Importing youtubeDL video %s to %s', this.url, pathWithoutExtension) - return res(path) + let options = [ '-f', this.getYoutubeDLVideoFormat(), '-o', pathWithoutExtension ] + options = this.wrapWithProxyOptions(options) + + if (process.env.FFMPEG_PATH) { + options = options.concat([ '--ffmpeg-location', process.env.FFMPEG_PATH ]) + } + + logger.debug('YoutubeDL options for %s.', this.url, { options }) + + return new Promise((res, rej) => { + YoutubeDL.safeGetYoutubeDL() + .then(youtubeDL => { + youtubeDL.exec(this.url, options, processOptions, async err => { + clearTimeout(timer) + + try { + // If youtube-dl did not guess an extension for our file, just use .mp4 as default + if (await pathExists(pathWithoutExtension)) { + await move(pathWithoutExtension, pathWithoutExtension + '.mp4') + } + + const path = await this.guessVideoPathWithExtension(pathWithoutExtension, fileExt) + + if (err) { + remove(path) + .catch(err => logger.error('Cannot delete path on YoutubeDL error.', { err })) + + return rej(err) + } + + return res(path) + } catch (err) { + return rej(err) + } + }) + + timer = setTimeout(() => { + const err = new Error('YoutubeDL download timeout.') + + this.guessVideoPathWithExtension(pathWithoutExtension, fileExt) + .then(path => remove(path)) + .finally(() => rej(err)) + .catch(err => { + logger.error('Cannot remove file in youtubeDL timeout.', { err }) + return rej(err) + }) + }, timeout) + }) + .catch(err => rej(err)) }) - }) -} + } -// Thanks: https://github.com/przemyslawpluta/node-youtube-dl/blob/master/lib/downloader.js -// We rewrote it to avoid sync calls -async function updateYoutubeDLBinary () { - logger.info('Updating youtubeDL binary.') + buildOriginallyPublishedAt (obj: any) { + let originallyPublishedAt: Date = null - const binDirectory = join(root(), 'node_modules', 'youtube-dl', 'bin') - const bin = join(binDirectory, 'youtube-dl') - const detailsPath = join(binDirectory, 'details') - const url = 'https://yt-dl.org/downloads/latest/youtube-dl' + const uploadDateMatcher = /^(\d{4})(\d{2})(\d{2})$/.exec(obj.upload_date) + if (uploadDateMatcher) { + originallyPublishedAt = new Date() + originallyPublishedAt.setHours(0, 0, 0, 0) - await ensureDir(binDirectory) + const year = parseInt(uploadDateMatcher[1], 10) + // Month starts from 0 + const month = parseInt(uploadDateMatcher[2], 10) - 1 + const day = parseInt(uploadDateMatcher[3], 10) - return new Promise(res => { - request.get(url, { followRedirect: false }, (err, result) => { - if (err) { - logger.error('Cannot update youtube-dl.', { err }) - return res() - } + originallyPublishedAt.setFullYear(year, month, day) + } - if (result.statusCode !== 302) { - logger.error('youtube-dl update error: did not get redirect for the latest version link. Status %d', result.statusCode) - return res() + return originallyPublishedAt + } + + private async guessVideoPathWithExtension (tmpPath: string, sourceExt: string) { + if (!isVideoFileExtnameValid(sourceExt)) { + throw new Error('Invalid video extension ' + sourceExt) + } + + const extensions = [ sourceExt, '.mp4', '.mkv', '.webm' ] + + for (const extension of extensions) { + const path = tmpPath + extension + + if (await pathExists(path)) return path + } + + throw new Error('Cannot guess path of ' + tmpPath) + } + + private normalizeObject (obj: any) { + const newObj: any = {} + + for (const key of Object.keys(obj)) { + // Deprecated key + if (key === 'resolution') continue + + const value = obj[key] + + if (typeof value === 'string') { + newObj[key] = value.normalize() + } else { + newObj[key] = value } + } - const url = result.headers.location - const downloadFile = request.get(url) - const newVersion = /yt-dl\.org\/downloads\/(\d{4}\.\d\d\.\d\d(\.\d)?)\/youtube-dl/.exec(url)[ 1 ] - - downloadFile.on('response', result => { - if (result.statusCode !== 200) { - logger.error('Cannot update youtube-dl: new version response is not 200, it\'s %d.', result.statusCode) - return res() - } - - downloadFile.pipe(createWriteStream(bin, { mode: 493 })) - }) - - downloadFile.on('error', err => { - logger.error('youtube-dl update error.', { err }) - return res() - }) - - downloadFile.on('end', () => { - const details = JSON.stringify({ version: newVersion, path: bin, exec: 'youtube-dl' }) - writeFile(detailsPath, details, { encoding: 'utf8' }, err => { - if (err) { - logger.error('youtube-dl update error: cannot write details.', { err }) - return res() - } - - logger.info('youtube-dl updated to version %s.', newVersion) - return res() - }) - }) + return newObj + } + + private buildVideoInfo (obj: any): YoutubeDLInfo { + return { + name: this.titleTruncation(obj.title), + description: this.descriptionTruncation(obj.description), + category: this.getCategory(obj.categories), + licence: this.getLicence(obj.license), + language: this.getLanguage(obj.language), + nsfw: this.isNSFW(obj), + tags: this.getTags(obj.tags), + thumbnailUrl: obj.thumbnail || undefined, + originallyPublishedAt: this.buildOriginallyPublishedAt(obj), + ext: obj.ext + } + } + + private titleTruncation (title: string) { + return peertubeTruncate(title, { + length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max, + separator: /,? +/, + omission: ' […]' }) - }) -} + } -async function safeGetYoutubeDL () { - let youtubeDL + private descriptionTruncation (description: string) { + if (!description || description.length < CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.min) return undefined - try { - youtubeDL = require('youtube-dl') - } catch (e) { - // Download binary - await updateYoutubeDLBinary() - youtubeDL = require('youtube-dl') + return peertubeTruncate(description, { + length: CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.max, + separator: /,? +/, + omission: ' […]' + }) } - return youtubeDL -} + private isNSFW (info: any) { + return info.age_limit && info.age_limit >= 16 + } -// --------------------------------------------------------------------------- + private getTags (tags: any) { + if (Array.isArray(tags) === false) return [] -export { - updateYoutubeDLBinary, - downloadYoutubeDLVideo, - getYoutubeDLInfo, - safeGetYoutubeDL -} + return tags + .filter(t => t.length < CONSTRAINTS_FIELDS.VIDEOS.TAG.max && t.length > CONSTRAINTS_FIELDS.VIDEOS.TAG.min) + .map(t => t.normalize()) + .slice(0, 5) + } -// --------------------------------------------------------------------------- + private getLicence (licence: string) { + if (!licence) return undefined + + if (licence.includes('Creative Commons Attribution')) return 1 + + for (const key of Object.keys(VIDEO_LICENCES)) { + const peertubeLicence = VIDEO_LICENCES[key] + if (peertubeLicence.toLowerCase() === licence.toLowerCase()) return parseInt(key, 10) + } + + return undefined + } -function normalizeObject (obj: any) { - const newObj: any = {} + private getCategory (categories: string[]) { + if (!categories) return undefined - for (const key of Object.keys(obj)) { - // Deprecated key - if (key === 'resolution') continue + const categoryString = categories[0] + if (!categoryString || typeof categoryString !== 'string') return undefined - const value = obj[key] + if (categoryString === 'News & Politics') return 11 - if (typeof value === 'string') { - newObj[key] = value.normalize() - } else { - newObj[key] = value + for (const key of Object.keys(VIDEO_CATEGORIES)) { + const category = VIDEO_CATEGORIES[key] + if (categoryString.toLowerCase() === category.toLowerCase()) return parseInt(key, 10) } + + return undefined } - return newObj -} + private getLanguage (language: string) { + return VIDEO_LANGUAGES[language] ? language : undefined + } + + private wrapWithProxyOptions (options: string[]) { + if (CONFIG.IMPORT.VIDEOS.HTTP.PROXY.ENABLED) { + logger.debug('Using proxy for YoutubeDL') + + return [ '--proxy', CONFIG.IMPORT.VIDEOS.HTTP.PROXY.URL ].concat(options) + } -function buildVideoInfo (obj: any) { - return { - name: titleTruncation(obj.title), - description: descriptionTruncation(obj.description), - category: getCategory(obj.categories), - licence: getLicence(obj.license), - nsfw: isNSFW(obj), - tags: getTags(obj.tags), - thumbnailUrl: obj.thumbnail || undefined + return options } -} -function titleTruncation (title: string) { - return truncate(title, { - 'length': CONSTRAINTS_FIELDS.VIDEOS.NAME.max, - 'separator': /,? +/, - 'omission': ' […]' - }) -} + // Thanks: https://github.com/przemyslawpluta/node-youtube-dl/blob/master/lib/downloader.js + // We rewrote it to avoid sync calls + static async updateYoutubeDLBinary () { + logger.info('Updating youtubeDL binary.') -function descriptionTruncation (description: string) { - if (!description || description.length < CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.min) return undefined + const binDirectory = join(root(), 'node_modules', 'youtube-dl', 'bin') + const bin = join(binDirectory, 'youtube-dl') + const detailsPath = join(binDirectory, 'details') + const url = process.env.YOUTUBE_DL_DOWNLOAD_HOST || 'https://yt-dl.org/downloads/latest/youtube-dl' - return truncate(description, { - 'length': CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.max, - 'separator': /,? +/, - 'omission': ' […]' - }) -} + await ensureDir(binDirectory) -function isNSFW (info: any) { - return info.age_limit && info.age_limit >= 16 -} + try { + const result = await got(url, { followRedirect: false }) -function getTags (tags: any) { - if (Array.isArray(tags) === false) return [] + if (result.statusCode !== HttpStatusCode.FOUND_302) { + logger.error('youtube-dl update error: did not get redirect for the latest version link. Status %d', result.statusCode) + return + } - return tags - .filter(t => t.length < CONSTRAINTS_FIELDS.VIDEOS.TAG.max && t.length > CONSTRAINTS_FIELDS.VIDEOS.TAG.min) - .map(t => t.normalize()) - .slice(0, 5) -} + const newUrl = result.headers.location + const newVersion = /yt-dl\.org\/downloads\/(\d{4}\.\d\d\.\d\d(\.\d)?)\/youtube-dl/.exec(newUrl)[1] -function getLicence (licence: string) { - if (!licence) return undefined + const downloadFileStream = got.stream(newUrl) + const writeStream = createWriteStream(bin, { mode: 493 }) - if (licence.indexOf('Creative Commons Attribution') !== -1) return 1 + await pipelinePromise( + downloadFileStream, + writeStream + ) - return undefined -} + const details = JSON.stringify({ version: newVersion, path: bin, exec: 'youtube-dl' }) + await writeFile(detailsPath, details, { encoding: 'utf8' }) -function getCategory (categories: string[]) { - if (!categories) return undefined + logger.info('youtube-dl updated to version %s.', newVersion) + } catch (err) { + logger.error('Cannot update youtube-dl.', { err }) + } + } - const categoryString = categories[0] - if (!categoryString || typeof categoryString !== 'string') return undefined + static async safeGetYoutubeDL () { + let youtubeDL - if (categoryString === 'News & Politics') return 11 + try { + youtubeDL = require('youtube-dl') + } catch (e) { + // Download binary + await this.updateYoutubeDLBinary() + youtubeDL = require('youtube-dl') + } - for (const key of Object.keys(VIDEO_CATEGORIES)) { - const category = VIDEO_CATEGORIES[key] - if (categoryString.toLowerCase() === category.toLowerCase()) return parseInt(key, 10) + return youtubeDL } +} - return undefined +// --------------------------------------------------------------------------- + +export { + YoutubeDL }