server/helpers/youtube-dl.ts

   1 import { CONSTRAINTS_FIELDS, VIDEO_CATEGORIES, VIDEO_LANGUAGES, VIDEO_LICENCES } from '../initializers/constants'
   2 import { logger } from './logger'
   3 import { generateVideoImportTmpPath } from './utils'
   4 import { getEnabledResolutions } from '../lib/video-transcoding'
   5 import { join } from 'path'
   6 import { peertubeTruncate, root } from './core-utils'
   7 import { ensureDir, remove, writeFile } from 'fs-extra'
   8 import * as request from 'request'
   9 import { createWriteStream } from 'fs'
  10 import { CONFIG } from '@server/initializers/config'
  11 import { HttpStatusCode } from '../../shared/core-utils/miscs/http-error-codes'
  12 import { VideoResolution } from '../../shared/models/videos'
  13
  14 export type YoutubeDLInfo = {
  15   name?: string
  16   description?: string
  17   category?: number
  18   language?: string
  19   licence?: number
  20   nsfw?: boolean
  21   tags?: string[]
  22   thumbnailUrl?: string
  23   ext?: string
  24   mergeExt?: string
  25   originallyPublishedAt?: Date
  26 }
  27
  28 export type YoutubeDLSubs = {
  29   language: string
  30   filename: string
  31   path: string
  32 }[]
  33
  34 const processOptions = {
  35   maxBuffer: 1024 * 1024 * 10 // 10MB
  36 }
  37
  38 function getYoutubeDLInfo (url: string, opts?: string[]): Promise<YoutubeDLInfo> {
  39   return new Promise<YoutubeDLInfo>((res, rej) => {
  40     let args = opts || [ '-j', '--flat-playlist' ]
  41
  42     if (CONFIG.IMPORT.VIDEOS.HTTP.FORCE_IPV4) {
  43       args.push('--force-ipv4')
  44     }
  45
  46     args = wrapWithProxyOptions(args)
  47     args = [ '-f', getYoutubeDLVideoFormat() ].concat(args)
  48
  49     safeGetYoutubeDL()
  50       .then(youtubeDL => {
  51         youtubeDL.getInfo(url, args, processOptions, (err, info) => {
  52           if (err) return rej(err)
  53           if (info.is_live === true) return rej(new Error('Cannot download a live streaming.'))
  54           if (info.format_id?.includes('+')) {
  55             // this is a merge format and its extension will be appended
  56             if (info.ext === 'mp4') {
  57               info.mergeExt = 'mp4'
  58             } else {
  59               info.mergeExt = 'mkv'
  60             }
  61           }
  62
  63           const obj = buildVideoInfo(normalizeObject(info))
  64           if (obj.name && obj.name.length < CONSTRAINTS_FIELDS.VIDEOS.NAME.min) obj.name += ' video'
  65
  66           return res(obj)
  67         })
  68       })
  69       .catch(err => rej(err))
  70   })
  71 }
  72
  73 function getYoutubeDLSubs (url: string, opts?: object): Promise<YoutubeDLSubs> {
  74   return new Promise<YoutubeDLSubs>((res, rej) => {
  75     const cwd = CONFIG.STORAGE.TMP_DIR
  76     const options = opts || { all: true, format: 'vtt', cwd }
  77
  78     safeGetYoutubeDL()
  79       .then(youtubeDL => {
  80         youtubeDL.getSubs(url, options, (err, files) => {
  81           if (err) return rej(err)
  82           if (!files) return []
  83
  84           logger.debug('Get subtitles from youtube dl.', { url, files })
  85
  86           const subtitles = files.reduce((acc, filename) => {
  87             const matched = filename.match(/\.([a-z]{2})\.(vtt|ttml)/i)
  88             if (!matched || !matched[1]) return acc
  89
  90             return [
  91               ...acc,
  92               {
  93                 language: matched[1],
  94                 path: join(cwd, filename),
  95                 filename
  96               }
  97             ]
  98           }, [])
  99
 100           return res(subtitles)
 101         })
 102       })
 103       .catch(err => rej(err))
 104   })
 105 }
 106
 107 function getYoutubeDLVideoFormat () {
 108   /**
 109    * list of format selectors in order or preference
 110    * see https://github.com/ytdl-org/youtube-dl#format-selection
 111    *
 112    * case #1 asks for a mp4 using h264 (avc1) and the exact resolution in the hope
 113    * of being able to do a "quick-transcode"
 114    * case #2 is the first fallback. No "quick-transcode" means we can get anything else (like vp9)
 115    * case #3 is the resolution-degraded equivalent of #1, and already a pretty safe fallback
 116    *
 117    * in any case we avoid AV1, see https://github.com/Chocobozzz/PeerTube/issues/3499
 118    **/
 119   const enabledResolutions = getEnabledResolutions('vod')
 120   const resolution = enabledResolutions.length === 0
 121     ? VideoResolution.H_720P
 122     : Math.max(...enabledResolutions)
 123
 124   return [
 125     `bestvideo[vcodec^=avc1][height=${resolution}]+bestaudio[ext=m4a]`, // case #1
 126     `bestvideo[vcodec!*=av01][vcodec!*=vp9.2][height=${resolution}]+bestaudio`, // case #2
 127     `bestvideo[vcodec^=avc1][height<=${resolution}]+bestaudio[ext=m4a]`, // case #3
 128     `bestvideo[vcodec!*=av01][vcodec!*=vp9.2]+bestaudio`,
 129     'best[vcodec!*=av01][vcodec!*=vp9.2]', // case fallback for known formats
 130     'best' // Ultimate fallback
 131   ].join('/')
 132 }
 133
 134 function downloadYoutubeDLVideo (url: string, extension: string, timeout: number, mergeExtension?: string) {
 135   const path = generateVideoImportTmpPath(url, extension)
 136   const finalPath = mergeExtension ? path.replace(new RegExp(`${extension}$`), mergeExtension) : path
 137   let timer
 138
 139   logger.info('Importing youtubeDL video %s to %s', url, finalPath)
 140
 141   let options = [ '-f', getYoutubeDLVideoFormat(), '-o', path ]
 142   options = wrapWithProxyOptions(options)
 143
 144   if (process.env.FFMPEG_PATH) {
 145     options = options.concat([ '--ffmpeg-location', process.env.FFMPEG_PATH ])
 146   }
 147
 148   return new Promise<string>((res, rej) => {
 149     safeGetYoutubeDL()
 150       .then(youtubeDL => {
 151         youtubeDL.exec(url, options, processOptions, err => {
 152           clearTimeout(timer)
 153
 154           if (err) {
 155             remove(path)
 156               .catch(err => logger.error('Cannot delete path on YoutubeDL error.', { err }))
 157
 158             return rej(err)
 159           }
 160
 161           return res(finalPath)
 162         })
 163
 164         timer = setTimeout(() => {
 165           const err = new Error('YoutubeDL download timeout.')
 166
 167           remove(path)
 168             .finally(() => rej(err))
 169             .catch(err => {
 170               logger.error('Cannot remove %s in youtubeDL timeout.', path, { err })
 171               return rej(err)
 172             })
 173         }, timeout)
 174       })
 175       .catch(err => rej(err))
 176   })
 177 }
 178
 179 // Thanks: https://github.com/przemyslawpluta/node-youtube-dl/blob/master/lib/downloader.js
 180 // We rewrote it to avoid sync calls
 181 async function updateYoutubeDLBinary () {
 182   logger.info('Updating youtubeDL binary.')
 183
 184   const binDirectory = join(root(), 'node_modules', 'youtube-dl', 'bin')
 185   const bin = join(binDirectory, 'youtube-dl')
 186   const detailsPath = join(binDirectory, 'details')
 187   const url = process.env.YOUTUBE_DL_DOWNLOAD_HOST || 'https://yt-dl.org/downloads/latest/youtube-dl'
 188
 189   await ensureDir(binDirectory)
 190
 191   return new Promise(res => {
 192     request.get(url, { followRedirect: false }, (err, result) => {
 193       if (err) {
 194         logger.error('Cannot update youtube-dl.', { err })
 195         return res()
 196       }
 197
 198       if (result.statusCode !== HttpStatusCode.FOUND_302) {
 199         logger.error('youtube-dl update error: did not get redirect for the latest version link. Status %d', result.statusCode)
 200         return res()
 201       }
 202
 203       const url = result.headers.location
 204       const downloadFile = request.get(url)
 205       const newVersion = /yt-dl\.org\/downloads\/(\d{4}\.\d\d\.\d\d(\.\d)?)\/youtube-dl/.exec(url)[1]
 206
 207       downloadFile.on('response', result => {
 208         if (result.statusCode !== HttpStatusCode.OK_200) {
 209           logger.error('Cannot update youtube-dl: new version response is not 200, it\'s %d.', result.statusCode)
 210           return res()
 211         }
 212
 213         const writeStream = createWriteStream(bin, { mode: 493 }).on('error', err => {
 214           logger.error('youtube-dl update error in write stream', { err })
 215           return res()
 216         })
 217
 218         downloadFile.pipe(writeStream)
 219       })
 220
 221       downloadFile.on('error', err => {
 222         logger.error('youtube-dl update error.', { err })
 223         return res()
 224       })
 225
 226       downloadFile.on('end', () => {
 227         const details = JSON.stringify({ version: newVersion, path: bin, exec: 'youtube-dl' })
 228         writeFile(detailsPath, details, { encoding: 'utf8' }, err => {
 229           if (err) {
 230             logger.error('youtube-dl update error: cannot write details.', { err })
 231             return res()
 232           }
 233
 234           logger.info('youtube-dl updated to version %s.', newVersion)
 235           return res()
 236         })
 237       })
 238     })
 239   })
 240 }
 241
 242 async function safeGetYoutubeDL () {
 243   let youtubeDL
 244
 245   try {
 246     youtubeDL = require('youtube-dl')
 247   } catch (e) {
 248     // Download binary
 249     await updateYoutubeDLBinary()
 250     youtubeDL = require('youtube-dl')
 251   }
 252
 253   return youtubeDL
 254 }
 255
 256 function buildOriginallyPublishedAt (obj: any) {
 257   let originallyPublishedAt: Date = null
 258
 259   const uploadDateMatcher = /^(\d{4})(\d{2})(\d{2})$/.exec(obj.upload_date)
 260   if (uploadDateMatcher) {
 261     originallyPublishedAt = new Date()
 262     originallyPublishedAt.setHours(0, 0, 0, 0)
 263
 264     const year = parseInt(uploadDateMatcher[1], 10)
 265     // Month starts from 0
 266     const month = parseInt(uploadDateMatcher[2], 10) - 1
 267     const day = parseInt(uploadDateMatcher[3], 10)
 268
 269     originallyPublishedAt.setFullYear(year, month, day)
 270   }
 271
 272   return originallyPublishedAt
 273 }
 274
 275 // ---------------------------------------------------------------------------
 276
 277 export {
 278   updateYoutubeDLBinary,
 279   getYoutubeDLVideoFormat,
 280   downloadYoutubeDLVideo,
 281   getYoutubeDLSubs,
 282   getYoutubeDLInfo,
 283   safeGetYoutubeDL,
 284   buildOriginallyPublishedAt
 285 }
 286
 287 // ---------------------------------------------------------------------------
 288
 289 function normalizeObject (obj: any) {
 290   const newObj: any = {}
 291
 292   for (const key of Object.keys(obj)) {
 293     // Deprecated key
 294     if (key === 'resolution') continue
 295
 296     const value = obj[key]
 297
 298     if (typeof value === 'string') {
 299       newObj[key] = value.normalize()
 300     } else {
 301       newObj[key] = value
 302     }
 303   }
 304
 305   return newObj
 306 }
 307
 308 function buildVideoInfo (obj: any): YoutubeDLInfo {
 309   return {
 310     name: titleTruncation(obj.title),
 311     description: descriptionTruncation(obj.description),
 312     category: getCategory(obj.categories),
 313     licence: getLicence(obj.license),
 314     language: getLanguage(obj.language),
 315     nsfw: isNSFW(obj),
 316     tags: getTags(obj.tags),
 317     thumbnailUrl: obj.thumbnail || undefined,
 318     originallyPublishedAt: buildOriginallyPublishedAt(obj),
 319     ext: obj.ext,
 320     mergeExt: obj.mergeExt
 321   }
 322 }
 323
 324 function titleTruncation (title: string) {
 325   return peertubeTruncate(title, {
 326     length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
 327     separator: /,? +/,
 328     omission: ' […]'
 329   })
 330 }
 331
 332 function descriptionTruncation (description: string) {
 333   if (!description || description.length < CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.min) return undefined
 334
 335   return peertubeTruncate(description, {
 336     length: CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.max,
 337     separator: /,? +/,
 338     omission: ' […]'
 339   })
 340 }
 341
 342 function isNSFW (info: any) {
 343   return info.age_limit && info.age_limit >= 16
 344 }
 345
 346 function getTags (tags: any) {
 347   if (Array.isArray(tags) === false) return []
 348
 349   return tags
 350     .filter(t => t.length < CONSTRAINTS_FIELDS.VIDEOS.TAG.max && t.length > CONSTRAINTS_FIELDS.VIDEOS.TAG.min)
 351     .map(t => t.normalize())
 352     .slice(0, 5)
 353 }
 354
 355 function getLicence (licence: string) {
 356   if (!licence) return undefined
 357
 358   if (licence.includes('Creative Commons Attribution')) return 1
 359
 360   for (const key of Object.keys(VIDEO_LICENCES)) {
 361     const peertubeLicence = VIDEO_LICENCES[key]
 362     if (peertubeLicence.toLowerCase() === licence.toLowerCase()) return parseInt(key, 10)
 363   }
 364
 365   return undefined
 366 }
 367
 368 function getCategory (categories: string[]) {
 369   if (!categories) return undefined
 370
 371   const categoryString = categories[0]
 372   if (!categoryString || typeof categoryString !== 'string') return undefined
 373
 374   if (categoryString === 'News & Politics') return 11
 375
 376   for (const key of Object.keys(VIDEO_CATEGORIES)) {
 377     const category = VIDEO_CATEGORIES[key]
 378     if (categoryString.toLowerCase() === category.toLowerCase()) return parseInt(key, 10)
 379   }
 380
 381   return undefined
 382 }
 383
 384 function getLanguage (language: string) {
 385   return VIDEO_LANGUAGES[language] ? language : undefined
 386 }
 387
 388 function wrapWithProxyOptions (options: string[]) {
 389   if (CONFIG.IMPORT.VIDEOS.HTTP.PROXY.ENABLED) {
 390     logger.debug('Using proxy for YoutubeDL')
 391
 392     return [ '--proxy', CONFIG.IMPORT.VIDEOS.HTTP.PROXY.URL ].concat(options)
 393   }
 394
 395   return options
 396 }