server/helpers/youtube-dl.ts

   1 import { createWriteStream } from 'fs'
   2 import { ensureDir, pathExists, remove, writeFile } from 'fs-extra'
   3 import { join } from 'path'
   4 import * as request from 'request'
   5 import { CONFIG } from '@server/initializers/config'
   6 import { HttpStatusCode } from '../../shared/core-utils/miscs/http-error-codes'
   7 import { VideoResolution } from '../../shared/models/videos'
   8 import { CONSTRAINTS_FIELDS, VIDEO_CATEGORIES, VIDEO_LANGUAGES, VIDEO_LICENCES } from '../initializers/constants'
   9 import { getEnabledResolutions } from '../lib/video-transcoding'
  10 import { peertubeTruncate, root } from './core-utils'
  11 import { isVideoFileExtnameValid } from './custom-validators/videos'
  12 import { logger } from './logger'
  13 import { generateVideoImportTmpPath } from './utils'
  14
  15 export type YoutubeDLInfo = {
  16   name?: string
  17   description?: string
  18   category?: number
  19   language?: string
  20   licence?: number
  21   nsfw?: boolean
  22   tags?: string[]
  23   thumbnailUrl?: string
  24   ext?: 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
  55           const obj = buildVideoInfo(normalizeObject(info))
  56           if (obj.name && obj.name.length < CONSTRAINTS_FIELDS.VIDEOS.NAME.min) obj.name += ' video'
  57
  58           return res(obj)
  59         })
  60       })
  61       .catch(err => rej(err))
  62   })
  63 }
  64
  65 function getYoutubeDLSubs (url: string, opts?: object): Promise<YoutubeDLSubs> {
  66   return new Promise<YoutubeDLSubs>((res, rej) => {
  67     const cwd = CONFIG.STORAGE.TMP_DIR
  68     const options = opts || { all: true, format: 'vtt', cwd }
  69
  70     safeGetYoutubeDL()
  71       .then(youtubeDL => {
  72         youtubeDL.getSubs(url, options, (err, files) => {
  73           if (err) return rej(err)
  74           if (!files) return []
  75
  76           logger.debug('Get subtitles from youtube dl.', { url, files })
  77
  78           const subtitles = files.reduce((acc, filename) => {
  79             const matched = filename.match(/\.([a-z]{2})\.(vtt|ttml)/i)
  80             if (!matched || !matched[1]) return acc
  81
  82             return [
  83               ...acc,
  84               {
  85                 language: matched[1],
  86                 path: join(cwd, filename),
  87                 filename
  88               }
  89             ]
  90           }, [])
  91
  92           return res(subtitles)
  93         })
  94       })
  95       .catch(err => rej(err))
  96   })
  97 }
  98
  99 function getYoutubeDLVideoFormat () {
 100   /**
 101    * list of format selectors in order or preference
 102    * see https://github.com/ytdl-org/youtube-dl#format-selection
 103    *
 104    * case #1 asks for a mp4 using h264 (avc1) and the exact resolution in the hope
 105    * of being able to do a "quick-transcode"
 106    * case #2 is the first fallback. No "quick-transcode" means we can get anything else (like vp9)
 107    * case #3 is the resolution-degraded equivalent of #1, and already a pretty safe fallback
 108    *
 109    * in any case we avoid AV1, see https://github.com/Chocobozzz/PeerTube/issues/3499
 110    **/
 111   const enabledResolutions = getEnabledResolutions('vod')
 112   const resolution = enabledResolutions.length === 0
 113     ? VideoResolution.H_720P
 114     : Math.max(...enabledResolutions)
 115
 116   return [
 117     `bestvideo[vcodec^=avc1][height=${resolution}]+bestaudio[ext=m4a]`, // case #1
 118     `bestvideo[vcodec!*=av01][vcodec!*=vp9.2][height=${resolution}]+bestaudio`, // case #2
 119     `bestvideo[vcodec^=avc1][height<=${resolution}]+bestaudio[ext=m4a]`, // case #3
 120     `bestvideo[vcodec!*=av01][vcodec!*=vp9.2]+bestaudio`,
 121     'best[vcodec!*=av01][vcodec!*=vp9.2]', // case fallback for known formats
 122     'best' // Ultimate fallback
 123   ].join('/')
 124 }
 125
 126 function downloadYoutubeDLVideo (url: string, fileExt: string, timeout: number) {
 127   // Leave empty the extension, youtube-dl will add it
 128   const pathWithoutExtension = generateVideoImportTmpPath(url, '')
 129
 130   let timer
 131
 132   logger.info('Importing youtubeDL video %s to %s', url, pathWithoutExtension)
 133
 134   let options = [ '-f', getYoutubeDLVideoFormat(), '-o', pathWithoutExtension ]
 135   options = wrapWithProxyOptions(options)
 136
 137   if (process.env.FFMPEG_PATH) {
 138     options = options.concat([ '--ffmpeg-location', process.env.FFMPEG_PATH ])
 139   }
 140
 141   logger.debug('YoutubeDL options for %s.', url, { options })
 142
 143   return new Promise<string>((res, rej) => {
 144     safeGetYoutubeDL()
 145       .then(youtubeDL => {
 146         youtubeDL.exec(url, options, processOptions, async err => {
 147           clearTimeout(timer)
 148
 149           try {
 150             const path = await guessVideoPathWithExtension(pathWithoutExtension, fileExt)
 151
 152             if (err) {
 153               remove(path)
 154                 .catch(err => logger.error('Cannot delete path on YoutubeDL error.', { err }))
 155
 156               return rej(err)
 157             }
 158
 159             return res(path)
 160           } catch (err) {
 161             return rej(err)
 162           }
 163         })
 164
 165         timer = setTimeout(() => {
 166           const err = new Error('YoutubeDL download timeout.')
 167
 168           guessVideoPathWithExtension(pathWithoutExtension, fileExt)
 169             .then(path => remove(path))
 170             .finally(() => rej(err))
 171             .catch(err => {
 172               logger.error('Cannot remove file in youtubeDL timeout.', { err })
 173               return rej(err)
 174             })
 175         }, timeout)
 176       })
 177       .catch(err => rej(err))
 178   })
 179 }
 180
 181 // Thanks: https://github.com/przemyslawpluta/node-youtube-dl/blob/master/lib/downloader.js
 182 // We rewrote it to avoid sync calls
 183 async function updateYoutubeDLBinary () {
 184   logger.info('Updating youtubeDL binary.')
 185
 186   const binDirectory = join(root(), 'node_modules', 'youtube-dl', 'bin')
 187   const bin = join(binDirectory, 'youtube-dl')
 188   const detailsPath = join(binDirectory, 'details')
 189   const url = process.env.YOUTUBE_DL_DOWNLOAD_HOST || 'https://yt-dl.org/downloads/latest/youtube-dl'
 190
 191   await ensureDir(binDirectory)
 192
 193   return new Promise(res => {
 194     request.get(url, { followRedirect: false }, (err, result) => {
 195       if (err) {
 196         logger.error('Cannot update youtube-dl.', { err })
 197         return res()
 198       }
 199
 200       if (result.statusCode !== HttpStatusCode.FOUND_302) {
 201         logger.error('youtube-dl update error: did not get redirect for the latest version link. Status %d', result.statusCode)
 202         return res()
 203       }
 204
 205       const url = result.headers.location
 206       const downloadFile = request.get(url)
 207       const newVersion = /yt-dl\.org\/downloads\/(\d{4}\.\d\d\.\d\d(\.\d)?)\/youtube-dl/.exec(url)[1]
 208
 209       downloadFile.on('response', result => {
 210         if (result.statusCode !== HttpStatusCode.OK_200) {
 211           logger.error('Cannot update youtube-dl: new version response is not 200, it\'s %d.', result.statusCode)
 212           return res()
 213         }
 214
 215         const writeStream = createWriteStream(bin, { mode: 493 }).on('error', err => {
 216           logger.error('youtube-dl update error in write stream', { err })
 217           return res()
 218         })
 219
 220         downloadFile.pipe(writeStream)
 221       })
 222
 223       downloadFile.on('error', err => {
 224         logger.error('youtube-dl update error.', { err })
 225         return res()
 226       })
 227
 228       downloadFile.on('end', () => {
 229         const details = JSON.stringify({ version: newVersion, path: bin, exec: 'youtube-dl' })
 230         writeFile(detailsPath, details, { encoding: 'utf8' }, err => {
 231           if (err) {
 232             logger.error('youtube-dl update error: cannot write details.', { err })
 233             return res()
 234           }
 235
 236           logger.info('youtube-dl updated to version %s.', newVersion)
 237           return res()
 238         })
 239       })
 240     })
 241   })
 242 }
 243
 244 async function safeGetYoutubeDL () {
 245   let youtubeDL
 246
 247   try {
 248     youtubeDL = require('youtube-dl')
 249   } catch (e) {
 250     // Download binary
 251     await updateYoutubeDLBinary()
 252     youtubeDL = require('youtube-dl')
 253   }
 254
 255   return youtubeDL
 256 }
 257
 258 function buildOriginallyPublishedAt (obj: any) {
 259   let originallyPublishedAt: Date = null
 260
 261   const uploadDateMatcher = /^(\d{4})(\d{2})(\d{2})$/.exec(obj.upload_date)
 262   if (uploadDateMatcher) {
 263     originallyPublishedAt = new Date()
 264     originallyPublishedAt.setHours(0, 0, 0, 0)
 265
 266     const year = parseInt(uploadDateMatcher[1], 10)
 267     // Month starts from 0
 268     const month = parseInt(uploadDateMatcher[2], 10) - 1
 269     const day = parseInt(uploadDateMatcher[3], 10)
 270
 271     originallyPublishedAt.setFullYear(year, month, day)
 272   }
 273
 274   return originallyPublishedAt
 275 }
 276
 277 // ---------------------------------------------------------------------------
 278
 279 export {
 280   updateYoutubeDLBinary,
 281   getYoutubeDLVideoFormat,
 282   downloadYoutubeDLVideo,
 283   getYoutubeDLSubs,
 284   getYoutubeDLInfo,
 285   safeGetYoutubeDL,
 286   buildOriginallyPublishedAt
 287 }
 288
 289 // ---------------------------------------------------------------------------
 290
 291 async function guessVideoPathWithExtension (tmpPath: string, sourceExt: string) {
 292   if (!isVideoFileExtnameValid(sourceExt)) {
 293     throw new Error('Invalid video extension ' + sourceExt)
 294   }
 295
 296   const extensions = [ sourceExt, '.mp4', '.mkv', '.webm' ]
 297
 298   for (const extension of extensions) {
 299     const path = tmpPath + extension
 300
 301     if (await pathExists(path)) return path
 302   }
 303
 304   throw new Error('Cannot guess path of ' + tmpPath)
 305 }
 306
 307 function normalizeObject (obj: any) {
 308   const newObj: any = {}
 309
 310   for (const key of Object.keys(obj)) {
 311     // Deprecated key
 312     if (key === 'resolution') continue
 313
 314     const value = obj[key]
 315
 316     if (typeof value === 'string') {
 317       newObj[key] = value.normalize()
 318     } else {
 319       newObj[key] = value
 320     }
 321   }
 322
 323   return newObj
 324 }
 325
 326 function buildVideoInfo (obj: any): YoutubeDLInfo {
 327   return {
 328     name: titleTruncation(obj.title),
 329     description: descriptionTruncation(obj.description),
 330     category: getCategory(obj.categories),
 331     licence: getLicence(obj.license),
 332     language: getLanguage(obj.language),
 333     nsfw: isNSFW(obj),
 334     tags: getTags(obj.tags),
 335     thumbnailUrl: obj.thumbnail || undefined,
 336     originallyPublishedAt: buildOriginallyPublishedAt(obj),
 337     ext: obj.ext
 338   }
 339 }
 340
 341 function titleTruncation (title: string) {
 342   return peertubeTruncate(title, {
 343     length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
 344     separator: /,? +/,
 345     omission: ' […]'
 346   })
 347 }
 348
 349 function descriptionTruncation (description: string) {
 350   if (!description || description.length < CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.min) return undefined
 351
 352   return peertubeTruncate(description, {
 353     length: CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.max,
 354     separator: /,? +/,
 355     omission: ' […]'
 356   })
 357 }
 358
 359 function isNSFW (info: any) {
 360   return info.age_limit && info.age_limit >= 16
 361 }
 362
 363 function getTags (tags: any) {
 364   if (Array.isArray(tags) === false) return []
 365
 366   return tags
 367     .filter(t => t.length < CONSTRAINTS_FIELDS.VIDEOS.TAG.max && t.length > CONSTRAINTS_FIELDS.VIDEOS.TAG.min)
 368     .map(t => t.normalize())
 369     .slice(0, 5)
 370 }
 371
 372 function getLicence (licence: string) {
 373   if (!licence) return undefined
 374
 375   if (licence.includes('Creative Commons Attribution')) return 1
 376
 377   for (const key of Object.keys(VIDEO_LICENCES)) {
 378     const peertubeLicence = VIDEO_LICENCES[key]
 379     if (peertubeLicence.toLowerCase() === licence.toLowerCase()) return parseInt(key, 10)
 380   }
 381
 382   return undefined
 383 }
 384
 385 function getCategory (categories: string[]) {
 386   if (!categories) return undefined
 387
 388   const categoryString = categories[0]
 389   if (!categoryString || typeof categoryString !== 'string') return undefined
 390
 391   if (categoryString === 'News & Politics') return 11
 392
 393   for (const key of Object.keys(VIDEO_CATEGORIES)) {
 394     const category = VIDEO_CATEGORIES[key]
 395     if (categoryString.toLowerCase() === category.toLowerCase()) return parseInt(key, 10)
 396   }
 397
 398   return undefined
 399 }
 400
 401 function getLanguage (language: string) {
 402   return VIDEO_LANGUAGES[language] ? language : undefined
 403 }
 404
 405 function wrapWithProxyOptions (options: string[]) {
 406   if (CONFIG.IMPORT.VIDEOS.HTTP.PROXY.ENABLED) {
 407     logger.debug('Using proxy for YoutubeDL')
 408
 409     return [ '--proxy', CONFIG.IMPORT.VIDEOS.HTTP.PROXY.URL ].concat(options)
 410   }
 411
 412   return options
 413 }