server/helpers/youtube-dl.ts

   1 import { createWriteStream } from 'fs'
   2 import { ensureDir, move, 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             // If youtube-dl did not guess an extension for our file, just use .mp4 as default
 151             if (await pathExists(pathWithoutExtension)) {
 152               await move(pathWithoutExtension, pathWithoutExtension + '.mp4')
 153             }
 154
 155             const path = await guessVideoPathWithExtension(pathWithoutExtension, fileExt)
 156
 157             if (err) {
 158               remove(path)
 159                 .catch(err => logger.error('Cannot delete path on YoutubeDL error.', { err }))
 160
 161               return rej(err)
 162             }
 163
 164             return res(path)
 165           } catch (err) {
 166             return rej(err)
 167           }
 168         })
 169
 170         timer = setTimeout(() => {
 171           const err = new Error('YoutubeDL download timeout.')
 172
 173           guessVideoPathWithExtension(pathWithoutExtension, fileExt)
 174             .then(path => remove(path))
 175             .finally(() => rej(err))
 176             .catch(err => {
 177               logger.error('Cannot remove file in youtubeDL timeout.', { err })
 178               return rej(err)
 179             })
 180         }, timeout)
 181       })
 182       .catch(err => rej(err))
 183   })
 184 }
 185
 186 // Thanks: https://github.com/przemyslawpluta/node-youtube-dl/blob/master/lib/downloader.js
 187 // We rewrote it to avoid sync calls
 188 async function updateYoutubeDLBinary () {
 189   logger.info('Updating youtubeDL binary.')
 190
 191   const binDirectory = join(root(), 'node_modules', 'youtube-dl', 'bin')
 192   const bin = join(binDirectory, 'youtube-dl')
 193   const detailsPath = join(binDirectory, 'details')
 194   const url = process.env.YOUTUBE_DL_DOWNLOAD_HOST || 'https://yt-dl.org/downloads/latest/youtube-dl'
 195
 196   await ensureDir(binDirectory)
 197
 198   return new Promise(res => {
 199     request.get(url, { followRedirect: false }, (err, result) => {
 200       if (err) {
 201         logger.error('Cannot update youtube-dl.', { err })
 202         return res()
 203       }
 204
 205       if (result.statusCode !== HttpStatusCode.FOUND_302) {
 206         logger.error('youtube-dl update error: did not get redirect for the latest version link. Status %d', result.statusCode)
 207         return res()
 208       }
 209
 210       const url = result.headers.location
 211       const downloadFile = request.get(url)
 212       const newVersion = /yt-dl\.org\/downloads\/(\d{4}\.\d\d\.\d\d(\.\d)?)\/youtube-dl/.exec(url)[1]
 213
 214       downloadFile.on('response', result => {
 215         if (result.statusCode !== HttpStatusCode.OK_200) {
 216           logger.error('Cannot update youtube-dl: new version response is not 200, it\'s %d.', result.statusCode)
 217           return res()
 218         }
 219
 220         const writeStream = createWriteStream(bin, { mode: 493 }).on('error', err => {
 221           logger.error('youtube-dl update error in write stream', { err })
 222           return res()
 223         })
 224
 225         downloadFile.pipe(writeStream)
 226       })
 227
 228       downloadFile.on('error', err => {
 229         logger.error('youtube-dl update error.', { err })
 230         return res()
 231       })
 232
 233       downloadFile.on('end', () => {
 234         const details = JSON.stringify({ version: newVersion, path: bin, exec: 'youtube-dl' })
 235         writeFile(detailsPath, details, { encoding: 'utf8' }, err => {
 236           if (err) {
 237             logger.error('youtube-dl update error: cannot write details.', { err })
 238             return res()
 239           }
 240
 241           logger.info('youtube-dl updated to version %s.', newVersion)
 242           return res()
 243         })
 244       })
 245     })
 246   })
 247 }
 248
 249 async function safeGetYoutubeDL () {
 250   let youtubeDL
 251
 252   try {
 253     youtubeDL = require('youtube-dl')
 254   } catch (e) {
 255     // Download binary
 256     await updateYoutubeDLBinary()
 257     youtubeDL = require('youtube-dl')
 258   }
 259
 260   return youtubeDL
 261 }
 262
 263 function buildOriginallyPublishedAt (obj: any) {
 264   let originallyPublishedAt: Date = null
 265
 266   const uploadDateMatcher = /^(\d{4})(\d{2})(\d{2})$/.exec(obj.upload_date)
 267   if (uploadDateMatcher) {
 268     originallyPublishedAt = new Date()
 269     originallyPublishedAt.setHours(0, 0, 0, 0)
 270
 271     const year = parseInt(uploadDateMatcher[1], 10)
 272     // Month starts from 0
 273     const month = parseInt(uploadDateMatcher[2], 10) - 1
 274     const day = parseInt(uploadDateMatcher[3], 10)
 275
 276     originallyPublishedAt.setFullYear(year, month, day)
 277   }
 278
 279   return originallyPublishedAt
 280 }
 281
 282 // ---------------------------------------------------------------------------
 283
 284 export {
 285   updateYoutubeDLBinary,
 286   getYoutubeDLVideoFormat,
 287   downloadYoutubeDLVideo,
 288   getYoutubeDLSubs,
 289   getYoutubeDLInfo,
 290   safeGetYoutubeDL,
 291   buildOriginallyPublishedAt
 292 }
 293
 294 // ---------------------------------------------------------------------------
 295
 296 async function guessVideoPathWithExtension (tmpPath: string, sourceExt: string) {
 297   if (!isVideoFileExtnameValid(sourceExt)) {
 298     throw new Error('Invalid video extension ' + sourceExt)
 299   }
 300
 301   const extensions = [ sourceExt, '.mp4', '.mkv', '.webm' ]
 302
 303   for (const extension of extensions) {
 304     const path = tmpPath + extension
 305
 306     if (await pathExists(path)) return path
 307   }
 308
 309   throw new Error('Cannot guess path of ' + tmpPath)
 310 }
 311
 312 function normalizeObject (obj: any) {
 313   const newObj: any = {}
 314
 315   for (const key of Object.keys(obj)) {
 316     // Deprecated key
 317     if (key === 'resolution') continue
 318
 319     const value = obj[key]
 320
 321     if (typeof value === 'string') {
 322       newObj[key] = value.normalize()
 323     } else {
 324       newObj[key] = value
 325     }
 326   }
 327
 328   return newObj
 329 }
 330
 331 function buildVideoInfo (obj: any): YoutubeDLInfo {
 332   return {
 333     name: titleTruncation(obj.title),
 334     description: descriptionTruncation(obj.description),
 335     category: getCategory(obj.categories),
 336     licence: getLicence(obj.license),
 337     language: getLanguage(obj.language),
 338     nsfw: isNSFW(obj),
 339     tags: getTags(obj.tags),
 340     thumbnailUrl: obj.thumbnail || undefined,
 341     originallyPublishedAt: buildOriginallyPublishedAt(obj),
 342     ext: obj.ext
 343   }
 344 }
 345
 346 function titleTruncation (title: string) {
 347   return peertubeTruncate(title, {
 348     length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
 349     separator: /,? +/,
 350     omission: ' […]'
 351   })
 352 }
 353
 354 function descriptionTruncation (description: string) {
 355   if (!description || description.length < CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.min) return undefined
 356
 357   return peertubeTruncate(description, {
 358     length: CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.max,
 359     separator: /,? +/,
 360     omission: ' […]'
 361   })
 362 }
 363
 364 function isNSFW (info: any) {
 365   return info.age_limit && info.age_limit >= 16
 366 }
 367
 368 function getTags (tags: any) {
 369   if (Array.isArray(tags) === false) return []
 370
 371   return tags
 372     .filter(t => t.length < CONSTRAINTS_FIELDS.VIDEOS.TAG.max && t.length > CONSTRAINTS_FIELDS.VIDEOS.TAG.min)
 373     .map(t => t.normalize())
 374     .slice(0, 5)
 375 }
 376
 377 function getLicence (licence: string) {
 378   if (!licence) return undefined
 379
 380   if (licence.includes('Creative Commons Attribution')) return 1
 381
 382   for (const key of Object.keys(VIDEO_LICENCES)) {
 383     const peertubeLicence = VIDEO_LICENCES[key]
 384     if (peertubeLicence.toLowerCase() === licence.toLowerCase()) return parseInt(key, 10)
 385   }
 386
 387   return undefined
 388 }
 389
 390 function getCategory (categories: string[]) {
 391   if (!categories) return undefined
 392
 393   const categoryString = categories[0]
 394   if (!categoryString || typeof categoryString !== 'string') return undefined
 395
 396   if (categoryString === 'News & Politics') return 11
 397
 398   for (const key of Object.keys(VIDEO_CATEGORIES)) {
 399     const category = VIDEO_CATEGORIES[key]
 400     if (categoryString.toLowerCase() === category.toLowerCase()) return parseInt(key, 10)
 401   }
 402
 403   return undefined
 404 }
 405
 406 function getLanguage (language: string) {
 407   return VIDEO_LANGUAGES[language] ? language : undefined
 408 }
 409
 410 function wrapWithProxyOptions (options: string[]) {
 411   if (CONFIG.IMPORT.VIDEOS.HTTP.PROXY.ENABLED) {
 412     logger.debug('Using proxy for YoutubeDL')
 413
 414     return [ '--proxy', CONFIG.IMPORT.VIDEOS.HTTP.PROXY.URL ].concat(options)
 415   }
 416
 417   return options
 418 }