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