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