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