server/helpers/youtube-dl.ts

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