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