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