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