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