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