server/tools/peertube-import-videos.ts

   1 import { registerTSPaths } from '../helpers/register-ts-paths'
   2 registerTSPaths()
   3
   4 import * as program from 'commander'
   5 import { join } from 'path'
   6 import { doRequestAndSaveToFile } from '../helpers/requests'
   7 import { CONSTRAINTS_FIELDS } from '../initializers/constants'
   8 import { getClient, getVideoCategories, login, searchVideoWithSort, uploadVideo } from '../../shared/extra-utils/index'
   9 import { truncate } from 'lodash'
  10 import * as prompt from 'prompt'
  11 import { accessSync, constants } from 'fs'
  12 import { remove } from 'fs-extra'
  13 import { sha256 } from '../helpers/core-utils'
  14 import { buildOriginallyPublishedAt, safeGetYoutubeDL } from '../helpers/youtube-dl'
  15 import { buildCommonVideoOptions, buildVideoAttributesFromCommander, getLogger, getServerCredentials } from './cli'
  16
  17 type UserInfo = {
  18   username: string
  19   password: string
  20 }
  21
  22 const processOptions = {
  23   maxBuffer: Infinity
  24 }
  25
  26 let command = program
  27   .name('import-videos')
  28
  29 command = buildCommonVideoOptions(command)
  30
  31 command
  32   .option('-u, --url <url>', 'Server url')
  33   .option('-U, --username <username>', 'Username')
  34   .option('-p, --password <token>', 'Password')
  35   .option('--target-url <targetUrl>', 'Video target URL')
  36   .option('--since <since>', 'Publication date (inclusive) since which the videos can be imported (YYYY-MM-DD)', parseDate)
  37   .option('--until <until>', 'Publication date (inclusive) until which the videos can be imported (YYYY-MM-DD)', parseDate)
  38   .option('--first <first>', 'Process first n elements of returned playlist')
  39   .option('--last <last>', 'Process last n elements of returned playlist')
  40   .option('-T, --tmpdir <tmpdir>', 'Working directory', __dirname)
  41   .usage("[global options] [ -- youtube-dl options]")
  42   .parse(process.argv)
  43
  44 const log = getLogger(program['verbose'])
  45
  46 getServerCredentials(command)
  47   .then(({ url, username, password }) => {
  48     if (!program['targetUrl']) {
  49       exitError('--target-url field is required.')
  50     }
  51
  52     try {
  53       accessSync(program['tmpdir'], constants.R_OK | constants.W_OK)
  54     } catch (e) {
  55       exitError('--tmpdir %s: directory does not exist or is not accessible', program['tmpdir'])
  56     }
  57
  58     url = normalizeTargetUrl(url)
  59     program['targetUrl'] = normalizeTargetUrl(program['targetUrl'])
  60
  61     const user = { username, password }
  62
  63     run(url, user)
  64       .catch(err => exitError(err))
  65   })
  66   .catch(err => console.error(err))
  67
  68 async function run (url: string, user: UserInfo) {
  69   if (!user.password) {
  70     user.password = await promptPassword()
  71   }
  72
  73   const youtubeDL = await safeGetYoutubeDL()
  74
  75   let info = await getYoutubeDLInfo(youtubeDL, program['targetUrl'], command.args)
  76
  77   if (!Array.isArray(info)) info = [ info ]
  78
  79   // Try to fix youtube channels upload
  80   const uploadsObject = info.find(i => !i.ie_key && !i.duration && i.title === 'Uploads')
  81
  82   if (uploadsObject) {
  83     console.log('Fixing URL to %s.', uploadsObject.url)
  84
  85     info = await getYoutubeDLInfo(youtubeDL, uploadsObject.url, command.args)
  86   }
  87
  88   let infoArray: any[]
  89
  90   // Normalize utf8 fields
  91   infoArray = [].concat(info)
  92   if (program['first']) {
  93     infoArray = infoArray.slice(0, program['first'])
  94   } else if (program['last']) {
  95     infoArray = infoArray.slice(-program['last'])
  96   }
  97   infoArray = infoArray.map(i => normalizeObject(i))
  98
  99   log.info('Will download and upload %d videos.\n', infoArray.length)
 100
 101   for (const info of infoArray) {
 102     try {
 103       await processVideo({
 104         cwd: program['tmpdir'],
 105         url,
 106         user,
 107         youtubeInfo: info
 108       })
 109     } catch (err) {
 110       console.error('Cannot process video.', { info, url })
 111     }
 112   }
 113
 114   log.info('Video/s for user %s imported: %s', user.username, program['targetUrl'])
 115   process.exit(0)
 116 }
 117
 118 function processVideo (parameters: {
 119   cwd: string
 120   url: string
 121   user: { username: string, password: string }
 122   youtubeInfo: any
 123 }) {
 124   const { youtubeInfo, cwd, url, user } = parameters
 125
 126   return new Promise(async res => {
 127     log.debug('Fetching object.', youtubeInfo)
 128
 129     const videoInfo = await fetchObject(youtubeInfo)
 130     log.debug('Fetched object.', videoInfo)
 131
 132     const originallyPublishedAt = buildOriginallyPublishedAt(videoInfo)
 133
 134     if (program['since'] && originallyPublishedAt && originallyPublishedAt.getTime() < program['since'].getTime()) {
 135       log.info('Video "%s" has been published before "%s", don\'t upload it.\n',
 136         videoInfo.title, formatDate(program['since']))
 137       return res()
 138     }
 139
 140     if (program['until'] && originallyPublishedAt && originallyPublishedAt.getTime() > program['until'].getTime()) {
 141       log.info('Video "%s" has been published after "%s", don\'t upload it.\n',
 142         videoInfo.title, formatDate(program['until']))
 143       return res()
 144     }
 145
 146     const result = await searchVideoWithSort(url, videoInfo.title, '-match')
 147
 148     log.info('############################################################\n')
 149
 150     if (result.body.data.find(v => v.name === videoInfo.title)) {
 151       log.info('Video "%s" already exists, don\'t reupload it.\n', videoInfo.title)
 152       return res()
 153     }
 154
 155     const path = join(cwd, sha256(videoInfo.url) + '.mp4')
 156
 157     log.info('Downloading video "%s"...', videoInfo.title)
 158
 159     const options = [ '-f', 'bestvideo[ext=mp4]+bestaudio[ext=m4a]/best', ...command.args, '-o', path ]
 160     try {
 161       const youtubeDL = await safeGetYoutubeDL()
 162       youtubeDL.exec(videoInfo.url, options, processOptions, async (err, output) => {
 163         if (err) {
 164           log.error(err)
 165           return res()
 166         }
 167
 168         log.info(output.join('\n'))
 169         await uploadVideoOnPeerTube({
 170           cwd,
 171           url,
 172           user,
 173           videoInfo: normalizeObject(videoInfo),
 174           videoPath: path
 175         })
 176         return res()
 177       })
 178     } catch (err) {
 179       log.error(err.message)
 180       return res()
 181     }
 182   })
 183 }
 184
 185 async function uploadVideoOnPeerTube (parameters: {
 186   videoInfo: any
 187   videoPath: string
 188   cwd: string
 189   url: string
 190   user: { username: string, password: string }
 191 }) {
 192   const { videoInfo, videoPath, cwd, url, user } = parameters
 193
 194   const category = await getCategory(videoInfo.categories, url)
 195   const licence = getLicence(videoInfo.license)
 196   let tags = []
 197   if (Array.isArray(videoInfo.tags)) {
 198     tags = videoInfo.tags
 199                     .filter(t => t.length < CONSTRAINTS_FIELDS.VIDEOS.TAG.max && t.length > CONSTRAINTS_FIELDS.VIDEOS.TAG.min)
 200                     .map(t => t.normalize())
 201                     .slice(0, 5)
 202   }
 203
 204   let thumbnailfile
 205   if (videoInfo.thumbnail) {
 206     thumbnailfile = join(cwd, sha256(videoInfo.thumbnail) + '.jpg')
 207
 208     await doRequestAndSaveToFile({
 209       method: 'GET',
 210       uri: videoInfo.thumbnail
 211     }, thumbnailfile)
 212   }
 213
 214   const originallyPublishedAt = buildOriginallyPublishedAt(videoInfo)
 215
 216   const defaultAttributes = {
 217     name: truncate(videoInfo.title, {
 218       length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
 219       separator: /,? +/,
 220       omission: ' […]'
 221     }),
 222     category,
 223     licence,
 224     nsfw: isNSFW(videoInfo),
 225     description: videoInfo.description,
 226     tags
 227   }
 228
 229   const videoAttributes = await buildVideoAttributesFromCommander(url, program, defaultAttributes)
 230
 231   Object.assign(videoAttributes, {
 232     originallyPublishedAt: originallyPublishedAt ? originallyPublishedAt.toISOString() : null,
 233     thumbnailfile,
 234     previewfile: thumbnailfile,
 235     fixture: videoPath
 236   })
 237
 238   log.info('\nUploading on PeerTube video "%s".', videoAttributes.name)
 239
 240   let accessToken = await getAccessTokenOrDie(url, user)
 241
 242   try {
 243     await uploadVideo(url, accessToken, videoAttributes)
 244   } catch (err) {
 245     if (err.message.indexOf('401') !== -1) {
 246       log.info('Got 401 Unauthorized, token may have expired, renewing token and retry.')
 247
 248       accessToken = await getAccessTokenOrDie(url, user)
 249
 250       await uploadVideo(url, accessToken, videoAttributes)
 251     } else {
 252       exitError(err.message)
 253     }
 254   }
 255
 256   await remove(videoPath)
 257   if (thumbnailfile) await remove(thumbnailfile)
 258
 259   log.warn('Uploaded video "%s"!\n', videoAttributes.name)
 260 }
 261
 262 /* ---------------------------------------------------------- */
 263
 264 async function getCategory (categories: string[], url: string) {
 265   if (!categories) return undefined
 266
 267   const categoryString = categories[0]
 268
 269   if (categoryString === 'News & Politics') return 11
 270
 271   const res = await getVideoCategories(url)
 272   const categoriesServer = res.body
 273
 274   for (const key of Object.keys(categoriesServer)) {
 275     const categoryServer = categoriesServer[key]
 276     if (categoryString.toLowerCase() === categoryServer.toLowerCase()) return parseInt(key, 10)
 277   }
 278
 279   return undefined
 280 }
 281
 282 function getLicence (licence: string) {
 283   if (!licence) return undefined
 284
 285   if (licence.includes('Creative Commons Attribution licence')) return 1
 286
 287   return undefined
 288 }
 289
 290 function normalizeObject (obj: any) {
 291   const newObj: any = {}
 292
 293   for (const key of Object.keys(obj)) {
 294     // Deprecated key
 295     if (key === 'resolution') continue
 296
 297     const value = obj[key]
 298
 299     if (typeof value === 'string') {
 300       newObj[key] = value.normalize()
 301     } else {
 302       newObj[key] = value
 303     }
 304   }
 305
 306   return newObj
 307 }
 308
 309 function fetchObject (info: any) {
 310   const url = buildUrl(info)
 311
 312   return new Promise<any>(async (res, rej) => {
 313     const youtubeDL = await safeGetYoutubeDL()
 314     youtubeDL.getInfo(url, undefined, processOptions, (err, videoInfo) => {
 315       if (err) return rej(err)
 316
 317       const videoInfoWithUrl = Object.assign(videoInfo, { url })
 318       return res(normalizeObject(videoInfoWithUrl))
 319     })
 320   })
 321 }
 322
 323 function buildUrl (info: any) {
 324   const webpageUrl = info.webpage_url as string
 325   if (webpageUrl?.match(/^https?:\/\//)) return webpageUrl
 326
 327   const url = info.url as string
 328   if (url?.match(/^https?:\/\//)) return url
 329
 330   // It seems youtube-dl does not return the video url
 331   return 'https://www.youtube.com/watch?v=' + info.id
 332 }
 333
 334 function isNSFW (info: any) {
 335   return info.age_limit && info.age_limit >= 16
 336 }
 337
 338 function normalizeTargetUrl (url: string) {
 339   let normalizedUrl = url.replace(/\/+$/, '')
 340
 341   if (!normalizedUrl.startsWith('http://') && !normalizedUrl.startsWith('https://')) {
 342     normalizedUrl = 'https://' + normalizedUrl
 343   }
 344
 345   return normalizedUrl
 346 }
 347
 348 async function promptPassword () {
 349   return new Promise<string>((res, rej) => {
 350     prompt.start()
 351     const schema = {
 352       properties: {
 353         password: {
 354           hidden: true,
 355           required: true
 356         }
 357       }
 358     }
 359     prompt.get(schema, function (err, result) {
 360       if (err) {
 361         return rej(err)
 362       }
 363       return res(result.password)
 364     })
 365   })
 366 }
 367
 368 async function getAccessTokenOrDie (url: string, user: UserInfo) {
 369   const resClient = await getClient(url)
 370   const client = {
 371     id: resClient.body.client_id,
 372     secret: resClient.body.client_secret
 373   }
 374
 375   try {
 376     const res = await login(url, client, user)
 377     return res.body.access_token
 378   } catch (err) {
 379     exitError('Cannot authenticate. Please check your username/password.')
 380   }
 381 }
 382
 383 function parseDate (dateAsStr: string): Date {
 384   if (!/\d{4}-\d{2}-\d{2}/.test(dateAsStr)) {
 385     exitError(`Invalid date passed: ${dateAsStr}. Expected format: YYYY-MM-DD. See help for usage.`)
 386   }
 387   const date = new Date(dateAsStr)
 388   date.setHours(0, 0, 0)
 389   if (isNaN(date.getTime())) {
 390     exitError(`Invalid date passed: ${dateAsStr}. See help for usage.`)
 391   }
 392   return date
 393 }
 394
 395 function formatDate (date: Date): string {
 396   return date.toISOString().split('T')[0]
 397 }
 398
 399 function exitError (message: string, ...meta: any[]) {
 400   // use console.error instead of log.error here
 401   console.error(message, ...meta)
 402   process.exit(-1)
 403 }
 404
 405 function getYoutubeDLInfo (youtubeDL: any, url: string, args: string[]) {
 406   return new Promise<any>((res, rej) => {
 407     const options = [ '-j', '--flat-playlist', '--playlist-reverse', ...args ]
 408
 409     youtubeDL.getInfo(url, options, processOptions, async (err, info) => {
 410       if (err) return rej(err)
 411
 412       return res(info)
 413     })
 414   })
 415 }