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 { accessSync, constants } from 'fs'
   6 import { remove } from 'fs-extra'
   7 import { truncate } from 'lodash'
   8 import { join } from 'path'
   9 import * as prompt from 'prompt'
  10 import { promisify } from 'util'
  11 import { advancedVideosSearch, getClient, getVideoCategories, login, uploadVideo } from '../../shared/extra-utils/index'
  12 import { sha256 } from '../helpers/core-utils'
  13 import { doRequestAndSaveToFile } from '../helpers/requests'
  14 import { CONSTRAINTS_FIELDS } from '../initializers/constants'
  15 import { buildCommonVideoOptions, buildVideoAttributesFromCommander, getLogger, getServerCredentials } from './cli'
  16 import { YoutubeDL } from '@server/helpers/youtube-dl'
  17
  18 type UserInfo = {
  19   username: string
  20   password: string
  21 }
  22
  23 const processOptions = {
  24   maxBuffer: Infinity
  25 }
  26
  27 let command = program
  28   .name('import-videos')
  29
  30 command = buildCommonVideoOptions(command)
  31
  32 command
  33   .option('-u, --url <url>', 'Server url')
  34   .option('-U, --username <username>', 'Username')
  35   .option('-p, --password <token>', 'Password')
  36   .option('--target-url <targetUrl>', 'Video target URL')
  37   .option('--since <since>', 'Publication date (inclusive) since which the videos can be imported (YYYY-MM-DD)', parseDate)
  38   .option('--until <until>', 'Publication date (inclusive) until which the videos can be imported (YYYY-MM-DD)', parseDate)
  39   .option('--first <first>', 'Process first n elements of returned playlist')
  40   .option('--last <last>', 'Process last n elements of returned playlist')
  41   .option('--wait-interval <waitInterval>', 'Duration between two video imports (in seconds)', convertIntoMs)
  42   .option('-T, --tmpdir <tmpdir>', 'Working directory', __dirname)
  43   .usage("[global options] [ -- youtube-dl options]")
  44   .parse(process.argv)
  45
  46 const options = command.opts()
  47
  48 const log = getLogger(options.verbose)
  49
  50 getServerCredentials(command)
  51   .then(({ url, username, password }) => {
  52     if (!options.targetUrl) {
  53       exitError('--target-url field is required.')
  54     }
  55
  56     try {
  57       accessSync(options.tmpdir, constants.R_OK | constants.W_OK)
  58     } catch (e) {
  59       exitError('--tmpdir %s: directory does not exist or is not accessible', options.tmpdir)
  60     }
  61
  62     url = normalizeTargetUrl(url)
  63     options.targetUrl = normalizeTargetUrl(options.targetUrl)
  64
  65     const user = { username, password }
  66
  67     run(url, user)
  68       .catch(err => exitError(err))
  69   })
  70   .catch(err => console.error(err))
  71
  72 async function run (url: string, user: UserInfo) {
  73   if (!user.password) {
  74     user.password = await promptPassword()
  75   }
  76
  77   const youtubeDLBinary = await YoutubeDL.safeGetYoutubeDL()
  78
  79   let info = await getYoutubeDLInfo(youtubeDLBinary, options.targetUrl, command.args)
  80
  81   if (!Array.isArray(info)) info = [ info ]
  82
  83   // Try to fix youtube channels upload
  84   const uploadsObject = info.find(i => !i.ie_key && !i.duration && i.title === 'Uploads')
  85
  86   if (uploadsObject) {
  87     console.log('Fixing URL to %s.', uploadsObject.url)
  88
  89     info = await getYoutubeDLInfo(youtubeDLBinary, uploadsObject.url, command.args)
  90   }
  91
  92   let infoArray: any[]
  93
  94   infoArray = [].concat(info)
  95   if (options.first) {
  96     infoArray = infoArray.slice(0, options.first)
  97   } else if (options.last) {
  98     infoArray = infoArray.slice(-options.last)
  99   }
 100   // Normalize utf8 fields
 101   infoArray = infoArray.map(i => normalizeObject(i))
 102
 103   log.info('Will download and upload %d videos.\n', infoArray.length)
 104
 105   for (const [ index, info ] of infoArray.entries()) {
 106     try {
 107       if (index > 0 && options.waitInterval) {
 108         log.info("Wait for %d seconds before continuing.", options.waitInterval / 1000)
 109         await new Promise(res => setTimeout(res, options.waitInterval))
 110       }
 111       await processVideo({
 112         cwd: options.tmpdir,
 113         url,
 114         user,
 115         youtubeInfo: info
 116       })
 117     } catch (err) {
 118       console.error('Cannot process video.', { info, url, err })
 119     }
 120   }
 121
 122   log.info('Video/s for user %s imported: %s', user.username, options.targetUrl)
 123   process.exit(0)
 124 }
 125
 126 async function processVideo (parameters: {
 127   cwd: string
 128   url: string
 129   user: { username: string, password: string }
 130   youtubeInfo: any
 131 }) {
 132   const { youtubeInfo, cwd, url, user } = parameters
 133   const youtubeDL = new YoutubeDL('', [])
 134
 135   log.debug('Fetching object.', youtubeInfo)
 136
 137   const videoInfo = await fetchObject(youtubeInfo)
 138   log.debug('Fetched object.', videoInfo)
 139
 140   const originallyPublishedAt = youtubeDL.buildOriginallyPublishedAt(videoInfo)
 141   if (options.since && originallyPublishedAt && originallyPublishedAt.getTime() < options.since.getTime()) {
 142     log.info('Video "%s" has been published before "%s", don\'t upload it.\n',
 143       videoInfo.title, formatDate(options.since))
 144     return
 145   }
 146   if (options.until && originallyPublishedAt && originallyPublishedAt.getTime() > options.until.getTime()) {
 147     log.info('Video "%s" has been published after "%s", don\'t upload it.\n',
 148       videoInfo.title, formatDate(options.until))
 149     return
 150   }
 151
 152   const result = await advancedVideosSearch(url, { search: videoInfo.title, sort: '-match', searchTarget: 'local' })
 153
 154   log.info('############################################################\n')
 155
 156   if (result.body.data.find(v => v.name === videoInfo.title)) {
 157     log.info('Video "%s" already exists, don\'t reupload it.\n', videoInfo.title)
 158     return
 159   }
 160
 161   const path = join(cwd, sha256(videoInfo.url) + '.mp4')
 162
 163   log.info('Downloading video "%s"...', videoInfo.title)
 164
 165   const youtubeDLOptions = [ '-f', youtubeDL.getYoutubeDLVideoFormat(), ...command.args, '-o', path ]
 166   try {
 167     const youtubeDLBinary = await YoutubeDL.safeGetYoutubeDL()
 168     const youtubeDLExec = promisify(youtubeDLBinary.exec).bind(youtubeDLBinary)
 169     const output = await youtubeDLExec(videoInfo.url, youtubeDLOptions, processOptions)
 170     log.info(output.join('\n'))
 171     await uploadVideoOnPeerTube({
 172       youtubeDL,
 173       cwd,
 174       url,
 175       user,
 176       videoInfo: normalizeObject(videoInfo),
 177       videoPath: path
 178     })
 179   } catch (err) {
 180     log.error(err.message)
 181   }
 182 }
 183
 184 async function uploadVideoOnPeerTube (parameters: {
 185   youtubeDL: YoutubeDL
 186   videoInfo: any
 187   videoPath: string
 188   cwd: string
 189   url: string
 190   user: { username: string, password: string }
 191 }) {
 192   const { youtubeDL, 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(videoInfo.thumbnail, thumbnailfile)
 209   }
 210
 211   const originallyPublishedAt = youtubeDL.buildOriginallyPublishedAt(videoInfo)
 212
 213   const defaultAttributes = {
 214     name: truncate(videoInfo.title, {
 215       length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
 216       separator: /,? +/,
 217       omission: ' […]'
 218     }),
 219     category,
 220     licence,
 221     nsfw: isNSFW(videoInfo),
 222     description: videoInfo.description,
 223     tags
 224   }
 225
 226   const videoAttributes = await buildVideoAttributesFromCommander(url, program, defaultAttributes)
 227
 228   Object.assign(videoAttributes, {
 229     originallyPublishedAt: originallyPublishedAt ? originallyPublishedAt.toISOString() : null,
 230     thumbnailfile,
 231     previewfile: thumbnailfile,
 232     fixture: videoPath
 233   })
 234
 235   log.info('\nUploading on PeerTube video "%s".', videoAttributes.name)
 236
 237   let accessToken = await getAccessTokenOrDie(url, user)
 238
 239   try {
 240     await uploadVideo(url, accessToken, videoAttributes)
 241   } catch (err) {
 242     if (err.message.indexOf('401') !== -1) {
 243       log.info('Got 401 Unauthorized, token may have expired, renewing token and retry.')
 244
 245       accessToken = await getAccessTokenOrDie(url, user)
 246
 247       await uploadVideo(url, accessToken, videoAttributes)
 248     } else {
 249       exitError(err.message)
 250     }
 251   }
 252
 253   await remove(videoPath)
 254   if (thumbnailfile) await remove(thumbnailfile)
 255
 256   log.warn('Uploaded video "%s"!\n', videoAttributes.name)
 257 }
 258
 259 /* ---------------------------------------------------------- */
 260
 261 async function getCategory (categories: string[], url: string) {
 262   if (!categories) return undefined
 263
 264   const categoryString = categories[0]
 265
 266   if (categoryString === 'News & Politics') return 11
 267
 268   const res = await getVideoCategories(url)
 269   const categoriesServer = res.body
 270
 271   for (const key of Object.keys(categoriesServer)) {
 272     const categoryServer = categoriesServer[key]
 273     if (categoryString.toLowerCase() === categoryServer.toLowerCase()) return parseInt(key, 10)
 274   }
 275
 276   return undefined
 277 }
 278
 279 function getLicence (licence: string) {
 280   if (!licence) return undefined
 281
 282   if (licence.includes('Creative Commons Attribution licence')) return 1
 283
 284   return undefined
 285 }
 286
 287 function normalizeObject (obj: any) {
 288   const newObj: any = {}
 289
 290   for (const key of Object.keys(obj)) {
 291     // Deprecated key
 292     if (key === 'resolution') continue
 293
 294     const value = obj[key]
 295
 296     if (typeof value === 'string') {
 297       newObj[key] = value.normalize()
 298     } else {
 299       newObj[key] = value
 300     }
 301   }
 302
 303   return newObj
 304 }
 305
 306 function fetchObject (info: any) {
 307   const url = buildUrl(info)
 308
 309   return new Promise<any>(async (res, rej) => {
 310     const youtubeDL = await YoutubeDL.safeGetYoutubeDL()
 311     youtubeDL.getInfo(url, undefined, processOptions, (err, videoInfo) => {
 312       if (err) return rej(err)
 313
 314       const videoInfoWithUrl = Object.assign(videoInfo, { url })
 315       return res(normalizeObject(videoInfoWithUrl))
 316     })
 317   })
 318 }
 319
 320 function buildUrl (info: any) {
 321   const webpageUrl = info.webpage_url as string
 322   if (webpageUrl?.match(/^https?:\/\//)) return webpageUrl
 323
 324   const url = info.url as string
 325   if (url?.match(/^https?:\/\//)) return url
 326
 327   // It seems youtube-dl does not return the video url
 328   return 'https://www.youtube.com/watch?v=' + info.id
 329 }
 330
 331 function isNSFW (info: any) {
 332   return info.age_limit && info.age_limit >= 16
 333 }
 334
 335 function normalizeTargetUrl (url: string) {
 336   let normalizedUrl = url.replace(/\/+$/, '')
 337
 338   if (!normalizedUrl.startsWith('http://') && !normalizedUrl.startsWith('https://')) {
 339     normalizedUrl = 'https://' + normalizedUrl
 340   }
 341
 342   return normalizedUrl
 343 }
 344
 345 async function promptPassword () {
 346   return new Promise<string>((res, rej) => {
 347     prompt.start()
 348     const schema = {
 349       properties: {
 350         password: {
 351           hidden: true,
 352           required: true
 353         }
 354       }
 355     }
 356     prompt.get(schema, function (err, result) {
 357       if (err) {
 358         return rej(err)
 359       }
 360       return res(result.password)
 361     })
 362   })
 363 }
 364
 365 async function getAccessTokenOrDie (url: string, user: UserInfo) {
 366   const resClient = await getClient(url)
 367   const client = {
 368     id: resClient.body.client_id,
 369     secret: resClient.body.client_secret
 370   }
 371
 372   try {
 373     const res = await login(url, client, user)
 374     return res.body.access_token
 375   } catch (err) {
 376     exitError('Cannot authenticate. Please check your username/password.')
 377   }
 378 }
 379
 380 function parseDate (dateAsStr: string): Date {
 381   if (!/\d{4}-\d{2}-\d{2}/.test(dateAsStr)) {
 382     exitError(`Invalid date passed: ${dateAsStr}. Expected format: YYYY-MM-DD. See help for usage.`)
 383   }
 384   const date = new Date(dateAsStr)
 385   date.setHours(0, 0, 0)
 386   if (isNaN(date.getTime())) {
 387     exitError(`Invalid date passed: ${dateAsStr}. See help for usage.`)
 388   }
 389   return date
 390 }
 391
 392 function formatDate (date: Date): string {
 393   return date.toISOString().split('T')[0]
 394 }
 395
 396 function convertIntoMs (secondsAsStr: string): number {
 397   const seconds = parseInt(secondsAsStr, 10)
 398   if (seconds <= 0) {
 399     exitError(`Invalid duration passed: ${seconds}. Expected duration to be strictly positive and in seconds`)
 400   }
 401   return Math.round(seconds * 1000)
 402 }
 403
 404 function exitError (message: string, ...meta: any[]) {
 405   // use console.error instead of log.error here
 406   console.error(message, ...meta)
 407   process.exit(-1)
 408 }
 409
 410 function getYoutubeDLInfo (youtubeDL: any, url: string, args: string[]) {
 411   return new Promise<any>((res, rej) => {
 412     const options = [ '-j', '--flat-playlist', '--playlist-reverse', ...args ]
 413
 414     youtubeDL.getInfo(url, options, processOptions, async (err, info) => {
 415       if (err) return rej(err)
 416
 417       return res(info)
 418     })
 419   })
 420 }