server/tools/peertube-import-videos.ts

   1 import { registerTSPaths } from '../helpers/register-ts-paths'
   2 registerTSPaths()
   3
   4 import { 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 { YoutubeDL } from '@server/helpers/youtube-dl'
  12 import { getVideoCategories, uploadVideo } from '../../shared/extra-utils/index'
  13 import { sha256 } from '../helpers/core-utils'
  14 import { doRequestAndSaveToFile } from '../helpers/requests'
  15 import { CONSTRAINTS_FIELDS } from '../initializers/constants'
  16 import {
  17   buildCommonVideoOptions,
  18   buildServer,
  19   buildVideoAttributesFromCommander,
  20   getAccessTokenOrDie,
  21   getLogger,
  22   getServerCredentials
  23 } from './cli'
  24
  25 const processOptions = {
  26   maxBuffer: Infinity
  27 }
  28
  29 let command = program
  30   .name('import-videos')
  31
  32 command = buildCommonVideoOptions(command)
  33
  34 command
  35   .option('-u, --url <url>', 'Server url')
  36   .option('-U, --username <username>', 'Username')
  37   .option('-p, --password <token>', 'Password')
  38   .option('--target-url <targetUrl>', 'Video target URL')
  39   .option('--since <since>', 'Publication date (inclusive) since which the videos can be imported (YYYY-MM-DD)', parseDate)
  40   .option('--until <until>', 'Publication date (inclusive) until which the videos can be imported (YYYY-MM-DD)', parseDate)
  41   .option('--first <first>', 'Process first n elements of returned playlist')
  42   .option('--last <last>', 'Process last n elements of returned playlist')
  43   .option('--wait-interval <waitInterval>', 'Duration between two video imports (in seconds)', convertIntoMs)
  44   .option('-T, --tmpdir <tmpdir>', 'Working directory', __dirname)
  45   .usage("[global options] [ -- youtube-dl options]")
  46   .parse(process.argv)
  47
  48 const options = command.opts()
  49
  50 const log = getLogger(options.verbose)
  51
  52 getServerCredentials(command)
  53   .then(({ url, username, password }) => {
  54     if (!options.targetUrl) {
  55       exitError('--target-url field is required.')
  56     }
  57
  58     try {
  59       accessSync(options.tmpdir, constants.R_OK | constants.W_OK)
  60     } catch (e) {
  61       exitError('--tmpdir %s: directory does not exist or is not accessible', options.tmpdir)
  62     }
  63
  64     url = normalizeTargetUrl(url)
  65     options.targetUrl = normalizeTargetUrl(options.targetUrl)
  66
  67     run(url, username, password)
  68       .catch(err => exitError(err))
  69   })
  70   .catch(err => console.error(err))
  71
  72 async function run (url: string, username: string, password: string) {
  73   if (!password) password = await promptPassword()
  74
  75   const youtubeDLBinary = await YoutubeDL.safeGetYoutubeDL()
  76
  77   let info = await getYoutubeDLInfo(youtubeDLBinary, options.targetUrl, command.args)
  78
  79   if (!Array.isArray(info)) info = [ info ]
  80
  81   // Try to fix youtube channels upload
  82   const uploadsObject = info.find(i => !i.ie_key && !i.duration && i.title === 'Uploads')
  83
  84   if (uploadsObject) {
  85     console.log('Fixing URL to %s.', uploadsObject.url)
  86
  87     info = await getYoutubeDLInfo(youtubeDLBinary, uploadsObject.url, command.args)
  88   }
  89
  90   let infoArray: any[]
  91
  92   infoArray = [].concat(info)
  93   if (options.first) {
  94     infoArray = infoArray.slice(0, options.first)
  95   } else if (options.last) {
  96     infoArray = infoArray.slice(-options.last)
  97   }
  98   // Normalize utf8 fields
  99   infoArray = infoArray.map(i => normalizeObject(i))
 100
 101   log.info('Will download and upload %d videos.\n', infoArray.length)
 102
 103   for (const [ index, info ] of infoArray.entries()) {
 104     try {
 105       if (index > 0 && options.waitInterval) {
 106         log.info("Wait for %d seconds before continuing.", options.waitInterval / 1000)
 107         await new Promise(res => setTimeout(res, options.waitInterval))
 108       }
 109       await processVideo({
 110         cwd: options.tmpdir,
 111         url,
 112         username,
 113         password,
 114         youtubeInfo: info
 115       })
 116     } catch (err) {
 117       console.error('Cannot process video.', { info, url, err })
 118     }
 119   }
 120
 121   log.info('Video/s for user %s imported: %s', username, options.targetUrl)
 122   process.exit(0)
 123 }
 124
 125 async function processVideo (parameters: {
 126   cwd: string
 127   url: string
 128   username: string
 129   password: string
 130   youtubeInfo: any
 131 }) {
 132   const { youtubeInfo, cwd, url, username, password } = 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
 142   if (options.since && originallyPublishedAt && originallyPublishedAt.getTime() < options.since.getTime()) {
 143     log.info('Video "%s" has been published before "%s", don\'t upload it.\n', videoInfo.title, formatDate(options.since))
 144     return
 145   }
 146
 147   if (options.until && originallyPublishedAt && originallyPublishedAt.getTime() > options.until.getTime()) {
 148     log.info('Video "%s" has been published after "%s", don\'t upload it.\n', videoInfo.title, formatDate(options.until))
 149     return
 150   }
 151
 152   const server = buildServer(url)
 153   const { data } = await server.searchCommand.advancedVideoSearch({
 154     search: {
 155       search: videoInfo.title,
 156       sort: '-match',
 157       searchTarget: 'local'
 158     }
 159   })
 160
 161   log.info('############################################################\n')
 162
 163   if (data.find(v => v.name === videoInfo.title)) {
 164     log.info('Video "%s" already exists, don\'t reupload it.\n', videoInfo.title)
 165     return
 166   }
 167
 168   const path = join(cwd, sha256(videoInfo.url) + '.mp4')
 169
 170   log.info('Downloading video "%s"...', videoInfo.title)
 171
 172   const youtubeDLOptions = [ '-f', youtubeDL.getYoutubeDLVideoFormat(), ...command.args, '-o', path ]
 173   try {
 174     const youtubeDLBinary = await YoutubeDL.safeGetYoutubeDL()
 175     const youtubeDLExec = promisify(youtubeDLBinary.exec).bind(youtubeDLBinary)
 176     const output = await youtubeDLExec(videoInfo.url, youtubeDLOptions, processOptions)
 177     log.info(output.join('\n'))
 178     await uploadVideoOnPeerTube({
 179       youtubeDL,
 180       cwd,
 181       url,
 182       username,
 183       password,
 184       videoInfo: normalizeObject(videoInfo),
 185       videoPath: path
 186     })
 187   } catch (err) {
 188     log.error(err.message)
 189   }
 190 }
 191
 192 async function uploadVideoOnPeerTube (parameters: {
 193   youtubeDL: YoutubeDL
 194   videoInfo: any
 195   videoPath: string
 196   cwd: string
 197   url: string
 198   username: string
 199   password: string
 200 }) {
 201   const { youtubeDL, videoInfo, videoPath, cwd, url, username, password } = parameters
 202
 203   const category = await getCategory(videoInfo.categories, url)
 204   const licence = getLicence(videoInfo.license)
 205   let tags = []
 206   if (Array.isArray(videoInfo.tags)) {
 207     tags = videoInfo.tags
 208                     .filter(t => t.length < CONSTRAINTS_FIELDS.VIDEOS.TAG.max && t.length > CONSTRAINTS_FIELDS.VIDEOS.TAG.min)
 209                     .map(t => t.normalize())
 210                     .slice(0, 5)
 211   }
 212
 213   let thumbnailfile
 214   if (videoInfo.thumbnail) {
 215     thumbnailfile = join(cwd, sha256(videoInfo.thumbnail) + '.jpg')
 216
 217     await doRequestAndSaveToFile(videoInfo.thumbnail, thumbnailfile)
 218   }
 219
 220   const originallyPublishedAt = youtubeDL.buildOriginallyPublishedAt(videoInfo)
 221
 222   const defaultAttributes = {
 223     name: truncate(videoInfo.title, {
 224       length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
 225       separator: /,? +/,
 226       omission: ' […]'
 227     }),
 228     category,
 229     licence,
 230     nsfw: isNSFW(videoInfo),
 231     description: videoInfo.description,
 232     tags
 233   }
 234
 235   let accessToken = await getAccessTokenOrDie(url, username, password)
 236   const server = buildServer(url, accessToken)
 237
 238   const videoAttributes = await buildVideoAttributesFromCommander(server, program, defaultAttributes)
 239
 240   Object.assign(videoAttributes, {
 241     originallyPublishedAt: originallyPublishedAt ? originallyPublishedAt.toISOString() : null,
 242     thumbnailfile,
 243     previewfile: thumbnailfile,
 244     fixture: videoPath
 245   })
 246
 247   log.info('\nUploading on PeerTube video "%s".', videoAttributes.name)
 248
 249   try {
 250     await uploadVideo(url, accessToken, videoAttributes)
 251   } catch (err) {
 252     if (err.message.indexOf('401') !== -1) {
 253       log.info('Got 401 Unauthorized, token may have expired, renewing token and retry.')
 254
 255       accessToken = await getAccessTokenOrDie(url, username, password)
 256
 257       await uploadVideo(url, accessToken, videoAttributes)
 258     } else {
 259       exitError(err.message)
 260     }
 261   }
 262
 263   await remove(videoPath)
 264   if (thumbnailfile) await remove(thumbnailfile)
 265
 266   log.warn('Uploaded video "%s"!\n', videoAttributes.name)
 267 }
 268
 269 /* ---------------------------------------------------------- */
 270
 271 async function getCategory (categories: string[], url: string) {
 272   if (!categories) return undefined
 273
 274   const categoryString = categories[0]
 275
 276   if (categoryString === 'News & Politics') return 11
 277
 278   const res = await getVideoCategories(url)
 279   const categoriesServer = res.body
 280
 281   for (const key of Object.keys(categoriesServer)) {
 282     const categoryServer = categoriesServer[key]
 283     if (categoryString.toLowerCase() === categoryServer.toLowerCase()) return parseInt(key, 10)
 284   }
 285
 286   return undefined
 287 }
 288
 289 function getLicence (licence: string) {
 290   if (!licence) return undefined
 291
 292   if (licence.includes('Creative Commons Attribution licence')) return 1
 293
 294   return undefined
 295 }
 296
 297 function normalizeObject (obj: any) {
 298   const newObj: any = {}
 299
 300   for (const key of Object.keys(obj)) {
 301     // Deprecated key
 302     if (key === 'resolution') continue
 303
 304     const value = obj[key]
 305
 306     if (typeof value === 'string') {
 307       newObj[key] = value.normalize()
 308     } else {
 309       newObj[key] = value
 310     }
 311   }
 312
 313   return newObj
 314 }
 315
 316 function fetchObject (info: any) {
 317   const url = buildUrl(info)
 318
 319   return new Promise<any>(async (res, rej) => {
 320     const youtubeDL = await YoutubeDL.safeGetYoutubeDL()
 321     youtubeDL.getInfo(url, undefined, processOptions, (err, videoInfo) => {
 322       if (err) return rej(err)
 323
 324       const videoInfoWithUrl = Object.assign(videoInfo, { url })
 325       return res(normalizeObject(videoInfoWithUrl))
 326     })
 327   })
 328 }
 329
 330 function buildUrl (info: any) {
 331   const webpageUrl = info.webpage_url as string
 332   if (webpageUrl?.match(/^https?:\/\//)) return webpageUrl
 333
 334   const url = info.url as string
 335   if (url?.match(/^https?:\/\//)) return url
 336
 337   // It seems youtube-dl does not return the video url
 338   return 'https://www.youtube.com/watch?v=' + info.id
 339 }
 340
 341 function isNSFW (info: any) {
 342   return info.age_limit && info.age_limit >= 16
 343 }
 344
 345 function normalizeTargetUrl (url: string) {
 346   let normalizedUrl = url.replace(/\/+$/, '')
 347
 348   if (!normalizedUrl.startsWith('http://') && !normalizedUrl.startsWith('https://')) {
 349     normalizedUrl = 'https://' + normalizedUrl
 350   }
 351
 352   return normalizedUrl
 353 }
 354
 355 async function promptPassword () {
 356   return new Promise<string>((res, rej) => {
 357     prompt.start()
 358     const schema = {
 359       properties: {
 360         password: {
 361           hidden: true,
 362           required: true
 363         }
 364       }
 365     }
 366     prompt.get(schema, function (err, result) {
 367       if (err) {
 368         return rej(err)
 369       }
 370       return res(result.password)
 371     })
 372   })
 373 }
 374
 375 function parseDate (dateAsStr: string): Date {
 376   if (!/\d{4}-\d{2}-\d{2}/.test(dateAsStr)) {
 377     exitError(`Invalid date passed: ${dateAsStr}. Expected format: YYYY-MM-DD. See help for usage.`)
 378   }
 379   const date = new Date(dateAsStr)
 380   date.setHours(0, 0, 0)
 381   if (isNaN(date.getTime())) {
 382     exitError(`Invalid date passed: ${dateAsStr}. See help for usage.`)
 383   }
 384   return date
 385 }
 386
 387 function formatDate (date: Date): string {
 388   return date.toISOString().split('T')[0]
 389 }
 390
 391 function convertIntoMs (secondsAsStr: string): number {
 392   const seconds = parseInt(secondsAsStr, 10)
 393   if (seconds <= 0) {
 394     exitError(`Invalid duration passed: ${seconds}. Expected duration to be strictly positive and in seconds`)
 395   }
 396   return Math.round(seconds * 1000)
 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 }