server/tools/peertube-import-videos.ts

   1 import { registerTSPaths } from '../helpers/register-ts-paths'
   2 registerTSPaths()
   3
   4 import { accessSync, constants } from 'fs'
   5 import { join } from 'path'
   6 import { promisify } from 'util'
   7
   8 import * as program from 'commander'
   9 import { remove } from 'fs-extra'
  10 import { truncate } from 'lodash'
  11 import * as prompt from 'prompt'
  12
  13 import { doRequestAndSaveToFile } from '../helpers/requests'
  14 import { CONSTRAINTS_FIELDS } from '../initializers/constants'
  15 import { getClient, getVideoCategories, login, searchVideoWithSort, uploadVideo } from '../../shared/extra-utils/index'
  16 import { sha256 } from '../helpers/core-utils'
  17 import { buildOriginallyPublishedAt, getYoutubeDLVideoFormat, safeGetYoutubeDL } from '../helpers/youtube-dl'
  18 import { buildCommonVideoOptions, buildVideoAttributesFromCommander, getLogger, getServerCredentials } from './cli'
  19
  20 type UserInfo = {
  21   username: string
  22   password: string
  23 }
  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     const user = { username, password }
  68
  69     run(url, user)
  70       .catch(err => exitError(err))
  71   })
  72   .catch(err => console.error(err))
  73
  74 async function run (url: string, user: UserInfo) {
  75   if (!user.password) {
  76     user.password = await promptPassword()
  77   }
  78
  79   const youtubeDL = await safeGetYoutubeDL()
  80
  81   let info = await getYoutubeDLInfo(youtubeDL, options.targetUrl, command.args)
  82
  83   if (!Array.isArray(info)) info = [ info ]
  84
  85   // Try to fix youtube channels upload
  86   const uploadsObject = info.find(i => !i.ie_key && !i.duration && i.title === 'Uploads')
  87
  88   if (uploadsObject) {
  89     console.log('Fixing URL to %s.', uploadsObject.url)
  90
  91     info = await getYoutubeDLInfo(youtubeDL, uploadsObject.url, command.args)
  92   }
  93
  94   let infoArray: any[]
  95
  96   infoArray = [].concat(info)
  97   if (options.first) {
  98     infoArray = infoArray.slice(0, options.first)
  99   } else if (options.last) {
 100     infoArray = infoArray.slice(-options.last)
 101   }
 102   // Normalize utf8 fields
 103   infoArray = infoArray.map(i => normalizeObject(i))
 104
 105   log.info('Will download and upload %d videos.\n', infoArray.length)
 106
 107   for (const [ index, info ] of infoArray.entries()) {
 108     try {
 109       if (index > 0 && options.waitInterval) {
 110         log.info("Wait for %d seconds before continuing.", options.waitInterval / 1000)
 111         await new Promise(res => setTimeout(res, options.waitInterval))
 112       }
 113       await processVideo({
 114         cwd: options.tmpdir,
 115         url,
 116         user,
 117         youtubeInfo: info
 118       })
 119     } catch (err) {
 120       console.error('Cannot process video.', { info, url, err })
 121     }
 122   }
 123
 124   log.info('Video/s for user %s imported: %s', user.username, options.targetUrl)
 125   process.exit(0)
 126 }
 127
 128 async function processVideo (parameters: {
 129   cwd: string
 130   url: string
 131   user: { username: string, password: string }
 132   youtubeInfo: any
 133 }) {
 134   const { youtubeInfo, cwd, url, user } = parameters
 135
 136   log.debug('Fetching object.', youtubeInfo)
 137
 138   const videoInfo = await fetchObject(youtubeInfo)
 139   log.debug('Fetched object.', videoInfo)
 140
 141   const originallyPublishedAt = buildOriginallyPublishedAt(videoInfo)
 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',
 144       videoInfo.title, formatDate(options.since))
 145     return
 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',
 149       videoInfo.title, formatDate(options.until))
 150     return
 151   }
 152
 153   const result = await searchVideoWithSort(url, videoInfo.title, '-match')
 154
 155   log.info('############################################################\n')
 156
 157   if (result.body.data.find(v => v.name === videoInfo.title)) {
 158     log.info('Video "%s" already exists, don\'t reupload it.\n', videoInfo.title)
 159     return
 160   }
 161
 162   const path = join(cwd, sha256(videoInfo.url) + '.mp4')
 163
 164   log.info('Downloading video "%s"...', videoInfo.title)
 165
 166   const youtubeDLOptions = [ '-f', getYoutubeDLVideoFormat(), ...command.args, '-o', path ]
 167   try {
 168     const youtubeDL = await safeGetYoutubeDL()
 169     const youtubeDLExec = promisify(youtubeDL.exec).bind(youtubeDL)
 170     const output = await youtubeDLExec(videoInfo.url, youtubeDLOptions, processOptions)
 171     log.info(output.join('\n'))
 172     await uploadVideoOnPeerTube({
 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   videoInfo: any
 186   videoPath: string
 187   cwd: string
 188   url: string
 189   user: { username: string, password: string }
 190 }) {
 191   const { videoInfo, videoPath, cwd, url, user } = parameters
 192
 193   const category = await getCategory(videoInfo.categories, url)
 194   const licence = getLicence(videoInfo.license)
 195   let tags = []
 196   if (Array.isArray(videoInfo.tags)) {
 197     tags = videoInfo.tags
 198                     .filter(t => t.length < CONSTRAINTS_FIELDS.VIDEOS.TAG.max && t.length > CONSTRAINTS_FIELDS.VIDEOS.TAG.min)
 199                     .map(t => t.normalize())
 200                     .slice(0, 5)
 201   }
 202
 203   let thumbnailfile
 204   if (videoInfo.thumbnail) {
 205     thumbnailfile = join(cwd, sha256(videoInfo.thumbnail) + '.jpg')
 206
 207     await doRequestAndSaveToFile({
 208       method: 'GET',
 209       uri: videoInfo.thumbnail
 210     }, thumbnailfile)
 211   }
 212
 213   const originallyPublishedAt = buildOriginallyPublishedAt(videoInfo)
 214
 215   const defaultAttributes = {
 216     name: truncate(videoInfo.title, {
 217       length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
 218       separator: /,? +/,
 219       omission: ' […]'
 220     }),
 221     category,
 222     licence,
 223     nsfw: isNSFW(videoInfo),
 224     description: videoInfo.description,
 225     tags
 226   }
 227
 228   const videoAttributes = await buildVideoAttributesFromCommander(url, program, defaultAttributes)
 229
 230   Object.assign(videoAttributes, {
 231     originallyPublishedAt: originallyPublishedAt ? originallyPublishedAt.toISOString() : null,
 232     thumbnailfile,
 233     previewfile: thumbnailfile,
 234     fixture: videoPath
 235   })
 236
 237   log.info('\nUploading on PeerTube video "%s".', videoAttributes.name)
 238
 239   let accessToken = await getAccessTokenOrDie(url, user)
 240
 241   try {
 242     await uploadVideo(url, accessToken, videoAttributes)
 243   } catch (err) {
 244     if (err.message.indexOf('401') !== -1) {
 245       log.info('Got 401 Unauthorized, token may have expired, renewing token and retry.')
 246
 247       accessToken = await getAccessTokenOrDie(url, user)
 248
 249       await uploadVideo(url, accessToken, videoAttributes)
 250     } else {
 251       exitError(err.message)
 252     }
 253   }
 254
 255   await remove(videoPath)
 256   if (thumbnailfile) await remove(thumbnailfile)
 257
 258   log.warn('Uploaded video "%s"!\n', videoAttributes.name)
 259 }
 260
 261 /* ---------------------------------------------------------- */
 262
 263 async function getCategory (categories: string[], url: string) {
 264   if (!categories) return undefined
 265
 266   const categoryString = categories[0]
 267
 268   if (categoryString === 'News & Politics') return 11
 269
 270   const res = await getVideoCategories(url)
 271   const categoriesServer = res.body
 272
 273   for (const key of Object.keys(categoriesServer)) {
 274     const categoryServer = categoriesServer[key]
 275     if (categoryString.toLowerCase() === categoryServer.toLowerCase()) return parseInt(key, 10)
 276   }
 277
 278   return undefined
 279 }
 280
 281 function getLicence (licence: string) {
 282   if (!licence) return undefined
 283
 284   if (licence.includes('Creative Commons Attribution licence')) return 1
 285
 286   return undefined
 287 }
 288
 289 function normalizeObject (obj: any) {
 290   const newObj: any = {}
 291
 292   for (const key of Object.keys(obj)) {
 293     // Deprecated key
 294     if (key === 'resolution') continue
 295
 296     const value = obj[key]
 297
 298     if (typeof value === 'string') {
 299       newObj[key] = value.normalize()
 300     } else {
 301       newObj[key] = value
 302     }
 303   }
 304
 305   return newObj
 306 }
 307
 308 function fetchObject (info: any) {
 309   const url = buildUrl(info)
 310
 311   return new Promise<any>(async (res, rej) => {
 312     const youtubeDL = await safeGetYoutubeDL()
 313     youtubeDL.getInfo(url, undefined, processOptions, (err, videoInfo) => {
 314       if (err) return rej(err)
 315
 316       const videoInfoWithUrl = Object.assign(videoInfo, { url })
 317       return res(normalizeObject(videoInfoWithUrl))
 318     })
 319   })
 320 }
 321
 322 function buildUrl (info: any) {
 323   const webpageUrl = info.webpage_url as string
 324   if (webpageUrl?.match(/^https?:\/\//)) return webpageUrl
 325
 326   const url = info.url as string
 327   if (url?.match(/^https?:\/\//)) return url
 328
 329   // It seems youtube-dl does not return the video url
 330   return 'https://www.youtube.com/watch?v=' + info.id
 331 }
 332
 333 function isNSFW (info: any) {
 334   return info.age_limit && info.age_limit >= 16
 335 }
 336
 337 function normalizeTargetUrl (url: string) {
 338   let normalizedUrl = url.replace(/\/+$/, '')
 339
 340   if (!normalizedUrl.startsWith('http://') && !normalizedUrl.startsWith('https://')) {
 341     normalizedUrl = 'https://' + normalizedUrl
 342   }
 343
 344   return normalizedUrl
 345 }
 346
 347 async function promptPassword () {
 348   return new Promise<string>((res, rej) => {
 349     prompt.start()
 350     const schema = {
 351       properties: {
 352         password: {
 353           hidden: true,
 354           required: true
 355         }
 356       }
 357     }
 358     prompt.get(schema, function (err, result) {
 359       if (err) {
 360         return rej(err)
 361       }
 362       return res(result.password)
 363     })
 364   })
 365 }
 366
 367 async function getAccessTokenOrDie (url: string, user: UserInfo) {
 368   const resClient = await getClient(url)
 369   const client = {
 370     id: resClient.body.client_id,
 371     secret: resClient.body.client_secret
 372   }
 373
 374   try {
 375     const res = await login(url, client, user)
 376     return res.body.access_token
 377   } catch (err) {
 378     exitError('Cannot authenticate. Please check your username/password.')
 379   }
 380 }
 381
 382 function parseDate (dateAsStr: string): Date {
 383   if (!/\d{4}-\d{2}-\d{2}/.test(dateAsStr)) {
 384     exitError(`Invalid date passed: ${dateAsStr}. Expected format: YYYY-MM-DD. See help for usage.`)
 385   }
 386   const date = new Date(dateAsStr)
 387   date.setHours(0, 0, 0)
 388   if (isNaN(date.getTime())) {
 389     exitError(`Invalid date passed: ${dateAsStr}. See help for usage.`)
 390   }
 391   return date
 392 }
 393
 394 function formatDate (date: Date): string {
 395   return date.toISOString().split('T')[0]
 396 }
 397
 398 function convertIntoMs (secondsAsStr: string): number {
 399   const seconds = parseInt(secondsAsStr, 10)
 400   if (seconds <= 0) {
 401     exitError(`Invalid duration passed: ${seconds}. Expected duration to be strictly positive and in seconds`)
 402   }
 403   return Math.round(seconds * 1000)
 404 }
 405
 406 function exitError (message: string, ...meta: any[]) {
 407   // use console.error instead of log.error here
 408   console.error(message, ...meta)
 409   process.exit(-1)
 410 }
 411
 412 function getYoutubeDLInfo (youtubeDL: any, url: string, args: string[]) {
 413   return new Promise<any>((res, rej) => {
 414     const options = [ '-j', '--flat-playlist', '--playlist-reverse', ...args ]
 415
 416     youtubeDL.getInfo(url, options, processOptions, async (err, info) => {
 417       if (err) return rej(err)
 418
 419       return res(info)
 420     })
 421   })
 422 }