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 { promisify } from 'util'
  10 import { YoutubeDL } from '@server/helpers/youtube-dl'
  11 import { sha256 } from '../helpers/core-utils'
  12 import { doRequestAndSaveToFile } from '../helpers/requests'
  13 import { CONSTRAINTS_FIELDS } from '../initializers/constants'
  14 import {
  15   assignToken,
  16   buildCommonVideoOptions,
  17   buildServer,
  18   buildVideoAttributesFromCommander,
  19   getLogger,
  20   getServerCredentials
  21 } from './cli'
  22 import { PeerTubeServer } from '@shared/extra-utils'
  23
  24 import prompt = require('prompt')
  25
  26 const processOptions = {
  27   maxBuffer: Infinity
  28 }
  29
  30 let command = program
  31   .name('import-videos')
  32
  33 command = buildCommonVideoOptions(command)
  34
  35 command
  36   .option('-u, --url <url>', 'Server url')
  37   .option('-U, --username <username>', 'Username')
  38   .option('-p, --password <token>', 'Password')
  39   .option('--target-url <targetUrl>', 'Video target URL')
  40   .option('--since <since>', 'Publication date (inclusive) since which the videos can be imported (YYYY-MM-DD)', parseDate)
  41   .option('--until <until>', 'Publication date (inclusive) until which the videos can be imported (YYYY-MM-DD)', parseDate)
  42   .option('--first <first>', 'Process first n elements of returned playlist')
  43   .option('--last <last>', 'Process last n elements of returned playlist')
  44   .option('--wait-interval <waitInterval>', 'Duration between two video imports (in seconds)', convertIntoMs)
  45   .option('-T, --tmpdir <tmpdir>', 'Working directory', __dirname)
  46   .usage("[global options] [ -- youtube-dl options]")
  47   .parse(process.argv)
  48
  49 const options = command.opts()
  50
  51 const log = getLogger(options.verbose)
  52
  53 getServerCredentials(command)
  54   .then(({ url, username, password }) => {
  55     if (!options.targetUrl) {
  56       exitError('--target-url field is required.')
  57     }
  58
  59     try {
  60       accessSync(options.tmpdir, constants.R_OK | constants.W_OK)
  61     } catch (e) {
  62       exitError('--tmpdir %s: directory does not exist or is not accessible', options.tmpdir)
  63     }
  64
  65     url = normalizeTargetUrl(url)
  66     options.targetUrl = normalizeTargetUrl(options.targetUrl)
  67
  68     run(url, username, password)
  69       .catch(err => exitError(err))
  70   })
  71   .catch(err => console.error(err))
  72
  73 async function run (url: string, username: string, password: string) {
  74   if (!password) password = await promptPassword()
  75
  76   const youtubeDLBinary = await YoutubeDL.safeGetYoutubeDL()
  77
  78   let info = await getYoutubeDLInfo(youtubeDLBinary, options.targetUrl, command.args)
  79
  80   if (!Array.isArray(info)) info = [ info ]
  81
  82   // Try to fix youtube channels upload
  83   const uploadsObject = info.find(i => !i.ie_key && !i.duration && i.title === 'Uploads')
  84
  85   if (uploadsObject) {
  86     console.log('Fixing URL to %s.', uploadsObject.url)
  87
  88     info = await getYoutubeDLInfo(youtubeDLBinary, uploadsObject.url, command.args)
  89   }
  90
  91   let infoArray: any[]
  92
  93   infoArray = [].concat(info)
  94   if (options.first) {
  95     infoArray = infoArray.slice(0, options.first)
  96   } else if (options.last) {
  97     infoArray = infoArray.slice(-options.last)
  98   }
  99   // Normalize utf8 fields
 100   infoArray = infoArray.map(i => normalizeObject(i))
 101
 102   log.info('Will download and upload %d videos.\n', infoArray.length)
 103
 104   for (const [ index, info ] of infoArray.entries()) {
 105     try {
 106       if (index > 0 && options.waitInterval) {
 107         log.info("Wait for %d seconds before continuing.", options.waitInterval / 1000)
 108         await new Promise(res => setTimeout(res, options.waitInterval))
 109       }
 110       await processVideo({
 111         cwd: options.tmpdir,
 112         url,
 113         username,
 114         password,
 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', username, options.targetUrl)
 123   process.exit(0)
 124 }
 125
 126 async function processVideo (parameters: {
 127   cwd: string
 128   url: string
 129   username: string
 130   password: string
 131   youtubeInfo: any
 132 }) {
 133   const { youtubeInfo, cwd, url, username, password } = parameters
 134   const youtubeDL = new YoutubeDL('', [])
 135
 136   log.debug('Fetching object.', youtubeInfo)
 137
 138   const videoInfo = await fetchObject(youtubeInfo)
 139   log.debug('Fetched object.', videoInfo)
 140
 141   const originallyPublishedAt = youtubeDL.buildOriginallyPublishedAt(videoInfo)
 142
 143   if (options.since && originallyPublishedAt && originallyPublishedAt.getTime() < options.since.getTime()) {
 144     log.info('Video "%s" has been published before "%s", don\'t upload it.\n', videoInfo.title, formatDate(options.since))
 145     return
 146   }
 147
 148   if (options.until && originallyPublishedAt && originallyPublishedAt.getTime() > options.until.getTime()) {
 149     log.info('Video "%s" has been published after "%s", don\'t upload it.\n', videoInfo.title, formatDate(options.until))
 150     return
 151   }
 152
 153   const server = buildServer(url)
 154   const { data } = await server.search.advancedVideoSearch({
 155     search: {
 156       search: videoInfo.title,
 157       sort: '-match',
 158       searchTarget: 'local'
 159     }
 160   })
 161
 162   log.info('############################################################\n')
 163
 164   if (data.find(v => v.name === videoInfo.title)) {
 165     log.info('Video "%s" already exists, don\'t reupload it.\n', videoInfo.title)
 166     return
 167   }
 168
 169   const path = join(cwd, sha256(videoInfo.url) + '.mp4')
 170
 171   log.info('Downloading video "%s"...', videoInfo.title)
 172
 173   const youtubeDLOptions = [ '-f', youtubeDL.getYoutubeDLVideoFormat(), ...command.args, '-o', path ]
 174   try {
 175     const youtubeDLBinary = await YoutubeDL.safeGetYoutubeDL()
 176     const youtubeDLExec = promisify(youtubeDLBinary.exec).bind(youtubeDLBinary)
 177     const output = await youtubeDLExec(videoInfo.url, youtubeDLOptions, processOptions)
 178     log.info(output.join('\n'))
 179     await uploadVideoOnPeerTube({
 180       youtubeDL,
 181       cwd,
 182       url,
 183       username,
 184       password,
 185       videoInfo: normalizeObject(videoInfo),
 186       videoPath: path
 187     })
 188   } catch (err) {
 189     log.error(err.message)
 190   }
 191 }
 192
 193 async function uploadVideoOnPeerTube (parameters: {
 194   youtubeDL: YoutubeDL
 195   videoInfo: any
 196   videoPath: string
 197   cwd: string
 198   url: string
 199   username: string
 200   password: string
 201 }) {
 202   const { youtubeDL, videoInfo, videoPath, cwd, url, username, password } = parameters
 203
 204   const server = buildServer(url)
 205   await assignToken(server, username, password)
 206
 207   const category = await getCategory(server, videoInfo.categories)
 208   const licence = getLicence(videoInfo.license)
 209   let tags = []
 210   if (Array.isArray(videoInfo.tags)) {
 211     tags = videoInfo.tags
 212                     .filter(t => t.length < CONSTRAINTS_FIELDS.VIDEOS.TAG.max && t.length > CONSTRAINTS_FIELDS.VIDEOS.TAG.min)
 213                     .map(t => t.normalize())
 214                     .slice(0, 5)
 215   }
 216
 217   let thumbnailfile
 218   if (videoInfo.thumbnail) {
 219     thumbnailfile = join(cwd, sha256(videoInfo.thumbnail) + '.jpg')
 220
 221     await doRequestAndSaveToFile(videoInfo.thumbnail, thumbnailfile)
 222   }
 223
 224   const originallyPublishedAt = youtubeDL.buildOriginallyPublishedAt(videoInfo)
 225
 226   const defaultAttributes = {
 227     name: truncate(videoInfo.title, {
 228       length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
 229       separator: /,? +/,
 230       omission: ' […]'
 231     }),
 232     category,
 233     licence,
 234     nsfw: isNSFW(videoInfo),
 235     description: videoInfo.description,
 236     tags
 237   }
 238
 239   const baseAttributes = await buildVideoAttributesFromCommander(server, program, defaultAttributes)
 240
 241   const attributes = {
 242     ...baseAttributes,
 243
 244     originallyPublishedAt: originallyPublishedAt ? originallyPublishedAt.toISOString() : null,
 245     thumbnailfile,
 246     previewfile: thumbnailfile,
 247     fixture: videoPath
 248   }
 249
 250   log.info('\nUploading on PeerTube video "%s".', attributes.name)
 251
 252   try {
 253     await server.videos.upload({ attributes })
 254   } catch (err) {
 255     if (err.message.indexOf('401') !== -1) {
 256       log.info('Got 401 Unauthorized, token may have expired, renewing token and retry.')
 257
 258       server.accessToken = await server.login.getAccessToken(username, password)
 259
 260       await server.videos.upload({ attributes })
 261     } else {
 262       exitError(err.message)
 263     }
 264   }
 265
 266   await remove(videoPath)
 267   if (thumbnailfile) await remove(thumbnailfile)
 268
 269   log.warn('Uploaded video "%s"!\n', attributes.name)
 270 }
 271
 272 /* ---------------------------------------------------------- */
 273
 274 async function getCategory (server: PeerTubeServer, categories: string[]) {
 275   if (!categories) return undefined
 276
 277   const categoryString = categories[0]
 278
 279   if (categoryString === 'News & Politics') return 11
 280
 281   const categoriesServer = await server.videos.getCategories()
 282
 283   for (const key of Object.keys(categoriesServer)) {
 284     const categoryServer = categoriesServer[key]
 285     if (categoryString.toLowerCase() === categoryServer.toLowerCase()) return parseInt(key, 10)
 286   }
 287
 288   return undefined
 289 }
 290
 291 function getLicence (licence: string) {
 292   if (!licence) return undefined
 293
 294   if (licence.includes('Creative Commons Attribution licence')) return 1
 295
 296   return undefined
 297 }
 298
 299 function normalizeObject (obj: any) {
 300   const newObj: any = {}
 301
 302   for (const key of Object.keys(obj)) {
 303     // Deprecated key
 304     if (key === 'resolution') continue
 305
 306     const value = obj[key]
 307
 308     if (typeof value === 'string') {
 309       newObj[key] = value.normalize()
 310     } else {
 311       newObj[key] = value
 312     }
 313   }
 314
 315   return newObj
 316 }
 317
 318 function fetchObject (info: any) {
 319   const url = buildUrl(info)
 320
 321   return new Promise<any>(async (res, rej) => {
 322     const youtubeDL = await YoutubeDL.safeGetYoutubeDL()
 323     youtubeDL.getInfo(url, undefined, processOptions, (err, videoInfo) => {
 324       if (err) return rej(err)
 325
 326       const videoInfoWithUrl = Object.assign(videoInfo, { url })
 327       return res(normalizeObject(videoInfoWithUrl))
 328     })
 329   })
 330 }
 331
 332 function buildUrl (info: any) {
 333   const webpageUrl = info.webpage_url as string
 334   if (webpageUrl?.match(/^https?:\/\//)) return webpageUrl
 335
 336   const url = info.url as string
 337   if (url?.match(/^https?:\/\//)) return url
 338
 339   // It seems youtube-dl does not return the video url
 340   return 'https://www.youtube.com/watch?v=' + info.id
 341 }
 342
 343 function isNSFW (info: any) {
 344   return info.age_limit && info.age_limit >= 16
 345 }
 346
 347 function normalizeTargetUrl (url: string) {
 348   let normalizedUrl = url.replace(/\/+$/, '')
 349
 350   if (!normalizedUrl.startsWith('http://') && !normalizedUrl.startsWith('https://')) {
 351     normalizedUrl = 'https://' + normalizedUrl
 352   }
 353
 354   return normalizedUrl
 355 }
 356
 357 async function promptPassword () {
 358   return new Promise<string>((res, rej) => {
 359     prompt.start()
 360     const schema = {
 361       properties: {
 362         password: {
 363           hidden: true,
 364           required: true
 365         }
 366       }
 367     }
 368     prompt.get(schema, function (err, result) {
 369       if (err) {
 370         return rej(err)
 371       }
 372       return res(result.password)
 373     })
 374   })
 375 }
 376
 377 function parseDate (dateAsStr: string): Date {
 378   if (!/\d{4}-\d{2}-\d{2}/.test(dateAsStr)) {
 379     exitError(`Invalid date passed: ${dateAsStr}. Expected format: YYYY-MM-DD. See help for usage.`)
 380   }
 381   const date = new Date(dateAsStr)
 382   date.setHours(0, 0, 0)
 383   if (isNaN(date.getTime())) {
 384     exitError(`Invalid date passed: ${dateAsStr}. See help for usage.`)
 385   }
 386   return date
 387 }
 388
 389 function formatDate (date: Date): string {
 390   return date.toISOString().split('T')[0]
 391 }
 392
 393 function convertIntoMs (secondsAsStr: string): number {
 394   const seconds = parseInt(secondsAsStr, 10)
 395   if (seconds <= 0) {
 396     exitError(`Invalid duration passed: ${seconds}. Expected duration to be strictly positive and in seconds`)
 397   }
 398   return Math.round(seconds * 1000)
 399 }
 400
 401 function exitError (message: string, ...meta: any[]) {
 402   // use console.error instead of log.error here
 403   console.error(message, ...meta)
 404   process.exit(-1)
 405 }
 406
 407 function getYoutubeDLInfo (youtubeDL: any, url: string, args: string[]) {
 408   return new Promise<any>((res, rej) => {
 409     const options = [ '-j', '--flat-playlist', '--playlist-reverse', ...args ]
 410
 411     youtubeDL.getInfo(url, options, processOptions, (err, info) => {
 412       if (err) return rej(err)
 413
 414       return res(info)
 415     })
 416   })
 417 }