Rewrite youtube-dl import

Use python3 binary
Allows to use a custom youtube-dl release URL
Allows to use yt-dlp (youtube-dl fork)
Remove proxy config from configuration to use HTTP_PROXY and HTTPS_PROXY
env variables

4 files changed, 490 insertions, 0 deletions

diff --git a/server/helpers/youtube-dl/index.ts b/server/helpers/youtube-dl/index.ts
new file mode 100644
index 000000000..6afc77dcf
--- /dev/null
+++ b/server/helpers/youtube-dl/index.ts
@@ -0,0 +1,3 @@
+export * from './youtube-dl-cli'
+export * from './youtube-dl-info-builder'
+export * from './youtube-dl-wrapper'
diff --git a/server/helpers/youtube-dl/youtube-dl-cli.ts b/server/helpers/youtube-dl/youtube-dl-cli.ts
new file mode 100644
index 000000000..440869205
--- /dev/null
+++ b/server/helpers/youtube-dl/youtube-dl-cli.ts
@@ -0,0 +1,198 @@
+import execa from 'execa'
+import { pathExists, writeFile } from 'fs-extra'
+import { join } from 'path'
+import { CONFIG } from '@server/initializers/config'
+import { VideoResolution } from '@shared/models'
+import { logger, loggerTagsFactory } from '../logger'
+import { getProxy, isProxyEnabled } from '../proxy'
+import { isBinaryResponse, peertubeGot } from '../requests'
+
+const lTags = loggerTagsFactory('youtube-dl')
+
+const youtubeDLBinaryPath = join(CONFIG.STORAGE.BIN_DIR, CONFIG.IMPORT.VIDEOS.HTTP.YOUTUBE_DL_RELEASE.NAME)
+
+export class YoutubeDLCLI {
+
+  static async safeGet () {
+    if (!await pathExists(youtubeDLBinaryPath)) {
+      await this.updateYoutubeDLBinary()
+    }
+
+    return new YoutubeDLCLI()
+  }
+
+  static async updateYoutubeDLBinary () {
+    const url = CONFIG.IMPORT.VIDEOS.HTTP.YOUTUBE_DL_RELEASE.URL
+
+    logger.info('Updating youtubeDL binary from %s.', url, lTags())
+
+    const gotOptions = { context: { bodyKBLimit: 20_000 }, responseType: 'buffer' as 'buffer' }
+
+    try {
+      let gotResult = await peertubeGot(url, gotOptions)
+
+      if (!isBinaryResponse(gotResult)) {
+        const json = JSON.parse(gotResult.body.toString())
+        const latest = json.filter(release => release.prerelease === false)[0]
+        if (!latest) throw new Error('Cannot find latest release')
+
+        const releaseName = CONFIG.IMPORT.VIDEOS.HTTP.YOUTUBE_DL_RELEASE.NAME
+        const releaseAsset = latest.assets.find(a => a.name === releaseName)
+        if (!releaseAsset) throw new Error(`Cannot find appropriate release with name ${releaseName} in release assets`)
+
+        gotResult = await peertubeGot(releaseAsset.browser_download_url, gotOptions)
+      }
+
+      if (!isBinaryResponse(gotResult)) {
+        throw new Error('Not a binary response')
+      }
+
+      await writeFile(youtubeDLBinaryPath, gotResult.body)
+
+      logger.info('youtube-dl updated %s.', youtubeDLBinaryPath, lTags())
+    } catch (err) {
+      logger.error('Cannot update youtube-dl from %s.', url, { err, ...lTags() })
+    }
+  }
+
+  static getYoutubeDLVideoFormat (enabledResolutions: VideoResolution[]) {
+    /**
+     * list of format selectors in order or preference
+     * see https://github.com/ytdl-org/youtube-dl#format-selection
+     *
+     * case #1 asks for a mp4 using h264 (avc1) and the exact resolution in the hope
+     * of being able to do a "quick-transcode"
+     * case #2 is the first fallback. No "quick-transcode" means we can get anything else (like vp9)
+     * case #3 is the resolution-degraded equivalent of #1, and already a pretty safe fallback
+     *
+     * in any case we avoid AV1, see https://github.com/Chocobozzz/PeerTube/issues/3499
+     **/
+    const resolution = enabledResolutions.length === 0
+      ? VideoResolution.H_720P
+      : Math.max(...enabledResolutions)
+
+    return [
+      `bestvideo[vcodec^=avc1][height=${resolution}]+bestaudio[ext=m4a]`, // case #1
+      `bestvideo[vcodec!*=av01][vcodec!*=vp9.2][height=${resolution}]+bestaudio`, // case #2
+      `bestvideo[vcodec^=avc1][height<=${resolution}]+bestaudio[ext=m4a]`, // case #3
+      `bestvideo[vcodec!*=av01][vcodec!*=vp9.2]+bestaudio`,
+      'best[vcodec!*=av01][vcodec!*=vp9.2]', // case fallback for known formats
+      'best' // Ultimate fallback
+    ].join('/')
+  }
+
+  private constructor () {
+
+  }
+
+  download (options: {
+    url: string
+    format: string
+    output: string
+    processOptions: execa.NodeOptions
+    additionalYoutubeDLArgs?: string[]
+  }) {
+    return this.run({
+      url: options.url,
+      processOptions: options.processOptions,
+      args: (options.additionalYoutubeDLArgs || []).concat([ '-f', options.format, '-o', options.output ])
+    })
+  }
+
+  async getInfo (options: {
+    url: string
+    format: string
+    processOptions: execa.NodeOptions
+    additionalYoutubeDLArgs?: string[]
+  }) {
+    const { url, format, additionalYoutubeDLArgs = [], processOptions } = options
+
+    const completeArgs = additionalYoutubeDLArgs.concat([ '--dump-json', '-f', format ])
+
+    const data = await this.run({ url, args: completeArgs, processOptions })
+    const info = data.map(this.parseInfo)
+
+    return info.length === 1
+      ? info[0]
+      : info
+  }
+
+  async getSubs (options: {
+    url: string
+    format: 'vtt'
+    processOptions: execa.NodeOptions
+  }) {
+    const { url, format, processOptions } = options
+
+    const args = [ '--skip-download', '--all-subs', `--sub-format=${format}` ]
+
+    const data = await this.run({ url, args, processOptions })
+    const files: string[] = []
+
+    const skipString = '[info] Writing video subtitles to: '
+
+    for (let i = 0, len = data.length; i < len; i++) {
+      const line = data[i]
+
+      if (line.indexOf(skipString) === 0) {
+        files.push(line.slice(skipString.length))
+      }
+    }
+
+    return files
+  }
+
+  private async run (options: {
+    url: string
+    args: string[]
+    processOptions: execa.NodeOptions
+  }) {
+    const { url, args, processOptions } = options
+
+    let completeArgs = this.wrapWithProxyOptions(args)
+    completeArgs = this.wrapWithIPOptions(completeArgs)
+    completeArgs = this.wrapWithFFmpegOptions(completeArgs)
+
+    const output = await execa('python', [ youtubeDLBinaryPath, ...completeArgs, url ], processOptions)
+
+    logger.debug('Runned youtube-dl command.', { command: output.command, stdout: output.stdout, ...lTags() })
+
+    return output.stdout
+      ? output.stdout.trim().split(/\r?\n/)
+      : undefined
+  }
+
+  private wrapWithProxyOptions (args: string[]) {
+    if (isProxyEnabled()) {
+      logger.debug('Using proxy %s for YoutubeDL', getProxy(), lTags())
+
+      return [ '--proxy', getProxy() ].concat(args)
+    }
+
+    return args
+  }
+
+  private wrapWithIPOptions (args: string[]) {
+    if (CONFIG.IMPORT.VIDEOS.HTTP.FORCE_IPV4) {
+      logger.debug('Force ipv4 for YoutubeDL')
+
+      return [ '--force-ipv4' ].concat(args)
+    }
+
+    return args
+  }
+
+  private wrapWithFFmpegOptions (args: string[]) {
+    if (process.env.FFMPEG_PATH) {
+      logger.debug('Using ffmpeg location %s for YoutubeDL', process.env.FFMPEG_PATH, lTags())
+
+      return [ '--ffmpeg-location', process.env.FFMPEG_PATH ].concat(args)
+    }
+
+    return args
+  }
+
+  private parseInfo (data: string) {
+    return JSON.parse(data)
+  }
+}
diff --git a/server/helpers/youtube-dl/youtube-dl-info-builder.ts b/server/helpers/youtube-dl/youtube-dl-info-builder.ts
new file mode 100644
index 000000000..9746a7067
--- /dev/null
+++ b/server/helpers/youtube-dl/youtube-dl-info-builder.ts
@@ -0,0 +1,154 @@
+import { CONSTRAINTS_FIELDS, VIDEO_CATEGORIES, VIDEO_LANGUAGES, VIDEO_LICENCES } from '../../initializers/constants'
+import { peertubeTruncate } from '../core-utils'
+
+type YoutubeDLInfo = {
+  name?: string
+  description?: string
+  category?: number
+  language?: string
+  licence?: number
+  nsfw?: boolean
+  tags?: string[]
+  thumbnailUrl?: string
+  ext?: string
+  originallyPublishedAt?: Date
+}
+
+class YoutubeDLInfoBuilder {
+  private readonly info: any
+
+  constructor (info: any) {
+    this.info = { ...info }
+  }
+
+  getInfo () {
+    const obj = this.buildVideoInfo(this.normalizeObject(this.info))
+    if (obj.name && obj.name.length < CONSTRAINTS_FIELDS.VIDEOS.NAME.min) obj.name += ' video'
+
+    return obj
+  }
+
+  private normalizeObject (obj: any) {
+    const newObj: any = {}
+
+    for (const key of Object.keys(obj)) {
+      // Deprecated key
+      if (key === 'resolution') continue
+
+      const value = obj[key]
+
+      if (typeof value === 'string') {
+        newObj[key] = value.normalize()
+      } else {
+        newObj[key] = value
+      }
+    }
+
+    return newObj
+  }
+
+  private buildOriginallyPublishedAt (obj: any) {
+    let originallyPublishedAt: Date = null
+
+    const uploadDateMatcher = /^(\d{4})(\d{2})(\d{2})$/.exec(obj.upload_date)
+    if (uploadDateMatcher) {
+      originallyPublishedAt = new Date()
+      originallyPublishedAt.setHours(0, 0, 0, 0)
+
+      const year = parseInt(uploadDateMatcher[1], 10)
+      // Month starts from 0
+      const month = parseInt(uploadDateMatcher[2], 10) - 1
+      const day = parseInt(uploadDateMatcher[3], 10)
+
+      originallyPublishedAt.setFullYear(year, month, day)
+    }
+
+    return originallyPublishedAt
+  }
+
+  private buildVideoInfo (obj: any): YoutubeDLInfo {
+    return {
+      name: this.titleTruncation(obj.title),
+      description: this.descriptionTruncation(obj.description),
+      category: this.getCategory(obj.categories),
+      licence: this.getLicence(obj.license),
+      language: this.getLanguage(obj.language),
+      nsfw: this.isNSFW(obj),
+      tags: this.getTags(obj.tags),
+      thumbnailUrl: obj.thumbnail || undefined,
+      originallyPublishedAt: this.buildOriginallyPublishedAt(obj),
+      ext: obj.ext
+    }
+  }
+
+  private titleTruncation (title: string) {
+    return peertubeTruncate(title, {
+      length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
+      separator: /,? +/,
+      omission: ' […]'
+    })
+  }
+
+  private descriptionTruncation (description: string) {
+    if (!description || description.length < CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.min) return undefined
+
+    return peertubeTruncate(description, {
+      length: CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.max,
+      separator: /,? +/,
+      omission: ' […]'
+    })
+  }
+
+  private isNSFW (info: any) {
+    return info?.age_limit >= 16
+  }
+
+  private getTags (tags: string[]) {
+    if (Array.isArray(tags) === false) return []
+
+    return tags
+      .filter(t => t.length < CONSTRAINTS_FIELDS.VIDEOS.TAG.max && t.length > CONSTRAINTS_FIELDS.VIDEOS.TAG.min)
+      .map(t => t.normalize())
+      .slice(0, 5)
+  }
+
+  private getLicence (licence: string) {
+    if (!licence) return undefined
+
+    if (licence.includes('Creative Commons Attribution')) return 1
+
+    for (const key of Object.keys(VIDEO_LICENCES)) {
+      const peertubeLicence = VIDEO_LICENCES[key]
+      if (peertubeLicence.toLowerCase() === licence.toLowerCase()) return parseInt(key, 10)
+    }
+
+    return undefined
+  }
+
+  private getCategory (categories: string[]) {
+    if (!categories) return undefined
+
+    const categoryString = categories[0]
+    if (!categoryString || typeof categoryString !== 'string') return undefined
+
+    if (categoryString === 'News & Politics') return 11
+
+    for (const key of Object.keys(VIDEO_CATEGORIES)) {
+      const category = VIDEO_CATEGORIES[key]
+      if (categoryString.toLowerCase() === category.toLowerCase()) return parseInt(key, 10)
+    }
+
+    return undefined
+  }
+
+  private getLanguage (language: string) {
+    return VIDEO_LANGUAGES[language] ? language : undefined
+  }
+}
+
+// ---------------------------------------------------------------------------
+
+export {
+  YoutubeDLInfo,
+  YoutubeDLInfoBuilder
+}
diff --git a/server/helpers/youtube-dl/youtube-dl-wrapper.ts b/server/helpers/youtube-dl/youtube-dl-wrapper.ts
new file mode 100644
index 000000000..6960fbae4
--- /dev/null
+++ b/server/helpers/youtube-dl/youtube-dl-wrapper.ts
@@ -0,0 +1,135 @@
+import { move, pathExists, readdir, remove } from 'fs-extra'
+import { dirname, join } from 'path'
+import { CONFIG } from '@server/initializers/config'
+import { isVideoFileExtnameValid } from '../custom-validators/videos'
+import { logger, loggerTagsFactory } from '../logger'
+import { generateVideoImportTmpPath } from '../utils'
+import { YoutubeDLCLI } from './youtube-dl-cli'
+import { YoutubeDLInfo, YoutubeDLInfoBuilder } from './youtube-dl-info-builder'
+
+const lTags = loggerTagsFactory('youtube-dl')
+
+export type YoutubeDLSubs = {
+  language: string
+  filename: string
+  path: string
+}[]
+
+const processOptions = {
+  maxBuffer: 1024 * 1024 * 10 // 10MB
+}
+
+class YoutubeDLWrapper {
+
+  constructor (private readonly url: string = '', private readonly enabledResolutions: number[] = []) {
+
+  }
+
+  async getInfoForDownload (youtubeDLArgs: string[] = []): Promise<YoutubeDLInfo> {
+    const youtubeDL = await YoutubeDLCLI.safeGet()
+
+    const info = await youtubeDL.getInfo({
+      url: this.url,
+      format: YoutubeDLCLI.getYoutubeDLVideoFormat(this.enabledResolutions),
+      additionalYoutubeDLArgs: youtubeDLArgs,
+      processOptions
+    })
+
+    if (info.is_live === true) throw new Error('Cannot download a live streaming.')
+
+    const infoBuilder = new YoutubeDLInfoBuilder(info)
+
+    return infoBuilder.getInfo()
+  }
+
+  async getSubtitles (): Promise<YoutubeDLSubs> {
+    const cwd = CONFIG.STORAGE.TMP_DIR
+
+    const youtubeDL = await YoutubeDLCLI.safeGet()
+
+    const files = await youtubeDL.getSubs({ url: this.url, format: 'vtt', processOptions: { cwd } })
+    if (!files) return []
+
+    logger.debug('Get subtitles from youtube dl.', { url: this.url, files, ...lTags() })
+
+    const subtitles = files.reduce((acc, filename) => {
+      const matched = filename.match(/\.([a-z]{2})(-[a-z]+)?\.(vtt|ttml)/i)
+      if (!matched || !matched[1]) return acc
+
+      return [
+        ...acc,
+        {
+          language: matched[1],
+          path: join(cwd, filename),
+          filename
+        }
+      ]
+    }, [])
+
+    return subtitles
+  }
+
+  async downloadVideo (fileExt: string, timeout: number): Promise<string> {
+    // Leave empty the extension, youtube-dl will add it
+    const pathWithoutExtension = generateVideoImportTmpPath(this.url, '')
+
+    let timer: NodeJS.Timeout
+
+    logger.info('Importing youtubeDL video %s to %s', this.url, pathWithoutExtension, lTags())
+
+    const youtubeDL = await YoutubeDLCLI.safeGet()
+
+    const timeoutPromise = new Promise<string>((_, rej) => {
+      timer = setTimeout(() => rej(new Error('YoutubeDL download timeout.')), timeout)
+    })
+
+    const downloadPromise = youtubeDL.download({
+      url: this.url,
+      format: YoutubeDLCLI.getYoutubeDLVideoFormat(this.enabledResolutions),
+      output: pathWithoutExtension,
+      processOptions
+    }).then(() => clearTimeout(timer))
+      .then(async () => {
+        // If youtube-dl did not guess an extension for our file, just use .mp4 as default
+        if (await pathExists(pathWithoutExtension)) {
+          await move(pathWithoutExtension, pathWithoutExtension + '.mp4')
+        }
+
+        return this.guessVideoPathWithExtension(pathWithoutExtension, fileExt)
+      })
+
+    return Promise.race([ downloadPromise, timeoutPromise ])
+      .catch(async err => {
+        const path = await this.guessVideoPathWithExtension(pathWithoutExtension, fileExt)
+
+        remove(path)
+          .catch(err => logger.error('Cannot remove file in youtubeDL timeout.', { err, ...lTags() }))
+
+        throw err
+      })
+  }
+
+  private async guessVideoPathWithExtension (tmpPath: string, sourceExt: string) {
+    if (!isVideoFileExtnameValid(sourceExt)) {
+      throw new Error('Invalid video extension ' + sourceExt)
+    }
+
+    const extensions = [ sourceExt, '.mp4', '.mkv', '.webm' ]
+
+    for (const extension of extensions) {
+      const path = tmpPath + extension
+
+      if (await pathExists(path)) return path
+    }
+
+    const directoryContent = await readdir(dirname(tmpPath))
+
+    throw new Error(`Cannot guess path of ${tmpPath}. Directory content: ${directoryContent.join(', ')}`)
+  }
+}
+
+// ---------------------------------------------------------------------------
+
+export {
+  YoutubeDLWrapper
+}