-import * as youtubeDL from 'youtube-dl'
-import { truncate } from 'lodash'
-import { CONSTRAINTS_FIELDS, VIDEO_CATEGORIES } from '../initializers'
+import { createWriteStream } from 'fs'
+import { ensureDir, move, pathExists, remove, writeFile } from 'fs-extra'
+import got from 'got'
+import { join } from 'path'
+import { CONFIG } from '@server/initializers/config'
+import { HttpStatusCode } from '../../shared/core-utils/miscs/http-error-codes'
+import { VideoResolution } from '../../shared/models/videos'
+import { CONSTRAINTS_FIELDS, VIDEO_CATEGORIES, VIDEO_LANGUAGES, VIDEO_LICENCES } from '../initializers/constants'
+import { getEnabledResolutions } from '../lib/video-transcoding'
+import { peertubeTruncate, pipelinePromise, root } from './core-utils'
+import { isVideoFileExtnameValid } from './custom-validators/videos'
import { logger } from './logger'
-import { generateVideoTmpPath } from './utils'
+import { generateVideoImportTmpPath } from './utils'
export type YoutubeDLInfo = {
name?: string
description?: string
category?: number
+ language?: string
licence?: number
nsfw?: boolean
tags?: string[]
thumbnailUrl?: string
+ ext?: string
+ originallyPublishedAt?: Date
}
-function getYoutubeDLInfo (url: string): Promise<YoutubeDLInfo> {
+export type YoutubeDLSubs = {
+ language: string
+ filename: string
+ path: string
+}[]
+
+const processOptions = {
+ maxBuffer: 1024 * 1024 * 10 // 10MB
+}
+
+function getYoutubeDLInfo (url: string, opts?: string[]): Promise<YoutubeDLInfo> {
return new Promise<YoutubeDLInfo>((res, rej) => {
- const options = [ '-j', '--flat-playlist' ]
+ let args = opts || [ '-j', '--flat-playlist' ]
+
+ if (CONFIG.IMPORT.VIDEOS.HTTP.FORCE_IPV4) {
+ args.push('--force-ipv4')
+ }
+
+ args = wrapWithProxyOptions(args)
+ args = [ '-f', getYoutubeDLVideoFormat() ].concat(args)
- youtubeDL.getInfo(url, options, (err, info) => {
- if (err) return rej(err)
+ safeGetYoutubeDL()
+ .then(youtubeDL => {
+ youtubeDL.getInfo(url, args, processOptions, (err, info) => {
+ if (err) return rej(err)
+ if (info.is_live === true) return rej(new Error('Cannot download a live streaming.'))
- const obj = normalizeObject(info)
+ const obj = buildVideoInfo(normalizeObject(info))
+ if (obj.name && obj.name.length < CONSTRAINTS_FIELDS.VIDEOS.NAME.min) obj.name += ' video'
- return res(buildVideoInfo(obj))
- })
+ return res(obj)
+ })
+ })
+ .catch(err => rej(err))
})
}
-function downloadYoutubeDLVideo (url: string) {
- const path = generateVideoTmpPath(url)
+function getYoutubeDLSubs (url: string, opts?: object): Promise<YoutubeDLSubs> {
+ return new Promise<YoutubeDLSubs>((res, rej) => {
+ const cwd = CONFIG.STORAGE.TMP_DIR
+ const options = opts || { all: true, format: 'vtt', cwd }
+
+ safeGetYoutubeDL()
+ .then(youtubeDL => {
+ youtubeDL.getSubs(url, options, (err, files) => {
+ if (err) return rej(err)
+ if (!files) return []
+
+ logger.debug('Get subtitles from youtube dl.', { url, files })
+
+ const subtitles = files.reduce((acc, filename) => {
+ const matched = filename.match(/\.([a-z]{2})\.(vtt|ttml)/i)
+ if (!matched || !matched[1]) return acc
+
+ return [
+ ...acc,
+ {
+ language: matched[1],
+ path: join(cwd, filename),
+ filename
+ }
+ ]
+ }, [])
+
+ return res(subtitles)
+ })
+ })
+ .catch(err => rej(err))
+ })
+}
- logger.info('Importing youtubeDL video %s', url)
+function getYoutubeDLVideoFormat () {
+ /**
+ * 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 enabledResolutions = getEnabledResolutions('vod')
+ 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('/')
+}
- const options = [ '-f', 'bestvideo[ext=mp4]+bestaudio[ext=m4a]/best', '-o', path ]
+function downloadYoutubeDLVideo (url: string, fileExt: string, timeout: number) {
+ // Leave empty the extension, youtube-dl will add it
+ const pathWithoutExtension = generateVideoImportTmpPath(url, '')
- return new Promise<string>((res, rej) => {
- youtubeDL.exec(url, options, async (err, output) => {
- if (err) return rej(err)
+ let timer
+
+ logger.info('Importing youtubeDL video %s to %s', url, pathWithoutExtension)
+
+ let options = [ '-f', getYoutubeDLVideoFormat(), '-o', pathWithoutExtension ]
+ options = wrapWithProxyOptions(options)
+
+ if (process.env.FFMPEG_PATH) {
+ options = options.concat([ '--ffmpeg-location', process.env.FFMPEG_PATH ])
+ }
+
+ logger.debug('YoutubeDL options for %s.', url, { options })
- return res(path)
- })
+ return new Promise<string>((res, rej) => {
+ safeGetYoutubeDL()
+ .then(youtubeDL => {
+ youtubeDL.exec(url, options, processOptions, async err => {
+ clearTimeout(timer)
+
+ try {
+ // 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')
+ }
+
+ const path = await guessVideoPathWithExtension(pathWithoutExtension, fileExt)
+
+ if (err) {
+ remove(path)
+ .catch(err => logger.error('Cannot delete path on YoutubeDL error.', { err }))
+
+ return rej(err)
+ }
+
+ return res(path)
+ } catch (err) {
+ return rej(err)
+ }
+ })
+
+ timer = setTimeout(() => {
+ const err = new Error('YoutubeDL download timeout.')
+
+ guessVideoPathWithExtension(pathWithoutExtension, fileExt)
+ .then(path => remove(path))
+ .finally(() => rej(err))
+ .catch(err => {
+ logger.error('Cannot remove file in youtubeDL timeout.', { err })
+ return rej(err)
+ })
+ }, timeout)
+ })
+ .catch(err => rej(err))
})
}
+// Thanks: https://github.com/przemyslawpluta/node-youtube-dl/blob/master/lib/downloader.js
+// We rewrote it to avoid sync calls
+async function updateYoutubeDLBinary () {
+ logger.info('Updating youtubeDL binary.')
+
+ const binDirectory = join(root(), 'node_modules', 'youtube-dl', 'bin')
+ const bin = join(binDirectory, 'youtube-dl')
+ const detailsPath = join(binDirectory, 'details')
+ const url = process.env.YOUTUBE_DL_DOWNLOAD_HOST || 'https://yt-dl.org/downloads/latest/youtube-dl'
+
+ await ensureDir(binDirectory)
+
+ try {
+ const result = await got(url, { followRedirect: false })
+
+ if (result.statusCode !== HttpStatusCode.FOUND_302) {
+ logger.error('youtube-dl update error: did not get redirect for the latest version link. Status %d', result.statusCode)
+ return
+ }
+
+ const newUrl = result.headers.location
+ const newVersion = /yt-dl\.org\/downloads\/(\d{4}\.\d\d\.\d\d(\.\d)?)\/youtube-dl/.exec(newUrl)[1]
+
+ const downloadFileStream = got.stream(newUrl)
+ const writeStream = createWriteStream(bin, { mode: 493 })
+
+ await pipelinePromise(
+ downloadFileStream,
+ writeStream
+ )
+
+ const details = JSON.stringify({ version: newVersion, path: bin, exec: 'youtube-dl' })
+ await writeFile(detailsPath, details, { encoding: 'utf8' })
+
+ logger.info('youtube-dl updated to version %s.', newVersion)
+ } catch (err) {
+ logger.error('Cannot update youtube-dl.', { err })
+ }
+}
+
+async function safeGetYoutubeDL () {
+ let youtubeDL
+
+ try {
+ youtubeDL = require('youtube-dl')
+ } catch (e) {
+ // Download binary
+ await updateYoutubeDLBinary()
+ youtubeDL = require('youtube-dl')
+ }
+
+ return youtubeDL
+}
+
+function 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
+}
+
// ---------------------------------------------------------------------------
export {
+ updateYoutubeDLBinary,
+ getYoutubeDLVideoFormat,
downloadYoutubeDLVideo,
- getYoutubeDLInfo
+ getYoutubeDLSubs,
+ getYoutubeDLInfo,
+ safeGetYoutubeDL,
+ buildOriginallyPublishedAt
}
// ---------------------------------------------------------------------------
+async function 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
+ }
+
+ throw new Error('Cannot guess path of ' + tmpPath)
+}
+
function normalizeObject (obj: any) {
const newObj: any = {}
return newObj
}
-function buildVideoInfo (obj: any) {
+function buildVideoInfo (obj: any): YoutubeDLInfo {
return {
name: titleTruncation(obj.title),
description: descriptionTruncation(obj.description),
category: getCategory(obj.categories),
licence: getLicence(obj.license),
+ language: getLanguage(obj.language),
nsfw: isNSFW(obj),
tags: getTags(obj.tags),
- thumbnailUrl: obj.thumbnail || undefined
+ thumbnailUrl: obj.thumbnail || undefined,
+ originallyPublishedAt: buildOriginallyPublishedAt(obj),
+ ext: obj.ext
}
}
function titleTruncation (title: string) {
- return truncate(title, {
- 'length': CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
- 'separator': /,? +/,
- 'omission': ' […]'
+ return peertubeTruncate(title, {
+ length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
+ separator: /,? +/,
+ omission: ' […]'
})
}
function descriptionTruncation (description: string) {
if (!description || description.length < CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.min) return undefined
- return truncate(description, {
- 'length': CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.max,
- 'separator': /,? +/,
- 'omission': ' […]'
+ return peertubeTruncate(description, {
+ length: CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.max,
+ separator: /,? +/,
+ omission: ' […]'
})
}
function getLicence (licence: string) {
if (!licence) return undefined
- if (licence.indexOf('Creative Commons Attribution') !== -1) return 1
+ 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
}
return undefined
}
+
+function getLanguage (language: string) {
+ return VIDEO_LANGUAGES[language] ? language : undefined
+}
+
+function wrapWithProxyOptions (options: string[]) {
+ if (CONFIG.IMPORT.VIDEOS.HTTP.PROXY.ENABLED) {
+ logger.debug('Using proxy for YoutubeDL')
+
+ return [ '--proxy', CONFIG.IMPORT.VIDEOS.HTTP.PROXY.URL ].concat(options)
+ }
+
+ return options
+}