PeerTube/server/helpers/youtube-dl.ts

290 lines
8.0 KiB
TypeScript
Raw Normal View History

import { CONSTRAINTS_FIELDS, VIDEO_CATEGORIES } from '../initializers/constants'
import { logger } from './logger'
2018-12-04 16:02:49 +01:00
import { generateVideoImportTmpPath } from './utils'
2018-09-19 16:24:24 +02:00
import { join } from 'path'
import { peertubeTruncate, root } from './core-utils'
import { ensureDir, remove, writeFile } from 'fs-extra'
2018-09-19 16:24:24 +02:00
import * as request from 'request'
import { createWriteStream } from 'fs'
import { CONFIG } from '@server/initializers/config'
export type YoutubeDLInfo = {
2018-08-06 17:13:39 +02:00
name?: string
description?: string
category?: number
licence?: number
nsfw?: boolean
tags?: string[]
thumbnailUrl?: string
originallyPublishedAt?: Date
}
const processOptions = {
maxBuffer: 1024 * 1024 * 10 // 10MB
}
function getYoutubeDLInfo (url: string, opts?: string[]): Promise<YoutubeDLInfo> {
2020-01-31 16:56:52 +01:00
return new Promise<YoutubeDLInfo>((res, rej) => {
2020-01-22 17:19:03 +01:00
let args = opts || [ '-j', '--flat-playlist' ]
args = wrapWithProxyOptions(args)
2020-01-31 16:56:52 +01:00
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.'))
2020-01-31 16:56:52 +01:00
const obj = buildVideoInfo(normalizeObject(info))
if (obj.name && obj.name.length < CONSTRAINTS_FIELDS.VIDEOS.NAME.min) obj.name += ' video'
2020-01-31 16:56:52 +01:00
return res(obj)
})
})
.catch(err => rej(err))
})
}
function downloadYoutubeDLVideo (url: string, timeout: number) {
2018-12-04 16:02:49 +01:00
const path = generateVideoImportTmpPath(url)
let timer
2018-08-06 17:13:39 +02:00
logger.info('Importing youtubeDL video %s', url)
let options = [ '-f', 'bestvideo[ext=mp4]+bestaudio[ext=m4a]/best', '-o', path ]
2020-01-22 17:19:03 +01:00
options = wrapWithProxyOptions(options)
2019-04-03 10:16:18 +02:00
if (process.env.FFMPEG_PATH) {
options = options.concat([ '--ffmpeg-location', process.env.FFMPEG_PATH ])
2019-04-03 10:16:18 +02:00
}
2020-01-31 16:56:52 +01:00
return new Promise<string>((res, rej) => {
safeGetYoutubeDL()
.then(youtubeDL => {
youtubeDL.exec(url, options, processOptions, err => {
clearTimeout(timer)
2020-01-31 16:56:52 +01:00
if (err) {
remove(path)
.catch(err => logger.error('Cannot delete path on YoutubeDL error.', { err }))
2020-01-31 16:56:52 +01:00
return rej(err)
}
2020-01-31 16:56:52 +01:00
return res(path)
})
2020-01-31 16:56:52 +01:00
timer = setTimeout(() => {
const err = new Error('YoutubeDL download timeout.')
2020-01-31 16:56:52 +01:00
remove(path)
.finally(() => rej(err))
.catch(err => {
logger.error('Cannot remove %s in youtubeDL timeout.', path, { err })
return rej(err)
})
}, timeout)
})
.catch(err => rej(err))
})
}
2018-09-19 16:24:24 +02:00
// 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 = 'https://yt-dl.org/downloads/latest/youtube-dl'
await ensureDir(binDirectory)
return new Promise(res => {
request.get(url, { followRedirect: false }, (err, result) => {
if (err) {
logger.error('Cannot update youtube-dl.', { err })
return res()
}
if (result.statusCode !== 302) {
logger.error('youtube-dl update error: did not get redirect for the latest version link. Status %d', result.statusCode)
return res()
}
const url = result.headers.location
const downloadFile = request.get(url)
2020-01-31 16:56:52 +01:00
const newVersion = /yt-dl\.org\/downloads\/(\d{4}\.\d\d\.\d\d(\.\d)?)\/youtube-dl/.exec(url)[1]
2018-09-19 16:24:24 +02:00
downloadFile.on('response', result => {
if (result.statusCode !== 200) {
logger.error('Cannot update youtube-dl: new version response is not 200, it\'s %d.', result.statusCode)
return res()
}
downloadFile.pipe(createWriteStream(bin, { mode: 493 }))
})
downloadFile.on('error', err => {
logger.error('youtube-dl update error.', { err })
return res()
})
downloadFile.on('end', () => {
const details = JSON.stringify({ version: newVersion, path: bin, exec: 'youtube-dl' })
writeFile(detailsPath, details, { encoding: 'utf8' }, err => {
if (err) {
logger.error('youtube-dl update error: cannot write details.', { err })
return res()
}
logger.info('youtube-dl updated to version %s.', newVersion)
return res()
})
})
})
})
}
async function safeGetYoutubeDL () {
let youtubeDL
try {
youtubeDL = require('youtube-dl')
} catch (e) {
// Download binary
2018-09-19 16:24:24 +02:00
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,
downloadYoutubeDLVideo,
getYoutubeDLInfo,
safeGetYoutubeDL,
buildOriginallyPublishedAt
}
// ---------------------------------------------------------------------------
function 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
}
function buildVideoInfo (obj: any) {
return {
name: titleTruncation(obj.title),
description: descriptionTruncation(obj.description),
category: getCategory(obj.categories),
licence: getLicence(obj.license),
nsfw: isNSFW(obj),
tags: getTags(obj.tags),
thumbnailUrl: obj.thumbnail || undefined,
originallyPublishedAt: buildOriginallyPublishedAt(obj)
}
}
function titleTruncation (title: string) {
return peertubeTruncate(title, {
length: CONSTRAINTS_FIELDS.VIDEOS.NAME.max,
separator: /,? +/,
omission: ' […]'
})
}
function descriptionTruncation (description: string) {
2018-08-02 17:48:50 +02:00
if (!description || description.length < CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.min) return undefined
return peertubeTruncate(description, {
length: CONSTRAINTS_FIELDS.VIDEOS.DESCRIPTION.max,
separator: /,? +/,
omission: ' […]'
})
}
function isNSFW (info: any) {
return info.age_limit && info.age_limit >= 16
}
function getTags (tags: any) {
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)
}
function getLicence (licence: string) {
if (!licence) return undefined
2020-02-28 16:03:39 +01:00
if (licence.includes('Creative Commons Attribution')) return 1
return undefined
}
function 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
}
2020-01-22 17:19:03 +01:00
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
}