PeerTube/server/tools/peertube-import-videos.ts

344 lines
9.6 KiB
TypeScript
Raw Normal View History

2021-06-25 17:48:27 +02:00
import { program } from 'commander'
import { accessSync, constants } from 'fs'
import { remove } from 'fs-extra'
import { join } from 'path'
import { YoutubeDLCLI, YoutubeDLInfo, YoutubeDLInfoBuilder } from '@server/helpers/youtube-dl'
2021-12-17 14:20:43 +01:00
import { wait } from '@shared/core-utils'
import { sha256 } from '@shared/extra-utils'
import { doRequestAndSaveToFile } from '../helpers/requests'
2021-07-09 15:03:44 +02:00
import {
2021-07-13 11:44:16 +02:00
assignToken,
2021-07-09 15:03:44 +02:00
buildCommonVideoOptions,
buildServer,
buildVideoAttributesFromCommander,
getLogger,
getServerCredentials
} from './cli'
2021-08-27 14:32:44 +02:00
import prompt = require('prompt')
const processOptions = {
maxBuffer: Infinity
}
2018-02-09 16:47:06 +01:00
let command = program
.name('import-videos')
command = buildCommonVideoOptions(command)
command
2018-02-09 16:47:06 +01:00
.option('-u, --url <url>', 'Server url')
.option('-U, --username <username>', 'Username')
.option('-p, --password <token>', 'Password')
.option('--target-url <targetUrl>', 'Video target URL')
.option('--since <since>', 'Publication date (inclusive) since which the videos can be imported (YYYY-MM-DD)', parseDate)
.option('--until <until>', 'Publication date (inclusive) until which the videos can be imported (YYYY-MM-DD)', parseDate)
.option('--first <first>', 'Process first n elements of returned playlist')
.option('--last <last>', 'Process last n elements of returned playlist')
.option('--wait-interval <waitInterval>', 'Duration between two video imports (in seconds)', convertIntoMs)
.option('-T, --tmpdir <tmpdir>', 'Working directory', __dirname)
.usage('[global options] [ -- youtube-dl options]')
2018-02-09 16:47:06 +01:00
.parse(process.argv)
2021-02-03 09:33:05 +01:00
const options = command.opts()
const log = getLogger(options.verbose)
getServerCredentials(command)
.then(({ url, username, password }) => {
2021-02-03 09:33:05 +01:00
if (!options.targetUrl) {
exitError('--target-url field is required.')
}
try {
2021-02-03 09:33:05 +01:00
accessSync(options.tmpdir, constants.R_OK | constants.W_OK)
} catch (e) {
2021-02-03 09:33:05 +01:00
exitError('--tmpdir %s: directory does not exist or is not accessible', options.tmpdir)
}
2018-05-10 23:59:28 +02:00
url = normalizeTargetUrl(url)
2021-02-03 09:33:05 +01:00
options.targetUrl = normalizeTargetUrl(options.targetUrl)
2021-07-09 15:03:44 +02:00
run(url, username, password)
2020-01-31 16:56:52 +01:00
.catch(err => exitError(err))
})
2020-01-31 16:56:52 +01:00
.catch(err => console.error(err))
2018-02-09 16:47:06 +01:00
2021-07-09 15:03:44 +02:00
async function run (url: string, username: string, password: string) {
if (!password) password = await promptPassword()
2018-05-11 11:26:50 +02:00
const youtubeDLBinary = await YoutubeDLCLI.safeGet()
2021-05-11 10:54:05 +02:00
let info = await getYoutubeDLInfo(youtubeDLBinary, options.targetUrl, command.args)
if (!Array.isArray(info)) info = [ info ]
2018-02-09 16:47:06 +01:00
// Try to fix youtube channels upload
const uploadsObject = info.find(i => !i.ie_key && !i.duration && i.title === 'Uploads')
if (uploadsObject) {
console.log('Fixing URL to %s.', uploadsObject.url)
2021-05-11 10:54:05 +02:00
info = await getYoutubeDLInfo(youtubeDLBinary, uploadsObject.url, command.args)
2020-11-17 15:28:24 +01:00
}
2018-02-09 16:47:06 +01:00
2020-11-17 15:28:24 +01:00
let infoArray: any[]
2020-11-17 15:28:24 +01:00
infoArray = [].concat(info)
2021-02-03 09:33:05 +01:00
if (options.first) {
infoArray = infoArray.slice(0, options.first)
} else if (options.last) {
infoArray = infoArray.slice(-options.last)
2020-11-17 15:28:24 +01:00
}
2018-02-09 16:47:06 +01:00
2020-11-17 15:28:24 +01:00
log.info('Will download and upload %d videos.\n', infoArray.length)
let skipInterval = true
for (const [ index, info ] of infoArray.entries()) {
2020-11-17 15:28:24 +01:00
try {
if (index > 0 && options.waitInterval && !skipInterval) {
log.info('Wait for %d seconds before continuing.', options.waitInterval / 1000)
await wait(options.waitInterval)
}
skipInterval = await processVideo({
2021-02-03 09:33:05 +01:00
cwd: options.tmpdir,
2020-11-17 15:28:24 +01:00
url,
2021-07-09 15:03:44 +02:00
username,
password,
2020-11-17 15:28:24 +01:00
youtubeInfo: info
})
} catch (err) {
console.error('Cannot process video.', { info, url, err })
2018-02-09 16:47:06 +01:00
}
2020-11-17 15:28:24 +01:00
}
2018-02-09 16:47:06 +01:00
2021-07-09 15:03:44 +02:00
log.info('Video/s for user %s imported: %s', username, options.targetUrl)
2020-11-17 15:28:24 +01:00
process.exit(0)
2018-02-09 16:47:06 +01:00
}
async function processVideo (parameters: {
2020-01-31 16:56:52 +01:00
cwd: string
url: string
2021-07-09 15:03:44 +02:00
username: string
password: string
2019-06-13 11:09:38 +02:00
youtubeInfo: any
}) {
2021-07-09 15:03:44 +02:00
const { youtubeInfo, cwd, url, username, password } = parameters
2019-06-13 11:09:38 +02:00
log.debug('Fetching object.', youtubeInfo)
const videoInfo = await fetchObject(youtubeInfo)
log.debug('Fetched object.', videoInfo)
if (options.since && videoInfo.originallyPublishedAt && videoInfo.originallyPublishedAt.getTime() < options.since.getTime()) {
log.info('Video "%s" has been published before "%s", don\'t upload it.\n', videoInfo.name, formatDate(options.since))
return true
}
2021-07-09 15:03:44 +02:00
if (options.until && videoInfo.originallyPublishedAt && videoInfo.originallyPublishedAt.getTime() > options.until.getTime()) {
log.info('Video "%s" has been published after "%s", don\'t upload it.\n', videoInfo.name, formatDate(options.until))
return true
}
2018-02-12 11:02:14 +01:00
2021-07-09 15:03:44 +02:00
const server = buildServer(url)
2021-07-16 09:04:35 +02:00
const { data } = await server.search.advancedVideoSearch({
2021-07-09 15:03:44 +02:00
search: {
search: videoInfo.name,
2021-07-09 15:03:44 +02:00
sort: '-match',
searchTarget: 'local'
}
})
2018-02-12 11:02:14 +01:00
log.info('############################################################\n')
2018-02-09 16:47:06 +01:00
if (data.find(v => v.name === videoInfo.name)) {
log.info('Video "%s" already exists, don\'t reupload it.\n', videoInfo.name)
return true
}
2018-02-09 16:47:06 +01:00
const path = join(cwd, sha256(videoInfo.url) + '.mp4')
2018-02-09 16:47:06 +01:00
log.info('Downloading video "%s"...', videoInfo.name)
try {
const youtubeDLBinary = await YoutubeDLCLI.safeGet()
const output = await youtubeDLBinary.download({
url: videoInfo.url,
2022-08-09 09:09:31 +02:00
format: YoutubeDLCLI.getYoutubeDLVideoFormat([], false),
output: path,
additionalYoutubeDLArgs: command.args,
processOptions
})
log.info(output.join('\n'))
await uploadVideoOnPeerTube({
cwd,
url,
2021-07-09 15:03:44 +02:00
username,
password,
videoInfo,
videoPath: path
})
} catch (err) {
log.error(err.message)
}
return false
2018-02-09 16:47:06 +01:00
}
2019-06-13 11:09:38 +02:00
async function uploadVideoOnPeerTube (parameters: {
videoInfo: YoutubeDLInfo
2020-01-31 16:56:52 +01:00
videoPath: string
cwd: string
url: string
2021-07-09 15:03:44 +02:00
username: string
password: string
2019-06-13 11:09:38 +02:00
}) {
const { videoInfo, videoPath, cwd, url, username, password } = parameters
2019-06-13 11:09:38 +02:00
2021-07-15 10:02:54 +02:00
const server = buildServer(url)
await assignToken(server, username, password)
let thumbnailfile: string
if (videoInfo.thumbnailUrl) {
thumbnailfile = join(cwd, sha256(videoInfo.thumbnailUrl) + '.jpg')
2018-02-14 15:56:07 +01:00
await doRequestAndSaveToFile(videoInfo.thumbnailUrl, thumbnailfile)
2018-02-14 15:56:07 +01:00
}
const baseAttributes = await buildVideoAttributesFromCommander(server, program, videoInfo)
2021-07-09 15:03:44 +02:00
2021-07-15 10:02:54 +02:00
const attributes = {
...baseAttributes,
originallyPublishedAt: videoInfo.originallyPublishedAt
? videoInfo.originallyPublishedAt.toISOString()
: null,
thumbnailfile,
previewfile: thumbnailfile,
fixture: videoPath
2021-07-15 10:02:54 +02:00
}
2019-06-13 11:09:38 +02:00
2021-07-15 10:02:54 +02:00
log.info('\nUploading on PeerTube video "%s".', attributes.name)
2019-06-13 11:09:38 +02:00
try {
2021-07-16 09:04:35 +02:00
await server.videos.upload({ attributes })
} catch (err) {
2018-02-20 18:18:47 +01:00
if (err.message.indexOf('401') !== -1) {
log.info('Got 401 Unauthorized, token may have expired, renewing token and retry.')
2021-07-16 09:04:35 +02:00
server.accessToken = await server.login.getAccessToken(username, password)
2021-07-16 09:04:35 +02:00
await server.videos.upload({ attributes })
} else {
exitError(err.message)
}
}
2018-02-14 15:56:07 +01:00
2018-08-27 16:23:34 +02:00
await remove(videoPath)
if (thumbnailfile) await remove(thumbnailfile)
2018-02-14 15:56:07 +01:00
log.info('Uploaded video "%s"!\n', attributes.name)
2018-02-09 16:47:06 +01:00
}
2019-06-13 11:09:38 +02:00
/* ---------------------------------------------------------- */
async function fetchObject (info: any) {
const url = buildUrl(info)
2018-02-12 11:02:14 +01:00
const youtubeDLCLI = await YoutubeDLCLI.safeGet()
const result = await youtubeDLCLI.getInfo({
url,
2022-08-09 09:09:31 +02:00
format: YoutubeDLCLI.getYoutubeDLVideoFormat([], false),
processOptions
})
const builder = new YoutubeDLInfoBuilder(result)
const videoInfo = builder.getInfo()
return { ...videoInfo, url }
}
function buildUrl (info: any) {
2018-02-20 18:56:43 +01:00
const webpageUrl = info.webpage_url as string
2020-01-31 16:56:52 +01:00
if (webpageUrl?.match(/^https?:\/\//)) return webpageUrl
2018-02-20 18:56:43 +01:00
const url = info.url as string
2020-01-31 16:56:52 +01:00
if (url?.match(/^https?:\/\//)) return url
// It seems youtube-dl does not return the video url
return 'https://www.youtube.com/watch?v=' + info.id
}
2018-02-20 18:56:43 +01:00
function normalizeTargetUrl (url: string) {
let normalizedUrl = url.replace(/\/+$/, '')
2019-12-30 10:44:50 +01:00
if (!normalizedUrl.startsWith('http://') && !normalizedUrl.startsWith('https://')) {
normalizedUrl = 'https://' + normalizedUrl
}
return normalizedUrl
}
2019-06-13 11:09:38 +02:00
async function promptPassword () {
return new Promise<string>((res, rej) => {
prompt.start()
const schema = {
properties: {
password: {
hidden: true,
required: true
}
}
}
prompt.get(schema, function (err, result) {
if (err) {
return rej(err)
}
return res(result.password)
})
})
}
function parseDate (dateAsStr: string): Date {
if (!/\d{4}-\d{2}-\d{2}/.test(dateAsStr)) {
exitError(`Invalid date passed: ${dateAsStr}. Expected format: YYYY-MM-DD. See help for usage.`)
}
const date = new Date(dateAsStr)
date.setHours(0, 0, 0)
if (isNaN(date.getTime())) {
exitError(`Invalid date passed: ${dateAsStr}. See help for usage.`)
}
return date
}
function formatDate (date: Date): string {
2020-01-31 16:56:52 +01:00
return date.toISOString().split('T')[0]
}
function convertIntoMs (secondsAsStr: string): number {
const seconds = parseInt(secondsAsStr, 10)
if (seconds <= 0) {
exitError(`Invalid duration passed: ${seconds}. Expected duration to be strictly positive and in seconds`)
}
return Math.round(seconds * 1000)
}
function exitError (message: string, ...meta: any[]) {
// use console.error instead of log.error here
console.error(message, ...meta)
process.exit(-1)
}
2020-11-17 15:28:24 +01:00
function getYoutubeDLInfo (youtubeDLCLI: YoutubeDLCLI, url: string, args: string[]) {
return youtubeDLCLI.getInfo({
url,
2022-08-09 09:09:31 +02:00
format: YoutubeDLCLI.getYoutubeDLVideoFormat([], false),
additionalYoutubeDLArgs: [ '-j', '--flat-playlist', '--playlist-reverse', ...args ],
processOptions
2020-11-17 15:28:24 +01:00
})
}