Add getSubs to YoutubeDL video import

2 files changed, 64 insertions, 1 deletions

diff --git a/server/controllers/api/videos/import.ts b/server/controllers/api/videos/import.ts
index da0832258..e9b9d68d7 100644
--- a/server/controllers/api/videos/import.ts
+++ b/server/controllers/api/videos/import.ts
@@ -3,11 +3,13 @@ import * as magnetUtil from 'magnet-uri'
 import { auditLoggerFactory, getAuditIdFromRes, VideoImportAuditView } from '../../../helpers/audit-logger'
 import { asyncMiddleware, asyncRetryTransactionMiddleware, authenticate, videoImportAddValidator } from '../../../middlewares'
 import { MIMETYPES } from '../../../initializers/constants'
-import { getYoutubeDLInfo, YoutubeDLInfo } from '../../../helpers/youtube-dl'
+import { getYoutubeDLInfo, YoutubeDLInfo, getYoutubeDLSubs } from '../../../helpers/youtube-dl'
 import { createReqFiles } from '../../../helpers/express-utils'
 import { logger } from '../../../helpers/logger'
 import { VideoImportCreate, VideoImportState, VideoPrivacy, VideoState } from '../../../../shared'
 import { VideoModel } from '../../../models/video/video'
+import { VideoCaptionModel } from '../../../models/video/video-caption'
+import { moveAndProcessCaptionFile } from '../../../helpers/captions-utils'
 import { getVideoActivityPubUrl } from '../../../lib/activitypub'
 import { TagModel } from '../../../models/video/tag'
 import { VideoImportModel } from '../../../models/video/video-import'
@@ -28,6 +30,7 @@ import {
   MThumbnail,
   MUser,
   MVideoAccountDefault,
+  MVideoCaptionVideo,
   MVideoTag,
   MVideoThumbnailAccountDefault,
   MVideoWithBlacklistLight
@@ -136,6 +139,7 @@ async function addYoutubeDLImport (req: express.Request, res: express.Response)
   const targetUrl = body.targetUrl
   const user = res.locals.oauth.token.User
 
+  // Get video infos
   let youtubeDLInfo: YoutubeDLInfo
   try {
     youtubeDLInfo = await getYoutubeDLInfo(targetUrl)
@@ -168,6 +172,29 @@ async function addYoutubeDLImport (req: express.Request, res: express.Response)
     user
   })
 
+
+  // Get video subtitles
+  try {
+    const subtitles = await getYoutubeDLSubs(targetUrl)
+
+    for (const subtitle of subtitles) {
+      const videoCaption = new VideoCaptionModel({
+        videoId: video.id,
+        language: subtitle.language
+      }) as MVideoCaptionVideo
+      videoCaption.Video = video
+
+      // Move physical file
+      await moveAndProcessCaptionFile(subtitle, videoCaption)
+
+      await sequelizeTypescript.transaction(async t => {
+        await VideoCaptionModel.insertOrReplaceLanguage(video.id, subtitle.language, null, t)
+      })
+    }
+  } catch (err) {
+    logger.warn('Cannot get video subtitles.', { err })
+  }
+
   // Create job to import the video
   const payload = {
     type: 'youtube-dl' as 'youtube-dl',
diff --git a/server/helpers/youtube-dl.ts b/server/helpers/youtube-dl.ts
index 07c85797a..277422645 100644
--- a/server/helpers/youtube-dl.ts
+++ b/server/helpers/youtube-dl.ts
@@ -20,6 +20,12 @@ export type YoutubeDLInfo = {
   originallyPublishedAt?: Date
 }
 
+export type YoutubeDLSubs = {
+  language: string,
+  filename: string,
+  path: string
+}[]
+
 const processOptions = {
   maxBuffer: 1024 * 1024 * 10 // 10MB
 }
@@ -45,6 +51,35 @@ function getYoutubeDLInfo (url: string, opts?: string[]): Promise<YoutubeDLInfo>
   })
 }
 
+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)
+
+          const subtitles = files.reduce((acc, filename) => {
+            const matched = filename.match(/\.([a-z]{2})\.(vtt|ttml)/i)
+
+            if (matched[1]) {
+              return [...acc, {
+                language: matched[1],
+                path: join(cwd, filename),
+                filename
+              }]
+            }
+          }, [])
+
+          return res(subtitles)
+        })
+      })
+      .catch(err => rej(err))
+  })
+}
+
 function downloadYoutubeDLVideo (url: string, extension: string, timeout: number) {
   const path = generateVideoImportTmpPath(url, extension)
   let timer
@@ -185,6 +220,7 @@ function buildOriginallyPublishedAt (obj: any) {
 export {
   updateYoutubeDLBinary,
   downloadYoutubeDLVideo,
+  getYoutubeDLSubs,
   getYoutubeDLInfo,
   safeGetYoutubeDL,
   buildOriginallyPublishedAt