feat: update youtube transcript fetch to allow all languages (#34436)

* feat: allow all languages

* feat: add youtube transcript import functions as drf

cms/djangoapps/contentstore/rest_api/v0/serializers/__init__.py

@@ -4,5 +4,5 @@ Serializers for v0 contentstore API.
 from .advanced_settings import AdvancedSettingsFieldSerializer, CourseAdvancedSettingsSerializer
 from .assets import AssetSerializer
 from .tabs import CourseTabSerializer, CourseTabUpdateSerializer, TabIDLocatorSerializer
-from .transcripts import TranscriptSerializer
+from .transcripts import TranscriptSerializer, YoutubeTranscriptCheckSerializer, YoutubeTranscriptUploadSerializer
 from .xblock import XblockSerializer

cms/djangoapps/contentstore/rest_api/v0/serializers/transcripts.py

@@ -13,3 +13,28 @@ class TranscriptSerializer(StrictSerializer):
     edx_video_id = serializers.CharField()
     language_code = serializers.CharField(required=False, allow_null=True)
     new_language_code = serializers.CharField(required=False, allow_null=True)
+
+
+class YoutubeTranscriptCheckSerializer(StrictSerializer):
+    """
+    Strict Serializer for YouTube transcripts check
+    """
+    html5_local = serializers.ListField(
+        child=serializers.CharField()
+    )
+    html5_equal = serializers.BooleanField()
+    is_youtube_mode = serializers.BooleanField()
+    youtube_local = serializers.BooleanField()
+    youtube_server = serializers.BooleanField()
+    youtube_diff = serializers.BooleanField()
+    current_item_subs = serializers.ListField(required=False, allow_null=True)
+    status = serializers.CharField()
+    command = serializers.CharField()
+
+
+class YoutubeTranscriptUploadSerializer(StrictSerializer):
+    """
+    Strict Serializer for YouTube transcripts upload
+    """
+    edx_video_id = serializers.CharField()
+    status = serializers.CharField()

cms/djangoapps/contentstore/rest_api/v0/urls.py

@@ -5,9 +5,16 @@ from django.urls import re_path, path
 
 from openedx.core.constants import COURSE_ID_PATTERN
 
-from .views import AdvancedCourseSettingsView, CourseTabSettingsView, CourseTabListView, CourseTabReorderView
+from .views import (
+    AdvancedCourseSettingsView,
+    CourseTabSettingsView,
+    CourseTabListView,
+    CourseTabReorderView,
+    TranscriptView,
+    YoutubeTranscriptCheckView,
+    YoutubeTranscriptUploadView,
+)
 from .views import assets
-from .views import transcripts
 from .views import authoring_videos
 from .views import xblock
 
@@ -68,7 +75,7 @@ urlpatterns = [
     ),
     re_path(
         fr'^video_transcripts/{settings.COURSE_ID_PATTERN}$',
-        transcripts.TranscriptView.as_view(), name='cms_api_video_transcripts'
+        TranscriptView.as_view(), name='cms_api_video_transcripts'
     ),
     re_path(
         fr'^xblock/{settings.COURSE_ID_PATTERN}$',
@@ -78,4 +85,12 @@ urlpatterns = [
         fr'^xblock/{settings.COURSE_ID_PATTERN}/{settings.USAGE_KEY_PATTERN}$',
         xblock.XblockView.as_view(), name='cms_api_xblock'
     ),
+    re_path(
+        fr'^youtube_transcripts/{settings.COURSE_ID_PATTERN}/check?$',
+        YoutubeTranscriptCheckView.as_view(), name='cms_api_youtube_transcripts_check'
+    ),
+    re_path(
+        fr'^youtube_transcripts/{settings.COURSE_ID_PATTERN}/upload?$',
+        YoutubeTranscriptUploadView.as_view(), name='cms_api_youtube_transcripts_upload'
+    ),
 ]

cms/djangoapps/contentstore/rest_api/v0/views/__init__.py

@@ -3,3 +3,4 @@ Views for v0 contentstore API.
 """
 from .advanced_settings import AdvancedCourseSettingsView
 from .tabs import CourseTabSettingsView, CourseTabListView, CourseTabReorderView
+from .transcripts import TranscriptView, YoutubeTranscriptCheckView, YoutubeTranscriptUploadView

cms/djangoapps/contentstore/rest_api/v0/views/transcripts.py

@@ -14,14 +14,14 @@ from openedx.core.lib.api.view_utils import DeveloperErrorViewMixin, view_auth_c
 from common.djangoapps.util.json_request import expect_json_in_class_view
 
 from cms.djangoapps.contentstore.api import course_author_access_required
-
+from cms.djangoapps.contentstore.views.transcripts_ajax import check_transcripts, replace_transcripts
 from cms.djangoapps.contentstore.transcript_storage_handlers import (
     upload_transcript,
     delete_video_transcript_or_404,
     handle_transcript_download,
 )
 import cms.djangoapps.contentstore.toggles as contentstore_toggles
-from ..serializers import TranscriptSerializer
+from ..serializers import TranscriptSerializer, YoutubeTranscriptCheckSerializer, YoutubeTranscriptUploadSerializer
 from rest_framework.parsers import (MultiPartParser, FormParser)
 from openedx.core.lib.api.parsers import TypedFileUploadParser
 
@@ -68,3 +68,50 @@ class TranscriptView(DeveloperErrorViewMixin, CreateAPIView, RetrieveAPIView, De
         """
 
         return delete_video_transcript_or_404(request)
+
+
+@view_auth_classes()
+class YoutubeTranscriptCheckView(DeveloperErrorViewMixin, RetrieveAPIView):
+    """
+    public rest API endpoints for the CMS API YouTube transcripts.
+    youtube_id: required argument, needed to authorize course authors and identify the video.
+    edx_video_id: required argument, needed to identify the transcript.
+    xblock_id: required argument, needed to identify the transcript.
+    """
+    serializer_class = YoutubeTranscriptCheckSerializer
+    parser_classes = (MultiPartParser, FormParser, TypedFileUploadParser)
+
+    def dispatch(self, request, *args, **kwargs):
+        if not toggles.use_studio_content_api():
+            raise Http404
+        return super().dispatch(request, *args, **kwargs)
+
+    @course_author_access_required
+    def retrieve(self, request, course_key_string):  # pylint: disable=arguments-differ
+        """
+        Get the status of youtube transcripts for a given youtube video
+        """
+        return check_transcripts(request)
+
+
+@view_auth_classes()
+class YoutubeTranscriptUploadView(DeveloperErrorViewMixin, RetrieveAPIView):
+    """
+    public rest API endpoints for the CMS API YouTube transcripts.
+    youtube_id: required argument, needed to authorize course authors and identify the video.
+    xblock_id: required argument, needed to identify the transcript.
+    """
+    serializer_class = YoutubeTranscriptUploadSerializer
+    parser_classes = (MultiPartParser, FormParser, TypedFileUploadParser)
+
+    def dispatch(self, request, *args, **kwargs):
+        if not toggles.use_studio_content_api():
+            raise Http404
+        return super().dispatch(request, *args, **kwargs)
+
+    @course_author_access_required
+    def retrieve(self, request, course_key_string):  # pylint: disable=arguments-differ
+        """
+        Get the  youtube transcripts for a give youtube video and add them to video block
+        """
+        return replace_transcripts(request)