feat: Video editor supports transcripts [FC-0076]

cms/djangoapps/contentstore/helpers.py

@@ -10,6 +10,7 @@
 import re
 
 from attrs import frozen, Factory
+from django.core.files.base import ContentFile
 from django.conf import settings
 from django.contrib.auth import get_user_model
 from django.utils.translation import gettext as _
@@ -23,6 +24,8 @@
 from xmodule.exceptions import NotFoundError
 from xmodule.modulestore.django import modulestore
 from xmodule.xml_block import XmlMixin
+from xmodule.video_block.transcripts_utils import Transcript, build_components_import_path
+from edxval.api import create_external_video, create_or_update_video_transcript
 
 from cms.djangoapps.models.settings.course_grading import CourseGradingModel
 from cms.lib.xblock.upstream_sync import UpstreamLink, UpstreamLinkException, fetch_customizable_fields
@@ -299,13 +302,21 @@ def import_staged_content_from_user_clipboard(parent_key: UsageKey, request) ->
             tags=user_clipboard.content.tags,
         )
 
+        usage_key = new_xblock.scope_ids.usage_id
+        if usage_key.block_type == 'video':
+            # The edx_video_id must always be new so as not
+            # to interfere with the data of the copied block
+            new_xblock.edx_video_id = create_external_video(display_name='external video')
+            store.update_item(new_xblock, request.user.id)
+
         # Now handle static files that need to go into Files & Uploads.
         static_files = content_staging_api.get_staged_content_static_files(user_clipboard.content.id)
         notices, substitutions = _import_files_into_course(
+            block=new_xblock,
             course_key=parent_key.context_key,
             staged_content_id=user_clipboard.content.id,
             static_files=static_files,
-            usage_key=new_xblock.scope_ids.usage_id,
+            usage_key=usage_key,
         )
 
         # Rewrite the OLX's static asset references to point to the new
@@ -504,6 +515,7 @@ def _import_xml_node_to_parent(
 
 
 def _import_files_into_course(
+    block: XBlock,
     course_key: CourseKey,
     staged_content_id: int,
     static_files: list[content_staging_api.StagedContentFileData],
@@ -540,6 +552,7 @@ def _import_files_into_course(
         # At this point, we know this is a "Files & Uploads" asset that we may need to copy into the course:
         try:
             result, substitution_for_file = _import_file_into_course(
+                block,
                 course_key,
                 staged_content_id,
                 file_data_obj,
@@ -566,6 +579,7 @@ def _import_files_into_course(
 
 
 def _import_file_into_course(
+    block: XBlock,
     course_key: CourseKey,
     staged_content_id: int,
     file_data_obj: content_staging_api.StagedContentFileData,
@@ -583,8 +597,8 @@ def _import_file_into_course(
     # we're not going to attempt to change.
     if clipboard_file_path.startswith('static/'):
         # If it's in this form, it came from a library and assumes component-local assets
-        file_path = clipboard_file_path.lstrip('static/')
-        import_path = f"components/{usage_key.block_type}/{usage_key.block_id}/{file_path}"
+        file_path = clipboard_file_path.removeprefix('static/')
+        import_path = build_components_import_path(usage_key, file_path)
         filename = pathlib.Path(file_path).name
         new_key = course_key.make_asset_key("asset", import_path.replace("/", "_"))
     else:
@@ -616,6 +630,24 @@ def _import_file_into_course(
         if thumbnail_content is not None:
             content.thumbnail_location = thumbnail_location
         contentstore().save(content)
+        if usage_key.block_type == 'video':
+            # Adding transcripts to VAL using the new edx_video_id
+            language_code = next((k for k, v in block.transcripts.items() if v == filename), None)
+            if language_code:
+                sjson_subs = Transcript.convert(
+                    content=data,
+                    input_format=Transcript.SRT,
+                    output_format=Transcript.SJSON
+                ).encode()
+                create_or_update_video_transcript(
+                    video_id=block.edx_video_id,
+                    language_code=language_code,
+                    metadata={
+                        'file_format': Transcript.SJSON,
+                        'language_code': language_code
+                    },
+                    file_data=ContentFile(sjson_subs),
+                )
         return True, {clipboard_file_path: f"static/{import_path}"}
     elif current_file.content_digest == file_data_obj.md5_hash:
         # The file already exists and matches exactly, so no action is needed

cms/djangoapps/contentstore/views/tests/test_transcripts.py

@@ -15,9 +15,11 @@
 from django.urls import reverse
 from edxval.api import create_video
 from opaque_keys.edx.keys import UsageKey
+from organizations.tests.factories import OrganizationFactory
 
 from cms.djangoapps.contentstore.tests.utils import CourseTestCase, setup_caption_responses
 from openedx.core.djangoapps.contentserver.caching import del_cached_content
+from openedx.core.djangoapps.content_libraries import api as lib_api
 from xmodule.contentstore.content import StaticContent  # lint-amnesty, pylint: disable=wrong-import-order
 from xmodule.contentstore.django import contentstore  # lint-amnesty, pylint: disable=wrong-import-order
 from xmodule.exceptions import NotFoundError  # lint-amnesty, pylint: disable=wrong-import-order
@@ -27,8 +29,10 @@
     GetTranscriptsFromYouTubeException,
     Transcript,
     get_video_transcript_content,
-    remove_subs_from_store
+    get_transcript,
+    remove_subs_from_store,
 )
+from openedx.core.djangoapps.xblock import api as xblock_api
 
 TEST_DATA_CONTENTSTORE = copy.deepcopy(settings.CONTENTSTORE)
 TEST_DATA_CONTENTSTORE['DOC_STORE_CONFIG']['db'] = 'test_xcontent_%s' % uuid4().hex
@@ -92,6 +96,21 @@ def setUp(self):
         resp = self.client.ajax_post('/xblock/', data)
         self.assertEqual(resp.status_code, 200)
 
+        self.library = lib_api.create_library(
+            org=OrganizationFactory.create(short_name="org1"),
+            slug="lib",
+            title="Library",
+        )
+        self.library_block_metadata = lib_api.create_library_block(
+            self.library.key,
+            "video",
+            "video-transcript",
+        )
+        self.library_block = xblock_api.load_block(
+            self.library_block_metadata.usage_key,
+            self.user,
+        )
+
         self.video_usage_key = self._get_usage_key(resp)
         self.item = modulestore().get_item(self.video_usage_key)
         # hI10vDNYz4M - valid Youtube ID with transcripts.
@@ -702,6 +721,25 @@ def test_replace_transcript_success(self, edx_video_id):
         expected_sjson_content = json.loads(SJSON_TRANSCRIPT_CONTENT)
         self.assertDictEqual(actual_sjson_content, expected_sjson_content)
 
+    def test_replace_transcript_library_content_success(self):
+        # Make call to replace transcripts from youtube
+        response = self.replace_transcript(self.library_block_metadata.usage_key, self.youtube_id)
+
+        # Verify the response
+        self.assert_response(response, expected_status_code=200, expected_message='Success')
+
+        # Obtain updated block
+        updated_block = xblock_api.load_block(
+            self.library_block_metadata.usage_key,
+            self.user,
+        )
+
+        # Verify transcript content
+        transcript = get_transcript(updated_block, 'en', Transcript.SJSON)
+        actual_sjson_content = json.loads(transcript[0])
+        expected_sjson_content = json.loads(SJSON_TRANSCRIPT_CONTENT)
+        self.assertDictEqual(actual_sjson_content, expected_sjson_content)
+
     def test_replace_transcript_fails_without_data(self):
         """
         Verify that replace transcript fails if we do not provide video data in request.