Update app.py

app.py

@@ -21,14 +21,14 @@ async def create_youtube_transcription(youtube_url: str) -> List[langchain_core.
     """
     try:
         loader = YoutubeLoader.from_youtube_url(
-            youtube_url, add_video_info=True
+            youtube_url, add_video_info=False
         ) # we can also pass an array of youtube urls to load multiple videos at once!
         youtube_docs = loader.load() # this loads the transcript
         return youtube_docs
     except Exception as e:
-        await cl.Message(content=f"failed to load youtube video: {e}").send() # display the error if we failed to load the youtube video
+        await cl.Message(content=f"failed to load youtube video: {e} Please refresh the page").send() # display the error if we failed to load the youtube video
 
-def create_text_splitter(docs: List[langchain_core.documents.Document]):
+async def create_text_splitter(docs: List[langchain_core.documents.Document]):
     """
     Create a text splitter from a list of documents
     More Info: ument_transformers/recursive_text_splitter/
@@ -102,13 +102,13 @@ async def start():
         # more on ask user message: https://docs.chainlit.io/api-reference/ask/ask-for-input
         await cl.Message(content=f"youtube link: {youtube_link}").send() # display and double check to make sure the link is correct
         youtube_docs = await create_youtube_transcription(youtube_link['content']) # create the youtube transcription
+        transcription = youtube_docs[0].page_content # get the transcription of the first document
+        await cl.Message(content=f"youtube docs: {transcription}").send() # display the transcription of the first document to show that we have the correct data
         split_docs = create_text_splitter(youtube_docs) # split the documents into chunks
         vector_db = create_faiss_vector_store(split_docs) # create the vector db
         bm25 = create_bm25_retreiver(split_docs) # create the BM25 retreiver
         ensemble_retriever = await create_ensemble_retriever(vector_db, bm25) # create the ensemble retriever
         cl.user_session.set("ensemble_retriever", ensemble_retriever) # store the ensemble retriever in the user session for our on message function
-        transcription = youtube_docs[0].page_content # get the transcription of the first document
-        await cl.Message(content=f"youtube docs: {transcription}").send() # display the transcription of the first document to show that we have the correct data
     except Exception as e:
         await cl.Message(content=f"failed to load model: {e}").send() # display the error if we failed to load the model