Update app.py

app.py

@@ -115,17 +115,15 @@ if "vectorstore" not in st.session_state:
     st.session_state.vectorstore = None
 if "index_created" not in st.session_state:
     st.session_state.index_created = False
-
-# Initialize embeddings once at the start (only for web URLs)
-if "embeddings" not in st.session_state:
-    st.session_state.embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+if "content_type" not in st.session_state:
+    st.session_state.content_type = None
 
 # Initialize LLM once at the start
 if "llm" not in st.session_state:
     st.session_state.llm = ChatGroq(
         api_key=GROQ_API_KEY,
         model="llama3-70b-8192",
-        max_tokens=1024  # Reduced to lower resource usage
+        max_tokens=512  # Further reduced to minimize resource usage
     )
 
 # Sidebar for URL and YouTube input
@@ -218,14 +216,8 @@ def extract_subtitles_with_ytdlp(video_url):
             'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36',
             'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8',
             'Accept-Language': 'en-US,en;q=0.5',
-            'Referer': 'https://www.youtube.com/',
-            'Origin': 'https://www.youtube.com',
         },
         'cookiefile': 'cookies.txt',
-        'retries': 5,  # Increased retries
-        'retry_sleep': 3,
-        'geo_bypass': True,  # Attempt to bypass geo-restrictions
-        'no_check_certificate': True,  # Bypass certificate issues
     }
     try:
         if not os.path.exists('cookies.txt'):
@@ -331,10 +323,15 @@ if process_url_clicked:
                         st.error("No content loaded from URL. Try a different URL (e.g., https://www.bbc.com/news/science-environment-67299122).")
                         st.stop()
 
+                    # Initialize embeddings only when needed
+                    if "embeddings" not in st.session_state:
+                        st.session_state.embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+
                     st.session_state.url_content = "\n".join([doc.page_content for doc in data])
                     embeddings = st.session_state.embeddings
                     st.session_state.vectorstore = process_content(st.session_state.url_content, embeddings, source=url.strip())
                     st.session_state.index_created = True
+                    st.session_state.content_type = "web"
                     st.session_state.summary = None
                     st.text("Content processed successfully! ✅✅✅")
                 except Exception as e:
@@ -370,10 +367,10 @@ if process_youtube_clicked:
                                 "1. Captions are not enabled for this video.\n"
                                 "2. YouTube detected this request as a bot (even with cookies.txt).\n"
                                 "Solutions:\n"
-                                "- Ensure captions are enabled for the video.\n"
+                                "- Ensure captions are enabled for the video by checking the video settings on YouTube (gear icon > Subtitles/CC > Enable if available).\n"
                                 "- Regenerate and upload a fresh cookies.txt file (see instructions above).\n"
-                                "- Try a different video (e.g., https://www.youtube.com/watch?v=dQw4w9WgXcQ).\n"
-                                "- Test locally to rule out Hugging Face Spaces IP restrictions."
+                                "- Try a different video (e.g., https://www.youtube.com/watch?v=dQw4w9WgXcQ, which has transcripts available).\n"
+                                "- Test locally to rule out Hugging Face Spaces IP restrictions by running: pip install -r requirements.txt && streamlit run app.py"
                             )
                             st.stop()
 
@@ -385,6 +382,7 @@ if process_youtube_clicked:
                     # No vector store for YouTube videos since we're not doing QA
                     st.session_state.vectorstore = None
                     st.session_state.index_created = False
+                    st.session_state.content_type = "youtube"
                     st.session_state.summary = None
                     st.text("YouTube video processed successfully! ✅✅✅")
                 except Exception as e:
@@ -395,10 +393,8 @@ if process_youtube_clicked:
 with main_container:
     if st.session_state.url_content and st.button("Generate Summary"):
         with st.spinner("Generating summary..."):
-            # Check if the content is from a YouTube video (based on last processed input)
-            is_youtube = youtube_url.strip() and youtube_url == (st.session_state.get('last_processed_url', ''))
+            is_youtube = st.session_state.content_type == "youtube"
             st.session_state.summary = summarize_content(st.session_state.url_content, st.session_state.llm, is_youtube=is_youtube)
-            st.session_state.last_processed_url = youtube_url if is_youtube else url
 
 # Display summary if generated
 if st.session_state.summary:
@@ -407,7 +403,7 @@ if st.session_state.summary:
         st.write(st.session_state.summary)
 
 # Query input with Ask button (only for web URLs)
-if st.session_state.url_content and not youtube_url.strip():
+if st.session_state.url_content and st.session_state.content_type == "web":
     with main_container:
         st.header("Ask a Question")
         query = st.text_input("Question", placeholder="e.g., What is the article about?")