Update app.py

app.py

@@ -4,7 +4,7 @@ from langchain.embeddings import HuggingFaceEmbeddings
 from langchain.chains import RetrievalQA
 from langchain_community.vectorstores import FAISS
 from langchain_core.prompts import PromptTemplate
-from langchain_community.llms import HuggingFaceEndpoint  # Corrected import
+from langchain_community.llms import HuggingFaceEndpoint
 import time
 import translators as ts
 from huggingface_hub import hf_hub_download
@@ -13,9 +13,9 @@ from huggingface_hub import hf_hub_download
 st.set_page_config(layout="wide")
 
 # Configuration
-HF_TOKEN = os.getenv("HF_TOKEN")  # From Spaces secrets
-VECTORSTORE_REPO_ID = "vashu2425/bhagavad-geeta-faiss-vectordb"  # Replace with your repo
-MODEL_REPO_ID = "mistralai/Mixtral-8x7B-Instruct-v0.1"  # Updated model
+HF_TOKEN = os.getenv("HF_TOKEN")  # First try to get from Spaces secrets
+VECTORSTORE_REPO_ID = "vashu2425/bhagavad-geeta-faiss-vectordb"
+MODEL_REPO_ID = "mistralai/Mixtral-8x7B-Instruct-v0.1"
 
 # Constants
 CUSTOM_PROMPT_TEMPLATE = """
@@ -30,6 +30,25 @@ Start The Answer Directly. No Small Talk, Please. The Answer Should Contain All
 Consider Yourself As God Krishna And Answer The Question
 """
 
+# Add manual token input if not found in environment
+if not HF_TOKEN:
+    HF_TOKEN = st.sidebar.text_input("Enter Hugging Face Token (hf_...)", type="password")
+    if not HF_TOKEN:
+        st.warning("Please enter your Hugging Face token to continue")
+        st.stop()
+
+def initialize_session_states():
+    session_defaults = {
+        "messages": [],
+        "selected_question": None,
+        "show_predefined": True,
+        "last_response": None,
+        "translation_done": False
+    }
+    for key, val in session_defaults.items():
+        if key not in st.session_state:
+            st.session_state[key] = val
+
 def translate_text(text, dest_language="hi"):
     try:
         return ts.google(text, to_language=dest_language)
@@ -41,13 +60,10 @@ def translate_text(text, dest_language="hi"):
 def get_vectorstore():
     embedding_model = HuggingFaceEmbeddings(model_name='sentence-transformers/all-MiniLM-L6-v2')
     
-    # Create directory structure
     os.makedirs("vectorstore/db_faiss", exist_ok=True)
     
-    # List of required FAISS files
     faiss_files = ["index.faiss", "index.pkl"]
     
-    # Download files from Hugging Face Hub
     for filename in faiss_files:
         if not os.path.exists(f"vectorstore/db_faiss/{filename}"):
             try:
@@ -71,11 +87,85 @@ def load_llm(huggingface_repo_id, hf_token):
     return HuggingFaceEndpoint(
         repo_id=huggingface_repo_id,
         temperature=0.5,
-        huggingfacehub_api_token=hf_token,  # Correct parameter
-        model_kwargs={"max_length": 512}    # Correct data type
+        huggingfacehub_api_token=hf_token,
+        model_kwargs={"max_length": 512}
     )
 
-# ... (keep other functions the same as in your original code)
+def format_source_docs(source_documents):
+    formatted_docs = []
+    for idx, doc in enumerate(source_documents, start=1):
+        content = doc.page_content.replace('\t', ' ').replace('\n', ' ').strip()
+        formatted_doc = f"**Source {idx}** (Page {doc.metadata['page']}):\n\n{content[:500]}..." 
+        formatted_docs.append(formatted_doc)
+    return "\n\n".join(formatted_docs)
+
+def render_chat_messages():
+    for message in st.session_state.messages:
+        with st.chat_message(message["role"], avatar="🐿" if message["role"] == "user" else "🪈"):
+            content = message["content"]
+            if "hindi-text" in content:
+                st.markdown(content, unsafe_allow_html=True)
+            else:
+                st.markdown(content)
+
+def render_predefined_questions():
+    predefined_questions = [
+        "Meaning of Dharma?",
+        "What is the purpose of life?",
+        "How to find inner peace?",
+        "How can I be a better person?",
+        "What is the meaning of life?",
+        "How can I be a better friend?"
+    ]
+    st.markdown("### Or, try one of these:")
+    buttons = st.columns(len(predefined_questions))
+    for idx, question in enumerate(predefined_questions):
+        if buttons[idx].button(question, key=f"predefined_{idx}"):
+            st.session_state.selected_question = question
+            st.session_state.show_predefined = False
+
+def handle_user_input(prompt, qa_chain):
+    if prompt:
+        with st.chat_message("user", avatar="🐿"):
+            st.markdown(prompt)
+        st.session_state.messages.append({"role": "user", "content": prompt})
+
+        try:
+            response = qa_chain.invoke({"query": prompt})
+            result = response["result"]
+            source_documents = response["source_documents"]
+
+            response_placeholder = st.empty()
+            accumulated_text = ""
+            for char in result:
+                accumulated_text += char
+                response_placeholder.markdown(f'<div class="english-text">{accumulated_text}</div>', unsafe_allow_html=True)
+                time.sleep(0.01)
+            
+            if "don't have information" not in result.lower():
+                with st.expander("Source Documents"):
+                    st.markdown(format_source_docs(source_documents))
+
+            st.session_state.messages.append({"role": "assistant", "content": accumulated_text})
+            st.session_state.last_response = accumulated_text
+            st.session_state.show_predefined = False
+            st.session_state.translation_done = False
+
+        except Exception as e:
+            st.error(f"Error: {str(e)}")
+
+def handle_translation():
+    if st.session_state.last_response and not st.session_state.translation_done:
+        try:
+            translated_text = translate_text(st.session_state.last_response, "hi")
+            for msg in reversed(st.session_state.messages):
+                if msg["role"] == "assistant":
+                    msg["content"] = f'<div class="hindi-text">{translated_text}</div>'
+                    break
+            st.session_state.translation_done = True
+            st.rerun()
+        except Exception as e:
+            st.error(f"Translation error: {str(e)}")
 
 def main():
     st.markdown("""