Delete streamlit_app.py

streamlit_app.py

@@ -1,229 +0,0 @@
-import streamlit as st
-import os
-import requests
-import re  # To help clean up leading whitespace
-import sys
-
-# Prefer pysqlite3 (if installed) for SQLite; otherwise fall back to stdlib sqlite3.
-# This mirrors the previous behavior but is robust when pysqlite3 isn't available.
-try:
-    # Try to import the pysqlite3 package which provides a replacement sqlite3 module
-    pysqlite3 = __import__('pysqlite3')
-    # If imported, make it available under the name 'sqlite3' to satisfy code expecting
-    # a sqlite3 module with pysqlite3's behavior.
-    if 'sqlite3' not in sys.modules:
-        sys.modules['sqlite3'] = pysqlite3
-except ModuleNotFoundError:
-    # pysqlite3 isn't installed; the standard library's sqlite3 will be used.
-    import sqlite3  # noqa: F401
-
-# Langchain and HuggingFace
-from langchain_community.vectorstores import Chroma
-from langchain_community.embeddings import HuggingFaceEmbeddings
-# Workaround: some versions of the `groq` package (used by `langchain_groq`) may pass
-# a `proxies` keyword into the underlying Client.__init__, which can cause
-# TypeError if the installed groq Client doesn't accept that kwarg. To make the
-# app more tolerant across versions, monkeypatch the groq Client constructor to
-# silently drop `proxies` if present.
-try:
-    import groq._base_client as _groq_base
-    _groq_client_init = getattr(_groq_base.Client, "__init__", None)
-    if _groq_client_init is not None:
-        def _client_init_no_proxies(self, *args, **kwargs):
-            kwargs.pop('proxies', None)
-            return _groq_client_init(self, *args, **kwargs)
-        _groq_base.Client.__init__ = _client_init_no_proxies
-except Exception:
-    # If groq isn't installed or the internal structure differs, let the import
-    # of ChatGroq attempt to initialize and raise its own errors. We don't
-    # want to crash here just for the monkeypatch.
-    pass
-
-from langchain_groq import ChatGroq
-from langchain.chains import RetrievalQA
-
-# Load the .env file (if using it)
-groq_api_key = os.getenv("GROQ_API_KEY")
-
-# Load embeddings, model, and vector store
-@st.cache_resource  # Singleton, prevent multiple initializations
-def init_chain():
-    model_kwargs = {'trust_remote_code': True}
-    embedding = HuggingFaceEmbeddings(model_name='nomic-ai/nomic-embed-text-v1.5', model_kwargs=model_kwargs)
-    llm = ChatGroq(groq_api_key=groq_api_key, model_name="openai/gpt-oss-20b", temperature=0.1)
-    vectordb = Chroma(persist_directory='updated_CSPCDB2', embedding_function=embedding)
-
-    # Create chain
-    chain = RetrievalQA.from_chain_type(llm=llm,
-                                        chain_type="stuff",
-                                        retriever=vectordb.as_retriever(k=5),
-                                        return_source_documents=True)
-    return chain
-
-# Streamlit app layout
-st.set_page_config(
-    page_title="CSPC Citizens Charter Conversational Agent",
-    page_icon="cspclogo.png"
-)
-
-# Custom CSS for styling
-st.markdown(
-    """
-    <style>
-    .subtitle {text-align: center; font-size: 18px; font-weight: bold;}
-    .details {font-size: 14px; color: #bbbbbb; padding-left: 20px; margin-bottom: -10px;}
-    .team {text-align: center; font-size: 20px; font-weight: bold; color: #777;}
-    </style>
-    """,
-    unsafe_allow_html=True
-)
-
-with st.sidebar:
-    # App title
-    st.title('CSPC Conversational Agent')
-    st.markdown('<p class="subtitle">Your go-to assistant for the Citizen’s Charter of CSPC!</p>', unsafe_allow_html=True)
-    
-    # Categories
-    st.markdown('''✔️**About CSPC:**''')
-    st.markdown('<p class="details">History, Core Values, Mission and Vision</p>', unsafe_allow_html=True)
-    
-    st.markdown('''✔️**Admission & Graduation:**''')
-    st.markdown('<p class="details">Apply, Requirements, Process, Graduation</p>', unsafe_allow_html=True)
-
-    st.markdown('''✔️**Student Services:**''')
-    st.markdown('<p class="details">Scholarships, Orgs, Facilities</p>', unsafe_allow_html=True)
-
-    st.markdown('''✔️**Academics:**''')
-    st.markdown('<p class="details">Degrees, Courses, Faculty</p>', unsafe_allow_html=True)
-
-    st.markdown('''✔️**Officials:**''')
-    st.markdown('<p class="details">President, VPs, Deans, Admin</p>', unsafe_allow_html=True)
-
-    # Links to resources
-    st.markdown("### 🔗 Quick Access to Resources")
-    st.markdown(
-        """
-        📄 [CSPC Citizen’s Charter](https://cspc.edu.ph/governance/citizens-charter/)  
-        🏛️ [About CSPC](https://cspc.edu.ph/about/)  
-        📋 [College Officials](https://cspc.edu.ph/college-officials/)  
-        """,
-        unsafe_allow_html=True
-    )
-
-# Store LLM generated responses
-if "messages" not in st.session_state:
-    st.session_state.chain = init_chain()
-    st.session_state.messages = [{"role": "assistant", "content": "Hello! I am your Conversational Agent for the Citizens Charter of Camarines Sur Polytechnic Colleges (CSPC). How may I assist you today?"}]
-    st.session_state.query_counter = 0  # Track the number of user queries
-    st.session_state.conversation_history = ""  # Keep track of history for the LLM
-
-def generate_response(prompt_input):
-    try:
-        # Retrieve vector database context using ONLY the current user input
-        retriever = st.session_state.chain.retriever
-        relevant_context = retriever.get_relevant_documents(prompt_input)  # Retrieve context only for the current prompt
-        
-        # Format the input for the chain with the retrieved context
-        formatted_input = (
-            f"You are a Conversational Agent for the Citizens Charter of Camarines Sur Polytechnic Colleges (CSPC). "
-            f"Your purpose is to provide accurate and helpful information about CSPC's policies, procedures, and services as outlined in the Citizens Charter. "
-            f"When responding to user queries:\n"
-            f"1. Always prioritize information from the provided context (Citizens Charter or other CSPC resources).\n"
-            f"2. Be concise, clear, and professional in your responses.\n"
-            f"3. If the user's question is outside the scope of the Citizens Charter, politely inform them and suggest relevant resources or departments they can contact.\n\n"
-            f"Context:\n"
-            f"{' '.join([doc.page_content for doc in relevant_context])}\n\n"
-            f"Conversation:\n{st.session_state.conversation_history}user: {prompt_input}\n"
-        )
-    
-        # Invoke the RetrievalQA chain directly with the formatted input
-        res = st.session_state.chain.invoke({"query": formatted_input})
-        
-        # Process the response text
-        result_text = res['result']
-        
-        # Clean up prefixing phrases and capitalize the first letter
-        if result_text.startswith('According to the provided context, '):
-            result_text = result_text[35:].strip()
-        elif result_text.startswith('Based on the provided context, '):
-            result_text = result_text[31:].strip()
-        elif result_text.startswith('According to the provided text, '):
-            result_text = result_text[34:].strip()
-        elif result_text.startswith('According to the context, '):
-            result_text = result_text[26:].strip()
-        
-        # Ensure the first letter is uppercase
-        result_text = result_text[0].upper() + result_text[1:] if result_text else result_text
-    
-        # Extract and format sources (if available)
-        sources = []
-        for doc in relevant_context:
-            source_path = doc.metadata.get('source', '')
-            formatted_source = source_path[122:-4] if source_path else "Unknown source"
-            sources.append(formatted_source)
-        
-        # Remove duplicates and combine into a single string
-        unique_sources = list(set(sources))
-        source_list = ", ".join(unique_sources)
-        
-        # # Combine response text with sources
-        # result_text += f"\n\n**Sources:** {source_list}" if source_list else "\n\n**Sources:** None"
-        
-        # Update conversation history
-        st.session_state.conversation_history += f"user: {prompt_input}\nassistant: {result_text}\n"
-    
-        return result_text
-
-    except Exception as e:
-        # Handle rate limit or other errors gracefully
-        if "rate_limit_exceeded" in str(e).lower():
-            return "⚠️ Rate limit exceeded. Please clear the chat history and try again."
-        else:
-            return f"❌ An error occurred: {str(e)}"
-
-
-# Display chat messages
-for message in st.session_state.messages:
-    with st.chat_message(message["role"]):
-        st.write(message["content"])
-
-# User-provided prompt for input box
-if prompt := st.chat_input(placeholder="Ask a question..."):
-    # Increment query counter
-    st.session_state.query_counter += 1
-    # Append user query to session state
-    st.session_state.messages.append({"role": "user", "content": prompt})
-    with st.chat_message("user"):
-        st.write(prompt)
-
-    # Generate and display placeholder for assistant response
-    with st.chat_message("assistant"):
-        message_placeholder = st.empty()  # Placeholder for response while it's being generated
-        with st.spinner("Generating response..."):
-            # Use conversation history when generating response
-            response = generate_response(prompt)
-        message_placeholder.markdown(response)  # Replace placeholder with actual response
-        st.session_state.messages.append({"role": "assistant", "content": response})
-
-    # Check if query counter has reached the limit
-    if st.session_state.query_counter >= 10:
-        st.sidebar.warning("Conversation context has been reset after 10 queries.")
-        st.session_state.query_counter = 0  # Reset the counter
-        st.session_state.conversation_history = ""  # Clear conversation history for the LLM
-
-# Clear chat history function
-def clear_chat_history():
-    # Clear chat messages (reset the assistant greeting)
-    st.session_state.messages = [{"role": "assistant", "content": "Hello! I am your Conversational Agent for the Citizens Charter of Camarines Sur Polytechnic Colleges (CSPC). How may I assist you today?"}]
-    
-    # Reinitialize the chain to clear any stored history (ensures it forgets previous user inputs)
-    st.session_state.chain = init_chain()
-
-    # Clear the query counter and conversation history
-    st.session_state.query_counter = 0
-    st.session_state.conversation_history = ""
-
-st.sidebar.button('Clear Chat History', on_click=clear_chat_history)
-
-# Footer
-st.sidebar.markdown('<p class="team">Developed by Team XceptionNet</p>', unsafe_allow_html=True)