chatbot/chatbot.py

import streamlit as st
from langchain_community.chat_models import ChatOllama
from langchain.schema import HumanMessage, AIMessage
from langchain.memory import ConversationBufferMemory
from langchain.chains import LLMChain
from langchain.prompts import PromptTemplate


# ---- Streamlit Setup ---- #
st.set_page_config(layout="wide")
st.markdown(
    "<h1 style='text-align: center; color: inherit;'>⎚-⎚ \nMikeyBot</h1>",
    
    unsafe_allow_html=True
)
st.markdown(
    
    """
    <style>
    .st-emotion-cache-4zpzjl {
        background-color: #bf498f;
        color: #ffffff;
    }
            
    .st-emotion-cache-jmw8un{
        background-color: #ffd1ec;
        color: #000000 ;
    }

    .st-emotion-cache-1k8897g{
        background-color: #d98dba;
    }
   
    </style>
""", unsafe_allow_html=True)

# ---- Sidebar Inputs ---- #
st.sidebar.header("⚙️ Settings")

# Dropdown for model selection
model_options = ["llama3.2", "deepseek-r1:1.5b"]
MODEL = st.sidebar.selectbox("Choose a Model", model_options, index=0)

# add advanced settings
with st.sidebar.expander("Advanced Settings"):
    temperature = st.slider("Temperature", 0.0, 1.0, 0.7)
    top_p       = st.slider("Top-P", 0.0, 1.0, 0.9)
    top_k       = st.slider("Top-K", 1, 100, 40)
    max_tokens  = st.slider("Max Tokens", 64, 2048, 512)

# max history and context size
# ----MAX_HISTORY = st.sidebar.number_input("Max History", min_value=1, max_value=10, value=2, step=1)
# ----CONTEXT_SIZE = st.sidebar.number_input("Context Size", min_value=1024, max_value=16384, value=8192, step=1024)
MAX_HISTORY = 2         
CONTEXT_SIZE = 8192   

# ---- Function to Clear Memory When Settings Change ---- #
def clear_memory():
    st.session_state.chat_history = []
    st.session_state.memory = ConversationBufferMemory(return_messages=True)  # Reset memory

# Clear memory if settings are changed
if "prev_context_size" not in st.session_state or st.session_state.prev_context_size != CONTEXT_SIZE:
    clear_memory()
    st.session_state.prev_context_size = CONTEXT_SIZE

# ---- Initialize Chat Memory ---- #
if "chat_history" not in st.session_state:
    st.session_state.chat_history = []

if "memory" not in st.session_state:
    st.session_state.memory = ConversationBufferMemory(return_messages=True)

# ---- LangChain LLM Setup ---- #
llm = ChatOllama(
    model=MODEL,
    streaming=True,
    temperature=temperature,
    top_p=top_p,
    top_k=top_k,
    num_predict=max_tokens,
    num_ctx=CONTEXT_SIZE,  
)

# for summarize button
if st.sidebar.button("Summarize Chat"):
    with st.spinner("Summarizing..."):
        # Format history nicely
        history_text = "\n".join(
            [f"{m['role']}: {m['content']}" for m in st.session_state.chat_history]
        )

        # Build prompt
        summary_prompt = [
            {"role": "system", "content": "You are a helpful assistant that summarizes conversations."},
            {"role": "user", "content": f"Please summarize this conversation:\n\n{history_text}"}
        ]

        # Call model using LangChain
        summary_result = llm.invoke(summary_prompt[1]["content"])
        summary = summary_result.content if hasattr(summary_result, 'content') else str(summary_result)

        # Save summary 
        st.session_state.chat_history.append(
            {"role": "assistant", "content": summary}
        )

        # Show in chat
        with st.chat_message("assistant"):
            st.markdown(summary)

# ---- Prompt Template ---- #
prompt_template = PromptTemplate(
    input_variables=["history", "human_input"],
    template="{history}\nUser: {human_input}\nAssistant:"
)

chain = LLMChain(llm=llm, prompt=prompt_template, memory=st.session_state.memory)

# ---- Display Chat History ---- #
for msg in st.session_state.chat_history:
    with st.chat_message(msg["role"]):
        st.markdown(msg["content"])

# ---- Trim Function (Removes Oldest Messages) ---- #
def trim_memory():
    while len(st.session_state.chat_history) > MAX_HISTORY * 2:  # Each cycle has 2 messages (User + AI)
        st.session_state.chat_history.pop(0)  # Remove oldest User message
        if st.session_state.chat_history:
            st.session_state.chat_history.pop(0)  # Remove oldest AI response

# ---- Handle User Input ---- #
if prompt := st.chat_input("Say something"):
    # Show User Input Immediately
    with st.chat_message("user"):
        st.markdown(prompt)

    st.session_state.chat_history.append({"role": "user", "content": prompt})  # Store user input

    # Trim chat history before generating response
    trim_memory()

    # ---- Get AI Response (Streaming) ---- #
    with st.chat_message("assistant"):
        response_container = st.empty()
        full_response = ""

        for chunk in chain.stream({"human_input": prompt}):
            if isinstance(chunk, dict) and "text" in chunk:
                text_chunk = chunk["text"]
                full_response += text_chunk
                response_container.markdown(full_response)

    # Store response in session_state
    st.session_state.chat_history.append({"role": "assistant", "content": full_response})

    # Trim history after storing the response
    trim_memory()