app.py

import uuid
import os
import sys
import tempfile
import streamlit as st
from langchain_core.messages import HumanMessage, AIMessage, ToolMessage

sys.path.insert(0, os.path.dirname(__file__))

st.set_page_config(
    page_title="Research Assistant",
    page_icon="🔬",
    layout="wide",
)

if "thread_id" not in st.session_state:
    st.session_state.thread_id = str(uuid.uuid4())
if "messages" not in st.session_state:
    st.session_state.messages = []
if "threads" not in st.session_state:
    st.session_state.threads = [st.session_state.thread_id]
if "titles" not in st.session_state:
    st.session_state.titles = {}
if "last_meta" not in st.session_state:
    st.session_state.last_meta = {}
if "docs_ingested" not in st.session_state:
    st.session_state.docs_ingested = []

@st.cache_resource(show_spinner="Loading agent...")
def load_agent():
    from src.agents.graph import get_graph
    return get_graph()

with st.sidebar:
    st.title("🔬 Research Assistant")
    st.caption("LangGraph · RAG · Multi-Agent")
    st.divider()

    st.subheader("📚 Upload Documents")
    uploaded = st.file_uploader(
        "Upload PDFs",
        type=["pdf"],
        accept_multiple_files=True,
    )

    if st.button("📥 Index Documents", type="primary"):
        if uploaded:
            from src.agents.graph import ingest_pdf
            bar = st.progress(0)
            for i, pdf in enumerate(uploaded):
                meta = ingest_pdf(
                    pdf.getvalue(),
                    thread_id=st.session_state.thread_id,
                    filename=pdf.name,
                )
                st.session_state.docs_ingested.append(pdf.name)
                bar.progress((i + 1) / len(uploaded))
            st.success(f"✅ Indexed {len(uploaded)} document(s)")
        else:
            st.warning("Please upload PDFs first.")

    if st.session_state.docs_ingested:
        st.caption(f"{len(st.session_state.docs_ingested)} doc(s) indexed")
        for name in st.session_state.docs_ingested[-5:]:
            st.caption(f"  📄 {name}")

    st.divider()

    st.subheader("💬 Chats")
    if st.button("➕ New Chat"):
        new_id = str(uuid.uuid4())
        st.session_state.thread_id = new_id
        st.session_state.messages = []
        st.session_state.threads.insert(0, new_id)
        st.rerun()

    for tid in st.session_state.threads[:10]:
        title = st.session_state.titles.get(tid, f"Chat {tid[:6]}")
        active = "▶ " if tid == st.session_state.thread_id else "   "
        if st.button(f"{active}{title}", key=f"t_{tid}", use_container_width=True):
            st.session_state.thread_id = tid
            try:
                graph = load_agent()
                state = graph.get_state({"configurable": {"thread_id": tid}})
                st.session_state.messages = [
                    {
                        "role": "user" if isinstance(m, HumanMessage) else "assistant",
                        "content": m.content,
                    }
                    for m in state.values.get("messages", [])
                    if isinstance(m, (HumanMessage, AIMessage)) and m.content
                ]
            except Exception:
                st.session_state.messages = []
            st.rerun()

st.title("🔬 Research Assistant")

if st.session_state.last_meta:
    m = st.session_state.last_meta
    model = m.get("model_used", "")
    label = "⚡ Fast (8B)" if "8b" in model else "🧠 Smart (70B)"
    qtype = m.get("query_type", "?")
    st.caption(f"{label} · {qtype} · {m.get('latency_ms', 0):.0f}ms")

for msg in st.session_state.messages:
    with st.chat_message(msg["role"]):
        st.markdown(msg["content"])

user_input = st.chat_input("Ask anything about your documents...")

if user_input:
    if st.session_state.thread_id not in st.session_state.titles:
        st.session_state.titles[st.session_state.thread_id] = user_input[:28]

    st.session_state.messages.append({"role": "user", "content": user_input})
    with st.chat_message("user"):
        st.markdown(user_input)

    graph = load_agent()
    config = {"configurable": {"thread_id": st.session_state.thread_id}}

    with st.chat_message("assistant"):
        tool_box = [None]

        def stream():
            for chunk, _ in graph.stream(
                {"messages": [HumanMessage(content=user_input)]},
                config=config,
                stream_mode="messages",
            ):
                if isinstance(chunk, ToolMessage) and tool_box[0] is None:
                    tool_box[0] = st.status("🔧 Using tools...", expanded=True)
                if isinstance(chunk, AIMessage) and chunk.content:
                    content = chunk.content
                    for prefix in ["simple", "complex", "calc"]:
                        if content.lower().startswith(prefix):
                            content = content[len(prefix):]
                    yield content

        answer = st.write_stream(stream())

        if tool_box[0]:
            tool_box[0].update(label="✅ Done", state="complete", expanded=False)

    try:
        state = graph.get_state(config)
        st.session_state.last_meta = {
            "query_type": state.values.get("query_type"),
            "model_used": state.values.get("model_used"),
            "latency_ms": state.values.get("latency_ms"),
        }
    except Exception:
        pass

    st.session_state.messages.append({"role": "assistant", "content": answer})
    st.rerun()