app.py

import uuid

import streamlit as st
from langchain_core.messages import HumanMessage, AIMessage

from llm.openai_llm import OpenAILLM
from management import ConversationsManager
from schemas import Message
from parsers import PDFParser


pdf_text = ""
prompt = None

# Initialize session states
if "chat_id" not in st.session_state:
    st.session_state.chat_id = uuid.uuid4().hex

if "summary" not in st.session_state:
    st.session_state.summary = None

if "flashcards" not in st.session_state:
    st.session_state.flashcards = []

if "current_flashcard" not in st.session_state:
    st.session_state.current_flashcard = 0

if "messages" not in st.session_state:
    st.session_state.messages = []
#

chatId = st.session_state.chat_id or "default"


# Placeholder for the feedback message
feedback_placeholder = st.empty()
#

# Initialize the OpenAI LLM client
llm = OpenAILLM(namespace="deep-searcher", ui=feedback_placeholder)
#

# Conversation Management
manager = ConversationsManager(chatId)
#


# --- Sidebar ---
with st.sidebar:
    if st.button("Novo", icon=":material/add:"):
        st.session_state.clear()
        st.rerun()

    st.write(f"_{st.session_state.chat_id}_")
    st.markdown("---")

    # Step 0: Upload PDF
    uploaded_file = st.file_uploader("Envie um PDF para extração de texto", type="pdf")

    if uploaded_file is not None:
        try:
            pdf_text = PDFParser().parse(uploaded_file.read())
            if st.button("Analisar PDF"):
                prompt = pdf_text
                pdf_text = ""
                uploaded_file = None
        except Exception as e:
            st.error(f"Erro ao ler o PDF: {e}")

    st.markdown("---")
    st.subheader("Histórico de Conversas")

#

manual_prompt = st.chat_input("Digite seu texto...")

if manual_prompt:
    prompt = manual_prompt

if prompt:
    with st.chat_message("user"):
        st.markdown(prompt)
        st.session_state.messages.append({"role": "user", "content": prompt})
        manager.add_message(
            Message(
                role="user",
                content=prompt,
            )
        )

    with st.chat_message("assistant"):
        st.session_state.summary = None
        st.session_state.flashcards = []
        st.session_state.current_flashcard = 0
        with st.spinner("Agentes gerando resumo..."):
            agent_output = llm.generate(
                [
                    HumanMessage(content=m['content'])
                    if m['role'] == "user"
                    else AIMessage(
                        content=m['content'],
                        additional_kwargs={"name": m['role']}
                    )
                    for m in st.session_state.messages
                ],
            )
            st.session_state.summary = agent_output
            st.session_state.messages.append({"role": "assistant", "content": agent_output})
            st.success("✅ Concluído!")
            manager.add_message(
                Message(
                    role="agent",
                    content=agent_output,
                )
            )

            # Write the summary
            st.write(st.session_state.summary)


if st.session_state.summary is not None and len(st.session_state.summary) >= 900:
    if st.button("Salvar", icon=":material/favorite:"):
        manager.favorite(st.session_state.summary)
        st.success("Resumo salvo com sucesso!")

    st.write("---")
    st.subheader("Crie flashcards para fixar o conteúdo.")

    quantities = st.number_input(
        "Quantas flashcards você gostaria de gerar?",
        min_value=1,
        max_value=10,
        value=5,
    )

    if st.button("Criar Flashcards"):
        with st.spinner(f"Gerando {quantities} flashcards..."):
            try:
                flashcards = llm.flashcard(st.session_state.summary, quantities)

                if isinstance(flashcards, list) and len(flashcards) > 0:
                    st.session_state.flashcards = flashcards
                    st.session_state.current_flashcard = 0
                    st.success(f"{len(flashcards)} Flashcards criados com sucesso!")
                else:
                    st.warning("Nenhum flashcard foi gerado.")
            except Exception as e:
                st.error(f"Erro ao gerar flashcards: {e}")


if st.session_state.flashcards:
    try:
        current_index = st.session_state.current_flashcard
        current_card = st.session_state.flashcards[current_index]

        st.subheader(f"Flashcard {current_index + 1} de {len(st.session_state.flashcards)}")

        if st.button("Mostrar Pergunta"):
            st.write(current_card.question)

        if st.button("Mostrar Resposta"):
            st.success(current_card.answer)

        # Navigation buttons
        col1, col2 = st.columns(2)
        with col1:
            if st.button("Voltar", disabled=current_index == 0):
                st.session_state.current_flashcard -= 1
        with col2:
            if st.button("Próximo", disabled=current_index == len(st.session_state.flashcards) - 1):
                st.session_state.current_flashcard += 1
    except IndexError:
        st.session_state.current_flashcard = 0
        st.error("Erro ao acessar o flashcard atual. Verifique se a lista de flashcards não está vazia.")
    except Exception as e:
        st.error(f"Erro inesperado: {e}")