Spaces:
Running
Running
File size: 4,775 Bytes
aabd32c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 |
import streamlit as st
import os
import logging
from processingPdf.loader import get_layout_extractor
from processingPdf.logicSections import extract_logical_sections
from processingPdf.chunker import split_sections_into_chunks
from processingPdf.indexer import Indexer
from db.graph_db import GraphDB
from dotenv import load_dotenv
# Importazione del workflow LangGraph
from agentLogic.graph import app as rag_app
# Configurazione Logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
load_dotenv()
st.set_page_config(page_title="RAG Chatbot - PDF", layout="wide")
# Inizializzazione componenti nel session_state
if "indexer" not in st.session_state:
st.session_state.indexer = Indexer()
if "layout_extractor" not in st.session_state:
st.session_state.layout_extractor = get_layout_extractor()
if "chat_history" not in st.session_state:
st.session_state.chat_history = []
if "pdf_ready" not in st.session_state:
st.session_state.pdf_ready = False
st.title("RAG Chatbot")
st.markdown("Carica un PDF per indicizzarlo su Neo4j e iniziare a chattare.")
# --- SIDEBAR: CARICAMENTO E INDICIZZAZIONE ---
with st.sidebar:
st.header("Profilo Utente")
user_name_input = st.text_input("Ciao! Inserisci il tuo nome/nickname", placeholder="es. Laura")
user_id_clean = user_name_input.strip().lower().replace(" ", "_")
if not user_id_clean:
st.warning("Inserisci un nome utente per procedere.")
else:
st.success(f"Loggato come: {user_id_clean}")
st.divider()
st.header("Caricamento PDF")
uploaded_file = st.file_uploader("Scegli un file PDF", type="pdf")
if uploaded_file is not None and user_id_clean:
if st.button("Indicizza Documento"):
with st.spinner("Elaborazione in corso..."):
try:
file_bytes = uploaded_file.getvalue()
filename = uploaded_file.name
doc = st.session_state.layout_extractor(file_bytes)
sections = extract_logical_sections(doc)
chunks = split_sections_into_chunks(sections, filename)
st.session_state.indexer.index_chunks_to_neo4j(filename, chunks, user_id_clean)
st.success(f"Ottimo {user_id_clean}! {len(chunks)} chunk indicizzati con successo.")
st.session_state.pdf_ready = True
st.session_state.current_user = user_id_clean
except Exception as e:
st.error(f"Errore durante l'indicizzazione: {e}")
logger.error(f"Errore: {e}", exc_info=True)
# --- AREA CHAT ---
st.divider()
# Visualizzazione cronologia
for message in st.session_state.chat_history:
with st.chat_message(message["role"]):
st.markdown(message["content"])
# Input utente
if prompt := st.chat_input("Fai una domanda sul documento..."):
st.session_state.chat_history.append({"role": "user", "content": prompt})
with st.chat_message("user"):
st.markdown(prompt)
with st.chat_message("assistant"):
response_placeholder = st.empty()
if not st.session_state.pdf_ready:
full_response = "Per favore, carica e indicizza un PDF prima di iniziare la chat."
response_placeholder.markdown(full_response)
else:
with st.spinner("Ricerca e generazione risposta in corso..."):
try:
# Preparazione dello stato iniziale per il grafo
initial_state = {
"query": prompt,
"user_id": st.session_state.current_user,
"filename": uploaded_file.name,
"intent_data": {},
"context_chunks": [],
"final_answer": ""
}
# Esecuzione del workflow LangGraph
# Usiamo invoke per ottenere il risultato finale dopo che tutti i nodi sono stati processati
result = rag_app.invoke(initial_state)
full_response = result.get("final_answer", "Non sono riuscito a generare una risposta.")
response_placeholder.markdown(full_response)
except Exception as e:
full_response = f"Errore durante l'elaborazione della domanda: {e}"
response_placeholder.error(full_response)
logger.error(f"Errore Chat: {e}", exc_info=True)
st.session_state.chat_history.append({"role": "assistant", "content": full_response}) |