database.py

import os
import re
import logging
from typing import List, Dict, Tuple
import chromadb
from chromadb.utils import embedding_functions
from config import EMBEDDING_MODEL, DATABASE_DIR

# Konfiguracja logowania
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')

class KodeksProcessor:
    def __init__(self):
        logging.info("Inicjalizacja klienta bazy danych...")
        self.client = chromadb.PersistentClient(path=DATABASE_DIR)
        try:
            self.collection = self.client.get_collection("kodeksy")
            logging.info("Pobrano istniejącą kolekcję 'kodeksy'.")
        except:
            self.collection = self.client.create_collection(
                name="kodeksy",
                embedding_function=embedding_functions.SentenceTransformerEmbeddingFunction(
                    model_name=EMBEDDING_MODEL
                )
            )
            logging.info("Utworzono nową kolekcję 'kodeksy'.")

    def extract_metadata(self, text: str) -> Dict:
        metadata = {}
        dz_u_match = re.search(r'Dz\.U\.(\d{4})\.(\d+)\.(\d+)', text)
        if dz_u_match:
            metadata['dz_u'] = f"Dz.U.{dz_u_match.group(1)}.{dz_u_match.group(2)}.{dz_u_match.group(3)}"
            metadata['rok'] = dz_u_match.group(1)

        nazwa_match = re.search(r'USTAWA\s+z dnia(.*?)\n(.*?)\n', text)
        if nazwa_match:
            metadata['data_ustawy'] = nazwa_match.group(1).strip()
            metadata['nazwa'] = nazwa_match.group(2).strip()

        logging.info("Wydobyto metadane: %s", metadata)
        return metadata

    def split_header_and_content(self, text: str) -> Tuple[str, str]:
        parts = text.split("USTAWA", 1)
        if len(parts) > 1:
            return parts[0], "USTAWA" + parts[1]
        return "", text

    def process_article(self, article_text: str) -> Dict:
        art_num_match = re.match(r'Art\.\s*(\d+)', article_text)
        article_num = art_num_match.group(1) if art_num_match else ""

        paragraphs = re.findall(r'§\s*(\d+)\.\s*(.*?)(?=§\s*\d+|Art\.\s*\d+|$)', article_text, re.DOTALL)

        if not paragraphs:
            return {
                "article_num": article_num,
                "content": article_text.strip(),
                "has_paragraphs": False
            }

        return {
            "article_num": article_num,
            "paragraphs": paragraphs,
            "has_paragraphs": True
        }

    def split_into_chunks(self, text: str, metadata: Dict) -> List[Dict]:
        chunks = []
        articles = re.split(r'(Art\.\s*\d+)', text)  # Podział na artykuły

        for i in range(1, len(articles), 2):  # Przechodzimy przez artykuły
            article_title = articles[i].strip()
            article_content = articles[i + 1].strip() if i + 1 < len(articles) else ""

            processed_article = self.process_article(article_title + " " + article_content)

            chunk_metadata = {
                **metadata,
                "article": processed_article["article_num"]
            }

            if processed_article["has_paragraphs"]:
                for par_num, par_content in processed_article["paragraphs"]:
                    chunks.append({
                        "text": f"{article_title} §{par_num}. {par_content.strip()}",
                        "metadata": {**chunk_metadata, "paragraph": par_num}
                    })
            else:
                chunks.append({
                    "text": processed_article["content"],
                    "metadata": chunk_metadata
                })

        logging.info("Podzielono tekst na %d chunków.", len(chunks))
        return chunks

    def process_file(self, filepath: str) -> None:
        logging.info("Przetwarzanie pliku: %s", filepath)

        with open(filepath, 'r', encoding='utf-8') as file:
            content = file.read()

        header, main_content = self.split_header_and_content(content)
        metadata = self.extract_metadata(main_content)
        metadata['filename'] = os.path.basename(filepath)

        chunks = self.split_into_chunks(main_content, metadata)

        for i, chunk in enumerate(chunks):
            self.collection.add(
                documents=[chunk["text"]],
                metadatas=[chunk["metadata"]],
                ids=[f"{metadata['filename']}_{chunk['metadata']['article']}_{i}"]
            )
            logging.info("Dodano chunk: %s", chunk["text"])  # Logowanie dodawanych chunków

        logging.info("Dodano %d chunków z pliku %s", len(chunks), metadata['filename'])

    def process_all_files(self, directory: str) -> None:
        logging.info("Rozpoczęcie przetwarzania wszystkich plików w katalogu: %s", directory)
        for filename in os.listdir(directory):
            if filename.endswith('.txt'):
                filepath = os.path.join(directory, filename)
                self.process_file(filepath)

    def search(self, query: str, n_results: int = 3) -> Dict:
        logging.info("Wyszukiwanie w bazie danych dla zapytania: %s", query)
        results = self.collection.query(
            query_texts=[query],
            n_results=n_results
        )
        logging.info("Znaleziono %d wyników dla zapytania: %s", len(results['documents'][0]), query)
        return results

    def list_all_documents(self) -> None:
        all_docs = self.collection.query(query_texts=[""], n_results=1000)  # Pobierz wszystkie dokumenty
        for doc in all_docs['documents'][0]:
            logging.info("Dokument: %s", doc)