Brancheneinstufung2/duplicate_checker.py

# duplicate_checker.py (v2.0 + Transparenz)

import logging
import pandas as pd
from thefuzz import fuzz
from config import Config
from helpers import normalize_company_name, simple_normalize_url, create_log_filename
from google_sheet_handler import GoogleSheetHandler
import time

# --- Konfiguration ---
CRM_SHEET_NAME = "CRM_Accounts"
MATCHING_SHEET_NAME = "Matching_Accounts"
SCORE_THRESHOLD = 80

# --- VOLLSTÄNDIGES LOGGING SETUP ---
LOG_LEVEL = logging.DEBUG if Config.DEBUG else logging.INFO
LOG_FORMAT = '%(asctime)s - %(levelname)-8s - %(name)s - %(message)s'

root_logger = logging.getLogger()
root_logger.setLevel(LOG_LEVEL)

# Handler nur hinzufügen, wenn noch keine konfiguriert sind, um Dopplung zu vermeiden
if not root_logger.handlers:
    stream_handler = logging.StreamHandler()
    stream_handler.setFormatter(logging.Formatter(LOG_FORMAT))
    root_logger.addHandler(stream_handler)

    log_file_path = create_log_filename("duplicate_check_v2_final")
    if log_file_path:
        file_handler = logging.FileHandler(log_file_path, mode='a', encoding='utf-8')
        file_handler.setFormatter(logging.Formatter(LOG_FORMAT))
        root_logger.addHandler(file_handler)
else:
    log_file_path = next((h.baseFilename for h in root_logger.handlers if isinstance(h, logging.FileHandler)), None)

logger = logging.getLogger(__name__)

def calculate_similarity_with_details(record1, record2):
    """
    Berechnet einen gewichteten Ähnlichkeits-Score und gibt den Score und den Grund zurück.
    Basierend auf der v2.0 Scoring-Logik.
    """
    scores = {'name': 0, 'location': 0, 'domain': 0}
    
    domain1 = record1.get('normalized_domain')
    domain2 = record2.get('normalized_domain')
    if domain1 and domain1 != 'k.a.' and domain1 == domain2:
        scores['domain'] = 100
    
    name1 = record1.get('normalized_name')
    name2 = record2.get('normalized_name')
    if name1 and name2:
        name_similarity = fuzz.token_set_ratio(name1, name2)
        scores['name'] = round(name_similarity * 0.7)

    ort1 = record1.get('CRM Ort')
    ort2 = record2.get('CRM Ort')
    land1 = record1.get('CRM Land')
    land2 = record2.get('CRM Land')
    if ort1 and ort1 == ort2 and land1 and land1 == land2:
        scores['location'] = 20
            
    total_score = sum(scores.values())
    
    reasons = []
    if scores['domain'] > 0: reasons.append(f"Domain({scores['domain']})")
    if scores['name'] > 0: reasons.append(f"Name({scores['name']})")
    if scores['location'] > 0: reasons.append(f"Ort({scores['location']})")
    reason_text = " + ".join(reasons) if reasons else "Keine Übereinstimmung"

    return round(total_score), reason_text

def main():
    """Hauptfunktion zum Laden, Vergleichen und Schreiben der Daten."""
    start_time = time.time()
    logger.info("Starte den Duplikats-Check (v2.0 mit Blocking und Maximum Logging)...")
    logger.info(f"Logdatei: {log_file_path}")
    
    try:
        sheet_handler = GoogleSheetHandler()
    except Exception as e:
        logger.critical(f"FEHLER bei Initialisierung des GoogleSheetHandler: {e}")
        return

    logger.info(f"Lade Master-Daten aus '{CRM_SHEET_NAME}'...")
    crm_df = sheet_handler.get_sheet_as_dataframe(CRM_SHEET_NAME)
    if crm_df is None or crm_df.empty: return

    logger.info(f"Lade zu prüfende Daten aus '{MATCHING_SHEET_NAME}'...")
    matching_df = sheet_handler.get_sheet_as_dataframe(MATCHING_SHEET_NAME)
    if matching_df is None or matching_df.empty: return
    original_matching_df = matching_df.copy()

    logger.info("Normalisiere Daten für den Vergleich...")
    for df in [crm_df, matching_df]:
        df['normalized_name'] = df['CRM Name'].astype(str).apply(normalize_company_name)
        df['normalized_domain'] = df['CRM Website'].astype(str).apply(simple_normalize_url)
        df['CRM Ort'] = df['CRM Ort'].astype(str).str.lower().str.strip()
        df['CRM Land'] = df['CRM Land'].astype(str).str.lower().str.strip()
        df['block_key'] = df['normalized_name'].apply(lambda x: x.split()[0] if x and x.split() else None)

    logger.info("Erstelle Index für CRM-Daten zur Beschleunigung...")
    crm_index = {}
    crm_records = crm_df.to_dict('records')
    for record in crm_records:
        key = record['block_key']
        if key:
            if key not in crm_index:
                crm_index[key] = []
            crm_index[key].append(record)

    logger.info("Starte Matching-Prozess...")
    results = []
    
    for match_record in matching_df.to_dict('records'):
        best_score = -1
        best_match_name = ""
        best_reason = ""
        
        logger.info(f"--- Prüfe: '{match_record.get('CRM Name', 'N/A')}' ---")
        logger.debug(f"  [Normalisiert: '{match_record.get('normalized_name')}', Domain: '{match_record.get('normalized_domain')}', Key: '{match_record.get('block_key')}']")

        block_key = match_record.get('block_key')
        candidates = crm_index.get(block_key, [])
        
        if not candidates:
            logger.debug("  -> Keine Kandidaten im Index gefunden. Überspringe Vergleich.")
            results.append({
                'Potenzieller Treffer im CRM': "", 'Ähnlichkeits-Score': 0, 'Matching-Grund': "Keine Kandidaten"
            })
            continue

        logger.debug(f"  -> Vergleiche mit {len(candidates)} Kandidaten aus Block '{block_key}'.")
        
        for crm_row in candidates:
            score, reason = calculate_similarity_with_details(match_record, crm_row)
            
            if score > 0:
                logger.debug(f"    - Kandidat: '{crm_row.get('CRM Name', 'N/A')}' -> Score: {score} (Grund: {reason})")
            
            if score > best_score:
                best_score = score
                best_match_name = crm_row.get('CRM Name', 'N/A')
                best_reason = reason
        
        logger.info(f"  --> Bester Treffer: '{best_match_name}' mit Score {best_score} (Grund: {best_reason})")
        
        results.append({
            'Potenzieller Treffer im CRM': best_match_name if best_score >= SCORE_THRESHOLD else "",
            'Ähnlichkeits-Score': best_score,
            'Matching-Grund': best_reason
        })

    logging.info("Matching abgeschlossen. Schreibe Ergebnisse zurück ins Sheet...")
    result_df = pd.DataFrame(results)
    
    output_df = pd.concat([original_matching_df.reset_index(drop=True), result_df], axis=1)
    
    data_to_write = [output_df.columns.values.tolist()] + output_df.values.tolist()
    
    success = sheet_handler.clear_and_write_data(MATCHING_SHEET_NAME, data_to_write)
    if success:
        logger.info(f"Ergebnisse erfolgreich in das Tabellenblatt '{MATCHING_SHEET_NAME}' geschrieben.")
    else:
        logger.error("FEHLER beim Schreiben der Ergebnisse ins Google Sheet.")

    end_time = time.time()
    logger.info(f"Gesamtdauer des Duplikats-Checks: {end_time - start_time:.2f} Sekunden.")
    logger.info(f"===== Skript beendet =====")

if __name__ == "__main__":
    main()