duplicate_checker.py aktualisiert

duplicate_checker.py

@@ -3,7 +3,6 @@ import sys
 import logging
 import pandas as pd
 from datetime import datetime
-import tldextract
 from thefuzz import fuzz
 from helpers import normalize_company_name, simple_normalize_url
 from google_sheet_handler import GoogleSheetHandler
@@ -11,14 +10,14 @@ from google_sheet_handler import GoogleSheetHandler
 # --- Konfiguration ---
 CRM_SHEET_NAME = "CRM_Accounts"
 MATCHING_SHEET_NAME = "Matching_Accounts"
-SCORE_THRESHOLD = 80  # ab hier automatisches Match
+SCORE_THRESHOLD = 80  # Score ab hier gilt als Match
 LOG_DIR = "Log"
 
 # --- Logging Setup mit Datum im Dateinamen ---
 if not os.path.exists(LOG_DIR):
     os.makedirs(LOG_DIR, exist_ok=True)
 now = datetime.now().strftime('%Y-%m-%d_%H-%M')
-log_path = os.path.join(LOG_DIR, f"{now}_Duplicate.txt")
+log_path = os.path.join(LOG_DIR, f"{now}_Duplicate.log")
 
 logger = logging.getLogger(__name__)
 logger.setLevel(logging.DEBUG)
@@ -39,28 +38,27 @@ logger.info(f"Logging in Datei: {log_path}")
 
 
 def calculate_similarity(record1, record2):
-    """Berechnet gewichteten Ähnlichkeits-Score (0–190) zwischen zwei Datensätzen."""
+    """Berechnet gewichteten Ähnlichkeits-Score zwischen zwei Datensätzen."""
     total = 0
-    # Domain-Check über registered domain
-    url1 = record1.get('CRM Website','')
-    url2 = record2.get('CRM Website','')
-    dom1 = tldextract.extract(url1).registered_domain or ''
-    dom2 = tldextract.extract(url2).registered_domain or ''
+    # Domain exact match über normalisierte Domain
+    dom1 = record1.get('normalized_domain', '')
+    dom2 = record2.get('normalized_domain', '')
     if dom1 and dom1 == dom2:
         total += 100
-    # Name-Fuzzy
-    name1 = record1['normalized_name']
-    name2 = record2['normalized_name']
+    # Name fuzzy (Token-Set Ratio)
+    name1 = record1.get('normalized_name', '')
+    name2 = record2.get('normalized_name', '')
     if name1 and name2:
-        total += fuzz.token_set_ratio(name1, name2) * 0.7
-    # Ort+Land exakt
-    if record1['CRM Ort'] == record2['CRM Ort'] and record1['CRM Land'] == record2['CRM Land']:
+        name_score = fuzz.token_set_ratio(name1, name2)
+        total += name_score * 0.7
+    # Ort+Land exact
+    if record1.get('CRM Ort') == record2.get('CRM Ort') and record1.get('CRM Land') == record2.get('CRM Land'):
         total += 20
     return round(total)
 
 
 def main():
-    logger.info("Starte Duplikats-Check (v2.0) mit Datum im Lognamen und verbessertem Domain-Match")
+    logger.info("Starte Duplikats-Check (v2.0 mit Kern-Syntax nach Entwurf)")
     try:
         sheet = GoogleSheetHandler()
         logger.info("GoogleSheetHandler initialisiert")
@@ -83,7 +81,7 @@ def main():
         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 else None)
-        logger.debug(f"{label}-Normierung Beispiel: {df.iloc[0][['normalized_name','normalized_domain','block_key']].to_dict()}")
+        logger.debug(f"{label}-Beispiel nach Normalisierung: {df.iloc[0][['normalized_name','normalized_domain','block_key']].to_dict()}")
 
     # Blocking-Index
     crm_index = {}
@@ -93,23 +91,22 @@ def main():
             crm_index.setdefault(key, []).append(row)
     logger.info(f"Blocking-Index erstellt: {len(crm_index)} Keys")
 
-    # Matching
+    # Matching mit Log relevanter Kandidaten
     results = []
     total = len(match_df)
     for i, mrow in match_df.iterrows():
         key = mrow['block_key']
-        cands = crm_index.get(key, [])
-        logger.info(f"Prüfe {i+1}/{total}: '{mrow['CRM Name']}' (Key='{key}') -> {len(cands)} Kandidaten")
-        if not cands:
+        candidates = crm_index.get(key, [])
+        logger.info(f"Prüfe {i+1}/{total}: '{mrow['CRM Name']}' (Key='{key}') -> {len(candidates)} Kandidaten")
+        if not candidates:
             results.append({'Match': '', 'Score': 0})
             continue
-        scored = []
-        for crow in cands:
-            score = calculate_similarity(mrow, crow)
-            scored.append((crow['CRM Name'], score))
-        # Log relevante Kandidaten mit Score>=SCORE_THRESHOLD-20
-        relevant = [(n,s) for n,s in scored if s >= SCORE_THRESHOLD-20]
-        logger.debug(f" Relevante Kandidaten (>= {SCORE_THRESHOLD-20}): {relevant}")
+        # Scores sammeln
+        scored = [(crow['CRM Name'], calculate_similarity(mrow, crow)) for crow in candidates]
+        # Top 3 relevante Kandidaten loggen
+        top3 = sorted(scored, key=lambda x: x[1], reverse=True)[:3]
+        logger.debug(f" Top 3 Kandidaten: {top3}")
+        # Besten Treffer wählen
         best_name, best_score = max(scored, key=lambda x: x[1])
         if best_score >= SCORE_THRESHOLD:
             results.append({'Match': best_name, 'Score': best_score})
@@ -118,7 +115,7 @@ def main():
             results.append({'Match': '', 'Score': best_score})
             logger.info(f" --> Kein Match (höchster Score {best_score})")
 
-    # Ergebnis zurück in Sheet
+    # Ergebnisse zurück ins Sheet
     out = pd.DataFrame(results)
     output = pd.concat([match_df[['CRM Name','CRM Website','CRM Ort','CRM Land']].reset_index(drop=True), out], axis=1)
     data = [output.columns.tolist()] + output.values.tolist()