Update presidio_helpers.py

presidio_helpers.py

@@ -1,21 +1,25 @@
 import spacy
 from presidio_analyzer import AnalyzerEngine, RecognizerRegistry, PatternRecognizer
 from presidio_anonymizer import AnonymizerEngine
-from presidio_anonymizer.entities import AnonymizerConfig
+from presidio_anonymizer.entities import OperatorConfig
 
 # Vytvoření českých rozpoznávačů
 def create_czech_recognizers():
     recognizers = []
     
+    # Rodné číslo
     rodne_cislo_recognizer = PatternRecognizer(supported_entity="RODNÉ_ČÍSLO", patterns=[r'\b\d{6}/\d{4}\b'])
     recognizers.append(rodne_cislo_recognizer)
     
+    # Telefonní číslo
     telefon_recognizer = PatternRecognizer(supported_entity="TELEFON", patterns=[r'\b(?:\+420)?\s?\d{3}\s?\d{3}\s?\d{3}\b'])
     recognizers.append(telefon_recognizer)
     
+    # Email
     email_recognizer = PatternRecognizer(supported_entity="EMAIL", patterns=[r'\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Z|a-z]{2,}\b'])
     recognizers.append(email_recognizer)
     
+    # IČO
     ico_recognizer = PatternRecognizer(supported_entity="IČO", patterns=[r'\b\d{8}\b'])
     recognizers.append(ico_recognizer)
     
@@ -23,11 +27,12 @@ def create_czech_recognizers():
 
 # Funkce pro vytvoření analyzer enginu
 def analyzer_engine():
-    nlp = spacy.load("en_core_web_sm")
+    nlp = spacy.load("cs_core_news_sm")  # Změna na český model
     
     registry = RecognizerRegistry()
     registry.load_predefined_recognizers(nlp_engine=nlp)
     
+    # Přidání českých rozpoznávačů
     czech_recognizers = create_czech_recognizers()
     for recognizer in czech_recognizers:
         registry.add_recognizer(recognizer)
@@ -45,15 +50,15 @@ def anonymize(text, anonymization_method="replace"):
     
     # Zvolení metody anonymizace
     if anonymization_method == "replace":
-        anonymizer_config = {"DEFAULT": AnonymizerConfig("replace", {"new_value": "<ANONYMIZED>"})}
+        operators = {"DEFAULT": OperatorConfig("replace", {"new_value": "<ANONYMIZED>"})}
     elif anonymization_method == "mask":
-        anonymizer_config = {"DEFAULT": AnonymizerConfig("mask", {"masking_char": "*", "chars_to_mask": 4, "from_end": True})}
+        operators = {"DEFAULT": OperatorConfig("mask", {"masking_char": "*", "chars_to_mask": 4})}
     elif anonymization_method == "hash":
-        anonymizer_config = {"DEFAULT": AnonymizerConfig("hash")}
+        operators = {"DEFAULT": OperatorConfig("hash", {"hash_type": "sha256"})}
     else:
         raise ValueError("Neznámá metoda anonymizace")
     
     # Anonymizuj rozpoznané údaje
-    anonymized_text = anonymizer.anonymize(text=text, analyzer_results=analyzer_results, anonymizers_config=anonymizer_config)
+    anonymized_text = anonymizer.anonymize(text=text, analyzer_results=analyzer_results, operators=operators)
     
     return anonymized_text