| 12345678910111213141516171819202122232425262728 |
- def anonymizeEmail(text_to_anonymize, country):
- # Initialize Faker
- fake = Faker(faker_locales_dict[country])
- fake.add_provider(internet)
- analyzer_results = analyzer.analyze(text=text_to_anonymize, entities=["EMAIL_ADDRESS"], language='en')
- matched_emails = {}
- for match in analyzer_results:
- email = text_to_anonymize[match.start:match.end]
- if email not in emails_dict:
- fake_email = fake.safe_email()
- while (fake_email in emails_dict.values()) or (fake_email in emails_dict):
- fake_email = fake.safe_email()
- emails_dict[email] = fake_email
- matched_emails[email] = fake_email
- else:
- fake_email = emails_dict[email]
- matched_emails[email] = fake_email
- anonymized_result = text_to_anonymize
- for email in matched_emails:
- anonymized_result = anonymized_result.replace(email, matched_emails[email])
- return anonymized_result
|
