gdpr_filter.py

import re
import logging

logger = logging.getLogger(__name__)


class GDPRFilter:
    
    PERSONNUMMER_PATTERN = r'\b(?:19|20)?\d{6}[-–]\d{4}\b'
    EMAIL_PATTERN = r'\b[\w\.\-]+@[\w\.\-]+\.\w{2,4}\b'
    PHONE_PATTERN = r'\b(?:\+46|0)[\s\-]?\d{1,3}[\s\-]?\d{3,4}[\s\-]?\d{2,4}\b'
    DATE_PATTERNS = [
        r'\b\d{4}[-–]\d{2}[-–]\d{2}\b',
        r'\b\d{1,2}[/\-\.]\d{1,2}[/\-\.]\d{2,4}\b'
    ]
    ADDRESS_PATTERN = r'\b[A-ZÅÄÖ][a-zåäö]+(?:gatan|vägen|gränd|torget|plan|allén|stigen|backen|torg)\s\d+[A-Za-z]?\b'
    
    MEDICATION_CONTEXT_KEYWORDS = [
        r'\b(läkemedel|medicin|mediciner|drog|droger)\b',
        r'\b(dos|doser|dosering|doseras)\b',
        r'\b(mg|ml|gram|tabletter?|kapslar?|ampuller?|dropp?ar?|injektioner?)\b',
        r'\b(ta|tar|tog|tar in|skall ta|ska ta|bör ta|måste ta|börja ta|slutat ta)\b',
        r'\b(recept|receptfritt|förskrivit|ordinerat)\b',
        r'\b(apotek|apoteket|farmaceut)\b',
        r'\b(två gånger|tre gånger|en gång|fyra gånger|dagligen|morgon|kväll|middag|natt)\b',
    ]
    
    MEDICATION_PATTERNS = [
        r'(ol|in|stat|pril|zol|am|um|ase|mycin|axin|cillin)$',
    ]
    
    MEDICATION_EXCLUSIONS = [
        'Metoprolol', 'Atorvastatin', 'Losartan', 'Omeprazol',
        'Lisinopril', 'Simvastatin', 'Ramipril', 'Bisoprolol',
        'Warfarin', 'Insulin', 'Aspirin', 'Paracetamol', 'Ibuprofen',
        'Amoxicillin', 'Penicillin', 'Tetracyclin', 'Doxycyclin',
    ]
    
    @staticmethod
    def has_medication_context(text: str) -> bool:
        for pattern in GDPRFilter.MEDICATION_CONTEXT_KEYWORDS:
            if re.search(pattern, text, re.IGNORECASE):
                return True
        return False
    
    @staticmethod
    def is_likely_medication(word: str) -> bool:
        if len(word) < 6:
            return False
        for pattern in GDPRFilter.MEDICATION_PATTERNS:
            if re.search(pattern, word, re.IGNORECASE):
                return True
        return False
    
    @staticmethod
    def apply_filter(text: str) -> str:
        if not text or not text.strip():
            return ""
        
        text = re.sub(GDPRFilter.PERSONNUMMER_PATTERN, '[PERSONNR]', text)
        text = re.sub(GDPRFilter.PHONE_PATTERN, '[TELEFON]', text)
        text = re.sub(
            r'(?i)(mobil(?:nummer)?|telefon(?:nummer)?|nummer)[\s:]+(\d[\d\s\-]{6,})',
            r'\1 [TELEFON]', text
        )
        text = re.sub(GDPRFilter.EMAIL_PATTERN, '[EMAIL]', text)
        for pattern in GDPRFilter.DATE_PATTERNS:
            text = re.sub(pattern, '[DATUM]', text)
        text = re.sub(GDPRFilter.ADDRESS_PATTERN, '[ADRESS]', text)
        text = re.sub(
            r'(?i)(bor\s+(?:i|på|kvar\s+i)|i\s+stadsdelen|i\s+området)\s+([A-ZÅÄÖ][a-zåäö]{2,}(?:\s+[A-ZÅÄÖ][a-zåäö]+)?)',
            r'\1 [ORT]', text
        )
        text = re.sub(r'(\[TELEFON\])\s+\d{1,4}\b', r'\1', text)
        text = re.sub(r'(\[PERSONNR\])\s+\d{1,4}\b', r'\1', text)
        
        has_med_context = GDPRFilter.has_medication_context(text)
        
        def replace_name(match):
            word = match.group(0)
            if GDPRFilter.is_likely_medication(word):
                if has_med_context:
                    return word
                return '[LÄKEMEDEL]'
            if any(word.startswith(ex) for ex in GDPRFilter.MEDICATION_EXCLUSIONS):
                return word
            return '[NAMN]'
        
        text = re.sub(
            r'\b[A-ZÅÄÖ][a-zåäö]{2,}\s[A-ZÅÄÖ][a-zåäö]{2,}\b',
            replace_name, text
        )
        
        return text


def apply_gdpr_filter(text: str) -> str:
    return GDPRFilter.apply_filter(text)


def apply_dual_layer_gdpr(input_text: str, output_text: str):
    return apply_gdpr_filter(input_text), apply_gdpr_filter(output_text)


if __name__ == "__main__":
    tests = [
        "Patienten tar Metoprolol 50mg två gånger dagligen",
        "Patienten är allergisk mot Metoprolol",
        "Jag heter Anna och mitt personnummer är 880415-1234",
        "Erik tar Atorvastatin 40mg enligt recept och är allergisk mot Penicillin",
    ]
    
    print("\n" + "="*80)
    print("GDPR FILTER TESTS")
    print("="*80)
    
    for text in tests:
        result = apply_gdpr_filter(text)
        print(f"\nINPUT:  {text}")
        print(f"OUTPUT: {result}")
    
    print("\n" + "="*80)