"""Data validation functions for ICD codes and KVNR."""

import re

ICD_PATTERN = re.compile(r"^[A-Z]\d{2}(\.\d{1,2})?$")
KVNR_PATTERN = re.compile(r"^[A-Z]\d{9}$")


def validate_icd(code: str) -> str:
    """Validate and normalize a single ICD code. Returns uppercase stripped code."""
    code = code.strip().upper()
    if not code:
        raise ValueError("Empty ICD code")
    if not ICD_PATTERN.match(code):
        raise ValueError(f"Invalid ICD code format: '{code}'")
    return code


def split_icd_codes(raw: str) -> list[str]:
    """Split a string of multiple ICD codes (comma or semicolon separated)."""
    if not raw or not raw.strip():
        return []
    # Split by comma, semicolon, or whitespace+comma combinations
    codes = re.split(r"[,;]\s*", raw.strip())
    return [c.strip() for c in codes if c.strip()]


def normalize_icd_hauptgruppe(code: str) -> str:
    """Extract hauptgruppe from ICD code: 'C50.1' -> 'C50'."""
    code = code.strip().upper()
    return code[:3] if len(code) >= 3 else code


def validate_kvnr(kvnr: str) -> str:
    """Validate KVNR format (letter + 9 digits). Returns stripped uppercase."""
    kvnr = kvnr.strip().upper()
    if not kvnr:
        raise ValueError("Empty KVNR")
    if not KVNR_PATTERN.match(kvnr):
        raise ValueError(f"Invalid KVNR format: '{kvnr}'")
    return kvnr