"""
Unified reference-number generator.

Format: {PREFIX}-{YYYYMM}-{HOSPITAL_TOKEN}-{SEQ:04d}
  e.g. CMP-202606-HHN-0001

  PREFIX        module prefix (CMP/INQ/OBS/APR/SGT)
  YYYYMM        creation month
  HOSPITAL_TOKEN hospital.code sanitized to uppercase alphanumerics
                 (HH-N -> HHN, E2E-HOSP -> E2EHOSP); "GEN" when no hospital
  SEQ           4-digit monthly sequence per (prefix, hospital, month)

The sequence is allocated atomically via ReferenceSequence so it is safe under
concurrent submissions. Legacy references keep their old format (no migration
of historical data); only new records use this generator.
"""

import re
from datetime import datetime

_TOKEN_RE = re.compile(r"[^A-Z0-9]")


def sanitize_hospital_token(hospital) -> str:
    """Derive a clean, format-safe token from a hospital instance."""
    code = getattr(hospital, "code", None) if hospital else None
    if not code:
        return "GEN"
    token = _TOKEN_RE.sub("", str(code).upper())
    return token or "GEN"


def generate_reference(prefix: str, hospital) -> str:
    """Generate a unified reference number for the given module prefix."""
    from apps.core.models import ReferenceSequence

    prefix = (prefix or "").upper()
    token = sanitize_hospital_token(hospital)
    year_month = datetime.now().strftime("%Y%m")
    number = ReferenceSequence.next_number(prefix, token, year_month)
    return f"{prefix}-{year_month}-{token}-{number:04d}"