"""Provenance / license classification for norm references.

Encodes the BreakPilot mapping methodology (idea/expression): a *reference* to
where a topic sits in a standard is a fact and citable; the paid normative *text*
is never stored or reproduced. This classifier marks which sources are freely
usable (EU law, US-gov public domain, open licenses) vs. paid standards we may
only REFERENCE by clause/control ID.

See docs-src/development/mapping-methodology.md.
"""

LAW = "eu_law"                  # EU legislation — public, reproducible (EUR-Lex)
PUBLIC_DOMAIN = "public_domain"  # e.g. NIST (US gov work) — reproducible
OPEN = "open"                   # OWASP (CC), ETSI EN 303 645, BSI — freely available
PAID_REFERENCE = "paid_reference"  # ISO/IEC/EN/DIN — REFERENCE ONLY, no text stored

LABEL = {
    LAW: "EU-Recht (frei)",
    PUBLIC_DOMAIN: "Public Domain (frei)",
    OPEN: "offen lizenziert",
    PAID_REFERENCE: "kostenpflichtige Norm — nur Verweis",
}

_LAW = ("2024/2847", "2023/1230", "verordnung (eu)", "maschinenverordnung", "(cra)",
        "anhang", "nis2", "nis-2", " art. ", "dsgvo", "2016/679", "2022/2555")
_PUBLIC = ("nist", "ntia", "nvd", "cisa")
_OPEN = ("owasp", "slsa", "etsi en 303 645", "bsi", "cyclonedx", "spdx",
         "nist privacy framework")
_PAID = ("iso", "iec", "din", "en iso", "62443", "27002", "27035", "29147",
         "30111", "15408", "18045", "13849", "13850", "13857", "14119", "14120",
         "61496", "61800", "62061", "60204", "82079", "15066", "10218", "13855", "62061")


def classify_norm_ref(ref: str) -> str:
    r = (ref or "").lower()
    # NIST Privacy Framework is open-ish; keep public-domain check after open guard.
    if "nist privacy" in r:
        return OPEN
    if any(k in r for k in _LAW):
        return LAW
    if any(k in r for k in _PUBLIC):
        return PUBLIC_DOMAIN
    if any(k in r for k in _OPEN):
        return OPEN
    if any(k in r for k in _PAID):
        return PAID_REFERENCE
    return PAID_REFERENCE  # conservative default: treat unknown as reference-only


def classify_refs(refs) -> list:
    """[{ref, license_class, label}] for each norm reference."""
    return [
        {"ref": r, "license_class": (lc := classify_norm_ref(r)), "label": LABEL[lc]}
        for r in (refs or [])
    ]