"""Customer implementation claim normaliser (spec §4.6).

Turns a free-text statement ("Wir haben einen Update-Prozess.") into structured
capabilities + related topics + weakness qualifiers. Deterministic substring
matching — the claim_id is a stable hash so the same statement always maps to
the same id (no randomness, replay-safe).
"""

from __future__ import annotations

import hashlib
from typing import List, Optional

from .schemas import CustomerImplementationClaim
from .taxonomy_claims import match_capabilities, match_qualifiers, topics_for


def _claim_id(raw_statement: str) -> str:
    digest = hashlib.sha1(raw_statement.strip().lower().encode("utf-8")).hexdigest()
    return "claim_%s" % digest[:10]


def _normalized(capabilities: List[str], qualifiers: List[str]) -> str:
    if not capabilities:
        return "Keine bekannte Compliance-Fähigkeit aus der Aussage ableitbar."
    text = "Fähigkeiten: " + ", ".join(capabilities)
    if qualifiers:
        text += " | Einschränkungen: " + ", ".join(qualifiers)
    return text


def normalize_claim(
    raw_statement: str, claim_id: Optional[str] = None, evidence_refs: Optional[List[str]] = None
) -> CustomerImplementationClaim:
    capabilities = match_capabilities(raw_statement)
    qualifiers = match_qualifiers(raw_statement)
    return CustomerImplementationClaim(
        claim_id=claim_id or _claim_id(raw_statement),
        raw_statement=raw_statement,
        normalized_claim=_normalized(capabilities, qualifiers),
        claimed_capability=capabilities,
        related_topics=topics_for(capabilities),
        qualifiers=qualifiers,
        evidence_refs=evidence_refs or [],
    )