from __future__ import annotations

import re

from .normalization import normalize_token

SUPPORT_TERMS = [
    "first line",
    "1st line",
    "helpdesk",
    "service desk",
    "telefonische support",
    "support utilisateurs",
]
CONSULTANCY_TERMS = [
    "consultancy",
    "consultant",
    "bij klanten",
    "chez nos clients",
    "customer sites",
]
TRAVEL_TERMS = ["verplaatsingen", "travel required", "déplacements", "rijbewijs b"]
PUBLIC_SECTOR_TERMS = ["overheid", "gemeente", "provincie", "publieke sector", "service public"]


def term_ratio(text: str, terms: list[str]) -> float:
    normalized = normalize_token(text)
    hits = sum(1 for term in terms if normalize_token(term) in normalized)
    return min(1.0, hits / max(1, len(terms) / 2))


def extract_deterministic_features(title: str, description: str) -> dict[str, object]:
    text = f"{title}\n{description}"
    normalized = normalize_token(text)
    experience_years = [
        int(v) for v in re.findall(r"\b(\d{1,2})\s*(?:jaar|years?|ans)\b", normalized)
    ]
    return {
        "support_ratio": term_ratio(text, SUPPORT_TERMS),
        "consultancy_ratio": term_ratio(text, CONSULTANCY_TERMS),
        "travel_ratio": term_ratio(text, TRAVEL_TERMS),
        "public_sector_signal": term_ratio(text, PUBLIC_SECTOR_TERMS),
        "experience_years_max": max(experience_years) if experience_years else None,
        "word_count": len(normalized.split()),
    }