# Trouve-KA — construction du document d'index
# Author: Simon-Pierre Boucher
# Contact: contact@spboucher.ai

"""Transforme une page parsée + signaux Québec en document OpenSearch (étape 1)."""

from datetime import UTC, datetime
from typing import Any

from trouveka.types import ParsedPage, QuebecSignals

# Catégorisation grossière par domaine — raffinée en asynchrone (étape 2)
_GOV_SUFFIXES = (".gouv.qc.ca", ".gc.ca", ".quebec.ca")
_GOV_DOMAINS = {"quebec.ca", "canada.ca", "montreal.ca", "laval.ca", "gatineau.ca", "sherbrooke.ca"}
_NEWS_DOMAINS = {
    "lapresse.ca", "ledevoir.com", "journaldemontreal.com", "journaldequebec.com",
    "radio-canada.ca", "tvanouvelles.ca", "lesoleil.com", "ledroit.com",
    "latribune.ca", "lenouvelliste.ca", "lequotidien.com", "lavoixdelest.ca",
    "noovo.info", "24heures.ca", "montrealgazette.com",
}
_EDU_SUFFIXES = (".ulaval.ca", ".umontreal.ca", ".mcgill.ca", ".uqam.ca", ".usherbrooke.ca",
                 ".concordia.ca", ".polymtl.ca", ".etsmtl.ca", ".hec.ca")


def categorize_domain(domain: str) -> list[str]:
    d = domain.lower()
    cats: list[str] = []
    if d in _GOV_DOMAINS or any(d.endswith(s) for s in _GOV_SUFFIXES) or ".gouv." in d:
        cats.append("government")
    if d in _NEWS_DOMAINS:
        cats.append("news")
    if any(d.endswith(s) or d == s.lstrip(".") for s in _EDU_SUFFIXES) or d.endswith(".edu"):
        cats.append("education")
    return cats


def build_search_document(
    page: ParsedPage,
    signals: QuebecSignals,
    *,
    domain: str,
    domain_quebec_score: float,
    authority_score: float = 0.0,
) -> dict[str, Any]:
    return {
        "url": page.url,
        "canonical_url": page.canonical_url or page.url,
        "domain": domain,
        "title": page.title,
        "description": page.description,
        "body": page.body[:100_000],
        "headings": page.headings,
        "language": page.language,
        "page_quebec_score": signals.score,
        "domain_quebec_score": round(domain_quebec_score, 4),
        "locations": signals.locations,
        "organizations": [],   # enrichissement étape 2
        "people": [],          # enrichissement étape 2
        "categories": categorize_domain(domain),
        "image_url": page.image_url,
        "published_at": page.published_at.isoformat() if page.published_at else None,
        "crawled_at": datetime.now(UTC).isoformat(),
        "authority_score": round(authority_score, 4),
        "freshness_score": 0.0,  # enrichissement étape 3
        "quality_score": 0.0,    # enrichissement étape 3
        "spam_score": 0.0,       # enrichissement étape 3
    }