# -----------------------------------------------------------------------------
# Lou-Ka — Location court terme
# connectors/boraboreal.py : Bora Boréal (boraboreal.com) — chalets FLOTTANTS
#   (minibora, boravilla) à Bury (Cantons-de-l'Est) et à Québec, plus un
#   chalet en bois rond ; ~13 unités réservables sur Lodgify
#   (reserver-boraboreal.lodgify.com, protégé Cloudflare → Scrapfly).
#
# Méthode :
#   1. SLUGS : le sitemap Lodgify est vide → la page « louer-maison-flottante »
#      (rendue via Scrapfly, la grille est en JS) liste les 12 chalets
#      flottants ; les pages vitrines boraboreal.com (récupérées en direct)
#      ajoutent les unités hors grille (bora-bois-rond).
#   2. DÉTAIL : chaque page unité Lodgify (Scrapfly sans render_js, cache
#      détail « v1 ») embarque un JSON-LD VacationRental complet :
#      identifier (external_id), priceRange « from 229 CAD/night »,
#      occupancy/chambres, amenityFeature, adresse, geo lat/lng, ~19 photos
#      icdbcdn ; le CITQ est extrait de la description (au-delà de la fenêtre
#      de 2000 caractères de finalize()). Animaux : frais au séjour
#      → pets = « conditions » si l'amenité l'indique.
# -----------------------------------------------------------------------------
from __future__ import annotations

import html as _html
import json
import re
import sys

from ..schema import StListing
from .airbnb import _region_from_latlng
from .base import StConnector

BOOKING = "https://reserver-boraboreal.lodgify.com"
GRID = BOOKING + "/fr/louer-maison-flottante"

# pages vitrines boraboreal.com (accessibles en direct) qui pointent vers des
# unités Lodgify absentes de la grille
SHOWCASE = [
    "https://boraboreal.com/chalet-flottant-quebec",
    "https://boraboreal.com/mini-chalet-a-louer-estrie",
    "https://boraboreal.com/chalet-6-personnes-estrie",
]

_TAG_RE = re.compile(r"<[^>]+>")

_LD_RE = re.compile(r'(?s)<script[^>]*type="application/ld\+json"[^>]*>'
                    r"(.*?)</script>")


def _strip_html(txt: str) -> str:
    txt = re.sub(r"</p>|<br\s*/?>|</li>", "\n", txt or "")
    txt = _html.unescape(_TAG_RE.sub(" ", txt))
    txt = re.sub(r"[ \t]+", " ", txt)
    return re.sub(r"\n\s+", "\n", txt).strip()


class BoraBoreal(StConnector):
    source_id = "boraboreal"

    # -- découverte des slugs --------------------------------------------------
    def _slugs(self) -> list[str]:
        slugs: list[str] = []

        def add(s: str):
            s = s.strip("/")
            if s and s != "louer-maison-flottante" and s not in slugs:
                slugs.append(s)

        try:
            grid = self.get_scrapfly(GRID, render_js=True,
                                     rendering_wait=3000)
            for s in re.findall(r'href="(?:%s)?/fr/([\w~-]+)"'
                                % re.escape(BOOKING), grid):
                add(s)
        except Exception as exc:  # noqa: BLE001
            print(f"[boraboreal] grille : {exc}", file=sys.stderr)

        for page in SHOWCASE:
            try:
                h = self.get(page).text
            except Exception:  # noqa: BLE001
                continue
            for s in re.findall(
                    r"reserver-boraboreal\.lodgify\.com/(?:fr/)?([\w~-]+)", h):
                add(s)
        return slugs

    # -- page unité Lodgify ----------------------------------------------------
    def _detail(self, slug: str) -> dict:
        h = self.get_scrapfly(f"{BOOKING}/fr/{slug}", render_js=False)
        for block in _LD_RE.findall(h):
            try:
                ld = json.loads(block)
            except ValueError:
                continue
            if ld.get("@type") == "VacationRental":
                return {"ld": ld}
        return {}

    # -- contrat ---------------------------------------------------------------
    def fetch(self) -> list[StListing]:
        listings: list[StListing] = []
        vus: set[str] = set()
        for slug in self._slugs():
            det = self.detail(slug, "v1", lambda s=slug: self._detail(s))
            ld = det.get("ld") or {}
            if not ld:
                continue
            ext_id = str(ld.get("identifier") or slug)
            if ext_id in vus:
                continue
            vus.add(ext_id)

            place = ld.get("containsPlace") or {}
            occupancy = (place.get("occupancy") or {}).get("value")
            addr = ld.get("address") or {}
            geo = ld.get("geo") or {}

            # « from 229 CAD/night » → price_night
            price = None
            m = re.search(r"from\s+([\d.]+)\s*CAD",
                          str(ld.get("priceRange") or ""))
            if m:
                v = float(m.group(1))
                if 20 <= v <= 20000:
                    price = v

            description = _strip_html(str(ld.get("description") or ""))
            m = re.search(r"CITQ\D{0,25}(\d{6})", description)
            citq = m.group(1) if m else ""

            amen = [a.get("name") for a in ld.get("amenityFeature") or []
                    if isinstance(a, dict) and a.get("name")
                    and a.get("value") is not False]
            pets = "conditions" if any("animaux" in a.lower()
                                       or "pet" in a.lower()
                                       for a in amen) else None

            imgs = ld.get("image") or []
            if isinstance(imgs, str):
                imgs = [imgs]

            city = (addr.get("addressLocality") or "").strip()
            region = (_region_from_latlng(geo.get("latitude"),
                                          geo.get("longitude"))
                      or ("Québec" if slug.endswith("---quebec")
                          else "Cantons-de-l'Est"))

            listings.append(StListing(
                source=self.source_id,
                external_id=ext_id,
                url=f"{BOOKING}/fr/{slug}",
                title=_html.unescape(str(ld.get("name") or slug)).strip(),
                property_type="Chalet",
                address=(addr.get("streetAddress") or "").strip(),
                city=city,
                region=region,
                price_night=price,
                price_label=(f"à partir de {price:g} $ / nuit"
                             if price else ""),
                capacity=float(occupancy) if occupancy else None,
                bedrooms=(float(place["numberOfBedrooms"])
                          if place.get("numberOfBedrooms") else None),
                pets=pets,
                citq=citq,
                description=description[:5000],
                amenities=amen,
                details={"floating": "bois-rond" not in slug,
                         "postal_code": addr.get("postalCode") or ""},
                images=[u for u in imgs if isinstance(u, str)
                        and u.startswith("https://")][:20],
                lat=geo.get("latitude"),
                lng=geo.get("longitude"),
            ))
        return listings