# -----------------------------------------------------------------------------
# Lou-Ka — Agrégateur de logements à louer (province de Québec)
# Auteur : Simon-Pierre Boucher — contact@spboucher.ai
# connectors/obohec.py : connecteur O'bohec (obohec.ca) — Voyer & Tremblay
#   Trois immeubles locatifs sur le boulevard Michèle-Bohec, Blainville
#   (27 unités affichées : studio, 1 et 2 chambres). WordPress (thème maison
#   « trema ») rendu serveur : /projets-locatifs/obohec/ liste des cartes
#   .item-listing-unite groupées par en-tête .item-unites-filter
#   (data-immeuble-id → Immeuble 1/2/3, classe filter-<id> sur la carte).
#   Carte : .title-unite (Studio/1 chambre/2 chambres), .number-unite
#   (« #105 Tout meublé Disponible septembre 2026 »), .disponiblite-unite
#   (Loué/Disponible), caractéristiques .car-* (chambres, bain, balcon,
#   stationnement, orientation, superficie) et plan image + PDF. AUCUN prix
#   publié (« Prix sur demande »). On exclut « BUREAU DE LOCATION » et les
#   unités louées. Granularité : unité. ⚠️ Le volet VENTE du site est ignoré.
# -----------------------------------------------------------------------------
from __future__ import annotations

import re

from bs4 import BeautifulSoup

from ..schema import Listing, normalize_unit_type

from .base import BaseConnector

BASE = "https://obohec.ca"
LIST_URL = f"{BASE}/projets-locatifs/obohec/"

ADDRESS = "Boulevard Michèle-Bohec, Blainville"
CITY = "Blainville"

FILTER_RE = re.compile(r"^filter-(\d+)$")
AREA_RE = re.compile(r"(\d[\d\s  ]*)\s*pi", re.I)
AVAIL_RE = re.compile(r"Disponible[^#<]*", re.I)


class ObohecConnector(BaseConnector):
    source_id = "obohec"
    request_delay = 0.6
    max_images = 10

    @staticmethod
    def _txt(node, sel: str) -> str:
        el = node.select_one(sel)
        return re.sub(r"\s+", " ", el.get_text(" ", strip=True)) if el else ""

    def fetch(self) -> list[Listing]:
        html = self.get(LIST_URL).text
        soup = BeautifulSoup(html, "html.parser")

        # data-immeuble-id -> « Immeuble N »
        buildings: dict[str, str] = {}
        for h in soup.select(".item-unites-filter[data-immeuble-id]"):
            buildings[str(h["data-immeuble-id"])] = re.sub(
                r"\s+", " ", h.get_text(" ", strip=True))

        listings: list[Listing] = []
        seen: set[str] = set()
        for it in soup.select(".item-listing-unite"):
            title_u = self._txt(it, ".title-unite")
            if re.search(r"bureau\s+de\s+location", title_u, re.I):
                continue                              # local commercial
            dispo = self._txt(it, ".disponiblite-unite")
            if re.search(r"lou[ée]", dispo, re.I):
                continue

            num_raw = self._txt(it, ".number-unite")
            m = re.search(r"#\s*([\w-]+)", num_raw)
            if m is None:
                continue
            num = m.group(1)

            bid = ""
            for cls in it.get("class", []):
                fm = FILTER_RE.match(cls)
                if fm:
                    bid = fm.group(1)
                    break
            ext_id = f"imm{bid}-{num}" if bid else num
            if ext_id in seen:
                continue
            seen.add(ext_id)
            building = buildings.get(bid, "")

            m = AVAIL_RE.search(num_raw)
            availability = m.group(0).strip() if m else (dispo or "Disponible")
            furnished = bool(re.search(r"meubl[ée]", num_raw, re.I))

            beds = None
            v = self._txt(it, ".car-chambres span")
            if v.isdigit():
                beds = float(v)
            baths = None
            v = self._txt(it, ".car-bain span")
            if v.replace(".", "", 1).isdigit():
                baths = float(v)

            area = None
            m = AREA_RE.search(self._txt(it, ".car-superficie span"))
            if m:
                try:
                    v2 = float(re.sub(r"[\s  ]", "", m.group(1)))
                    if 100 <= v2 <= 10000:
                        area = v2
                except ValueError:
                    pass

            amenities = [a for a in (
                "Tout meublé" if furnished else "",
                f"Balcon/terrasse : {self._txt(it, '.car-balcon-terrasse span')}"
                if self._txt(it, ".car-balcon-terrasse span") else "",
                f"Stationnement {self._txt(it, '.car-stationnement span')}"
                if self._txt(it, ".car-stationnement span") else "",
                f"Orientation {self._txt(it, '.car-orientation span')}"
                if self._txt(it, ".car-orientation span") else "",
            ) if a]

            images: list[str] = []
            img = it.select_one(".image-plan-unite img[src]")
            if img is not None and str(img["src"]).startswith("http"):
                images.append(str(img["src"]))
            details: dict = {}
            pdf = it.select_one('a.btn-imprimable[href], a[href$=".pdf"]')
            if pdf is not None:
                details["floor_plan"] = pdf["href"]

            desc_bits = [f"Unité #{num}", building, title_u,
                         "Tout meublé" if furnished else ""]
            listings.append(Listing(
                source=self.source_id,
                external_id=ext_id,
                url=LIST_URL,
                title=f"O'bohec — {building or 'Blainville'} — Unité #{num}"
                      f" ({title_u})",
                address=ADDRESS,
                city=CITY,
                unit_type=normalize_unit_type(title_u),
                bedrooms=beds,
                bathrooms=baths,
                price=None,
                price_label="Prix sur demande",
                availability=availability,
                area_sqft=area,
                description=" — ".join(b for b in desc_bits if b),
                amenities=amenities,
                details=details,
                images=images,
            ))
        return listings