# -----------------------------------------------------------------------------
# Lou-Ka — Agrégateur de logements à louer (province de Québec)
# Auteur : Simon-Pierre Boucher — contact@spboucher.ai
# connectors/tour_fides.py : connecteur La Tour Fides (latourfides.com)
#   290 unités locatives au 235, boulevard René-Lévesque Est (Ville-Marie,
#   Montréal). Le site WordPress ne publie aucun prix (le « budget » de la
#   page d'accueil est un menu de formulaire) : les vraies données vivent sur
#   le portail Yardi RentCafe/SecureCafe
#   latourfides.securecafe.com/onlineleasing/la-tour-fides/floorplans.aspx,
#   protégé par Cloudflare (403 direct) -> Scrapfly (ASP + rendu JS).
#   La page floorplans expose un accordéon par plan (data-selenium-id
#   FloorPlanName_n / Bed_Bath_n / Rent_n / Availability_n / floorplandesc_n)
#   avec fourchette de loyer réelle, superficie, salles de bain et carrousel
#   de plans par unité (id fp-myCarousel<FloorPlanID>, identifiant RentCafe
#   STABLE -> external_id). Une annonce par plan (granularité la plus fine
#   publiée, comme elite.py). Les photos du projet viennent du site vitrine
#   latourfides.com (accès direct, wp-content).
# -----------------------------------------------------------------------------
from __future__ import annotations

import re

from bs4 import BeautifulSoup

from ..schema import Listing, normalize_unit_type, strip_accents
from .base import BaseConnector

SITE = "https://latourfides.com"
FLOORPLANS_URL = ("https://latourfides.securecafe.com/onlineleasing/"
                  "la-tour-fides/floorplans.aspx?nocache=-1")

ADDRESS = "235, boulevard René-Lévesque Est, Montréal, QC H2X 1N8"
SECTOR = "Ville-Marie"
PHONE = "514-700-2589"

# « $1,920 - à $2,920 » / « $8,600 » (format RentCafe, virgule = milliers)
_RENT_RE = re.compile(r"\$\s*([\d,]+)(?:\s*-\s*(?:à\s*)?\$\s*([\d,]+))?")
# « 424 - à 766 Square Foot » (texte de la rangée d'accordéon)
_SQFT_RE = re.compile(r"([\d,]+)(?:\s*-\s*(?:à\s*)?([\d,]+))?\s*Square\s*F",
                      re.I)
# « Bed/Bath Studio / 1 » / « Bed/Bath 2 / 1.5 »
_BEDBATH_RE = re.compile(r"(Studio|\d+)\s*/\s*([\d.]+)", re.I)
# identifiant RentCafe stable du plan (carrousel de l'accordéon)
_CAROUSEL_RE = re.compile(r"fp-myCarousel(\d+)")
# photos du site vitrine (originaux seulement, pas les variantes -WxH)
_WP_IMG_RE = re.compile(
    r"https://latourfides\.com/wp-content/uploads/[^\"'\s\)]+\.(?:jpg|jpeg|webp)",
    re.I)
_SKIP_IMG = re.compile(r"logo|icon|favicon|-\d{2,4}x\d{2,4}\.", re.I)


def _slugify(s: str) -> str:
    return re.sub(r"[^a-z0-9]+", "-", strip_accents(s.lower())).strip("-")


def _fmt(n: float) -> str:
    return f"{int(n):,}".replace(",", " ") + " $"


class TourFidesConnector(BaseConnector):
    source_id = "tour_fides"
    request_delay = 1.0
    max_plans = 25
    max_images = 12

    # -- page floorplans SecureCafe (Cloudflare -> Scrapfly) --------------------
    def _floorplans_html(self) -> str:
        for wait in (4000, 9000):        # seconde chance si rendu incomplet
            html = self.get_scrapfly(FLOORPLANS_URL, rendering_wait=wait)
            if html and "FloorPlanName_1" in html:
                return html
        raise RuntimeError(
            "floorplans.aspx SecureCafe inaccessible via Scrapfly")

    # -- photos + description du site vitrine (accès direct) --------------------
    def _site_extras(self) -> tuple[list[str], str]:
        images: list[str] = []
        blurb = ""
        try:
            html = self.get(SITE).text
            for u in dict.fromkeys(_WP_IMG_RE.findall(html)):
                if not _SKIP_IMG.search(u) and u not in images:
                    images.append(u)
            og = BeautifulSoup(html, "html.parser").find(
                "meta", attrs={"property": "og:description"})
            if og:
                blurb = (og.get("content") or "").strip()[:400]
        except Exception:
            pass
        return images[: self.max_images], blurb

    def fetch(self) -> list[Listing]:
        # un échec transitoire doit être journalisé comme ÉCHEC par ingest
        # (ok=0), pas comme un passage « ok » à 0 annonce
        soup = BeautifulSoup(self._floorplans_html(), "html.parser")
        site_images, blurb = self._site_extras()

        listings: list[Listing] = []
        seen: set[str] = set()
        for n in range(1, self.max_plans + 1):
            name_el = soup.select_one(f"[data-selenium-id='FloorPlanName_{n}']")
            if not name_el:
                break
            try:
                lst = self._plan_listing(soup, n, name_el, site_images, blurb)
            except Exception:
                continue
            if lst and lst.external_id not in seen:
                seen.add(lst.external_id)
                listings.append(lst)
        return listings

    # -- une annonce par plan d'étage --------------------------------------------
    def _plan_listing(self, soup, n: int, name_el, site_images: list[str],
                      blurb: str) -> Listing | None:
        name = re.sub(r"(?i)^plan d.[ée]tage\s*", "",
                      name_el.get_text(" ", strip=True)).strip()
        if not name:
            return None

        def _txt(prefix: str) -> str:
            el = soup.select_one(f"[data-selenium-id='{prefix}_{n}']")
            return el.get_text(" ", strip=True) if el else ""

        # chambres / salles de bain (« Bed/Bath 2 / 1.5 »)
        unit_type = ""
        bathrooms = None
        m = _BEDBATH_RE.search(_txt("Bed_Bath"))
        if m:
            if m.group(1).lower() == "studio":
                unit_type = "Studio"
            else:
                unit_type = normalize_unit_type(f"{m.group(1)} chambres")
            try:
                bathrooms = float(m.group(2))
            except ValueError:
                pass

        # loyer : fourchette réelle RentCafe (borne basse retenue)
        rent_txt = _txt("Rent")
        price = None
        price_label = ""
        rm = _RENT_RE.search(rent_txt)
        if rm:
            lo = float(rm.group(1).replace(",", ""))
            hi = float(rm.group(2).replace(",", "")) if rm.group(2) else None
            if 300 <= lo <= 30000:
                price = lo
                price_label = (f"À partir de {_fmt(lo)} (max {_fmt(hi)})/mois"
                               if hi and hi > lo else f"{_fmt(lo)}/mois")
        elif re.search(r"(?i)appelez|call", rent_txt):
            price_label = "Appelez-nous pour plus de détails"

        availability = _txt("Availability")

        # superficie « 424 - à 766 Square Foot » de la rangée d'accordéon
        area = None
        row = soup.select_one(f"[data-selenium-id='tRow{n}_1']")
        row_txt = row.get_text(" ", strip=True) if row else ""
        sm = _SQFT_RE.search(row_txt)
        sqft_bits = ""
        if sm:
            lo_s = float(sm.group(1).replace(",", ""))
            hi_s = float(sm.group(2).replace(",", "")) if sm.group(2) else None
            if 80 <= lo_s <= 20000:
                area = lo_s
                sqft_bits = (f"Superficie : {int(lo_s)} à {int(hi_s)} pi²"
                             if hi_s and hi_s > lo_s
                             else f"Superficie : {int(lo_s)} pi²")

        # identifiant RentCafe stable du plan (id du carrousel de l'accordéon)
        ext_id = ""
        img_cell = soup.select_one(f"[data-selenium-id='Image_{n}']")
        if img_cell:
            cm = _CAROUSEL_RE.search(str(img_cell))
            if cm:
                ext_id = f"fp-{cm.group(1)}"
        ext_id = ext_id or _slugify(name)

        # plans des unités du carrousel (cdngeneralcf.rentcafe.com) + photos site
        plan_imgs: list[str] = []
        if img_cell:
            for im in img_cell.select("img"):
                src = (im.get("src") or im.get("data-src") or "").strip()
                if src.startswith("http") and src not in plan_imgs:
                    plan_imgs.append(src)
        images = (site_images + plan_imgs)[: self.max_images + 6]

        # description : blurb du site + texte du plan (« lire plus » exclu)
        desc = _txt("floorplandesc")
        desc = re.sub(r"(?i)\s*lire (plus|moins)\s*", " ", desc).strip()
        desc_parts = [b for b in (blurb, desc) if b]

        amenities = []
        if sqft_bits:
            amenities.append(sqft_bits)
        specials = _txt("Specials")
        if 3 < len(specials) < 120:
            amenities.append(f"Promotion : {specials}")

        return Listing(
            source=self.source_id,
            external_id=ext_id,
            url=FLOORPLANS_URL,
            title=f"La Tour Fides — {name}",
            address=ADDRESS,
            sector=SECTOR,
            city="Montréal",
            unit_type=unit_type,
            bathrooms=bathrooms,
            price=price,
            price_label=price_label,
            availability=availability,
            area_sqft=area,
            description=" — ".join(desc_parts)[:700],
            amenities=amenities,
            details={"contact": {"phone": PHONE}},
            images=images,
        )