feat(ontario v1): connecteur générique RealtyPress — 13 sites d'agences ON (CREA DDF) · 39ac48a · immo-ka

feat(ontario v1): connecteur générique RealtyPress — 13 sites d'agences ON (CREA DDF)

- connectors/realtypress.py : archive /listing/page/N/?posts_per_page=100,
  cookie disclaimer=accepted (mur CREA), 2 variantes de thème (ville avec/sans
  <strong>, ft²), fiches DDF (MLS®, tableaux, GPS JSON-LD, photos ddfcdn),
  external_id ddf<id> + infixe _ag_ → dédup inter-sites par external_id
- data/ontario_agencies.json : registre 13 sites (Ottawa, KW, Kingston,
  Grey-Bruce, Collingwood, PEC, Pembroke, Huron-Perth, Thunder Bay, Sault,
  North Bay) ; sources.json : 13 entrées rp_ag_*
- pipeline Ontario-aware : schema.finalize bbox 41.6–63/-95.5–-56,
  normalize._TYPE_MAP mots-clés anglais (detached, semi-detached, row house,
  multi-family, mobile home, farm, retail…), geocode _BBOX_KA + fallback
  Nominatim « city, Ontario, Canada » + run_batch exclut region='Ontario'
  (Adresses Québec ne couvre pas l'ON ; le DDF fournit lat/lng)

Testé : cbnorthbay ingéré (503 fiches, 150 enrichies publiées score 94,
stubs en quarantaine jusqu'à enrichissement), labrosse/sutton spot-checkés,
API /api/listings?region=Ontario OK, parc QC intact (94 528 publiées).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Simon-Pierre Boucher committed 1 mo ago (Aug 27, 2026) parent b0cfba6

6 changed files +569 −40

added data/ontario_agencies.json +80 −0

		`@@ -0,0 +1,80 @@`
	1	+[
	2	+ {
	3	+ "id": "rp_ag_suttonottawa",
	4	+ "name": "Sutton Group — Ottawa Realty",
	5	+ "site": "https://suttonottawa.ca",
	6	+ "note": "Recensement ON 2026-08-27 : ~10 080 fiches (OREB+). RealtyPress/DDF."
	7	+ },
	8	+ {
	9	+ "id": "rp_ag_helensteam",
	10	+ "name": "Helen's Team (Kitchener-Waterloo)",
	11	+ "site": "https://helensteam.ca",
	12	+ "note": "Recensement ON 2026-08-27 : ~9 910 fiches (Kitchener-Waterloo). RealtyPress/DDF."
	13	+ },
	14	+ {
	15	+ "id": "rp_ag_greybruce",
	16	+ "name": "Grey Bruce Real Estate",
	17	+ "site": "https://greybrucerealestate.ca",
	18	+ "note": "Recensement ON 2026-08-27 : ~8 268 fiches (Grey-Bruce/Georgian Bay). RealtyPress/DDF. Même feed que collaborativerealestate.ca (fallback)."
	19	+ },
	20	+ {
	21	+ "id": "rp_ag_remaxfinest",
	22	+ "name": "RE/MAX Finest Realty (Kingston)",
	23	+ "site": "https://remaxfinestrealty.com",
	24	+ "note": "Recensement ON 2026-08-27 : ~8 217 fiches (Kingston). RealtyPress/DDF."
	25	+ },
	26	+ {
	27	+ "id": "rp_ag_riouxbaker",
	28	+ "name": "Rioux Baker Real Estate Team (Collingwood)",
	29	+ "site": "https://riouxbakerteam.com",
	30	+ "note": "Recensement ON 2026-08-27 : ~7 716 fiches (Collingwood/South Georgian Bay). RealtyPress/DDF."
	31	+ },
	32	+ {
	33	+ "id": "rp_ag_countyguys",
	34	+ "name": "The County Guys (Prince Edward County)",
	35	+ "site": "https://thecountyguys.com",
	36	+ "note": "Recensement ON 2026-08-27 : ~6 878 fiches (Prince Edward County/Quinte). RealtyPress/DDF."
	37	+ },
	38	+ {
	39	+ "id": "rp_ag_labrosse",
	40	+ "name": "Labrosse Real Estate (Ottawa/Orléans)",
	41	+ "site": "https://labrosserealestate.com",
	42	+ "note": "Recensement ON 2026-08-27 : ~6 873 fiches (Ottawa/Orléans, équipe FRANCOPHONE). RealtyPress/DDF."
	43	+ },
	44	+ {
	45	+ "id": "rp_ag_ryanpattinson",
	46	+ "name": "Ryan Pattinson (Pembroke/Renfrew)",
	47	+ "site": "https://ryanpattinson.com",
	48	+ "note": "Recensement ON 2026-08-27 : ~6 601 fiches (Pembroke/vallée de l'Outaouais ON). RealtyPress/DDF."
	49	+ },
	50	+ {
	51	+ "id": "rp_ag_grapevine",
	52	+ "name": "Grapevine (Ottawa)",
	53	+ "site": "https://grapevine.ca",
	54	+ "note": "Recensement ON 2026-08-27 : ~6 550 fiches (Ottawa). RealtyPress/DDF. Suivre les redirections www/apex."
	55	+ },
	56	+ {
	57	+ "id": "rp_ag_rlpheartland",
	58	+ "name": "Royal LePage Heartland Realty",
	59	+ "site": "https://rlpheartland.ca",
	60	+ "note": "Recensement ON 2026-08-27 : ~5 577 fiches (Midwestern Ontario — Huron/Perth). RealtyPress/DDF."
	61	+ },
	62	+ {
	63	+ "id": "rp_ag_signaturenorth",
	64	+ "name": "Signature North Realty (Thunder Bay)",
	65	+ "site": "https://signaturenorthrealty.ca",
	66	+ "note": "Recensement ON 2026-08-27 : ~1 023 fiches (Thunder Bay). RealtyPress/DDF."
	67	+ },
	68	+ {
	69	+ "id": "rp_ag_saultstemarie",
	70	+ "name": "Sault Ste. Marie Real Estate (Century 21 Choice)",
	71	+ "site": "https://saultstemarierealestate.com",
	72	+ "note": "Recensement ON 2026-08-27 : ~864 fiches (Sault Ste. Marie). RealtyPress/DDF."
	73	+ },
	74	+ {
	75	+ "id": "rp_ag_cbnorthbay",
	76	+ "name": "Coldwell Banker Peter Minogue (North Bay)",
	77	+ "site": "https://cbnorthbay.com",
	78	+ "note": "Recensement ON 2026-08-27 : ~307 fiches (North Bay). RealtyPress/DDF."
	79	+ }
	80	+]

modified data/sources.json +144 −1

		`@@ -1,5 +1,5 @@`
1	1	{
2		− "_comment": "Immo-Ka — Registre des sources (agences de courtage immobilier, province de Québec). Un connecteur par agence. Auteur : Simon-Pierre Boucher — contact@spboucher.ai",
	2	+ "_comment": "Immo-Ka — Registre des sources (agences de courtage immobilier, Québec + Ontario). Un connecteur par agence. Auteur : Simon-Pierre Boucher — contact@spboucher.ai",
3	3	"sources": [
4	4	{
5	5	"id": "remax_quebec",
		`@@ -867,6 +867,149 @@`
867	867	"status": "actif",
868	868	"type": "agence",
869	869	"note": "Site web « classique » server-rendered, connecteur générique agences_web.py (registre data/web_agencies.json). Découverte via recensement OACIQ 2026-08-24. Fiches Next.js rendues via Scrapfly. OACIQ 2026-08 : 9 courtiers (Montréal). Next.js : fiches rendues via Scrapfly."
	870	+ },
	871	+ {
	872	+ "id": "rp_ag_suttonottawa",
	873	+ "name": "Sutton Group — Ottawa Realty",
	874	+ "url": "https://suttonottawa.ca",
	875	+ "listing_url": "https://suttonottawa.ca/listing/",
	876	+ "coverage": "Ottawa et l'Est ontarien (board OREB) — ~10 000 fiches DDF",
	877	+ "connector": "realtypress",
	878	+ "status": "actif",
	879	+ "type": "agence",
	880	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~10 080 fiches (OREB+). RealtyPress/DDF."
	881	+ },
	882	+ {
	883	+ "id": "rp_ag_helensteam",
	884	+ "name": "Helen's Team (Kitchener-Waterloo)",
	885	+ "url": "https://helensteam.ca",
	886	+ "listing_url": "https://helensteam.ca/listing/",
	887	+ "coverage": "Kitchener-Waterloo et région (ITSO) — ~9 900 fiches DDF",
	888	+ "connector": "realtypress",
	889	+ "status": "actif",
	890	+ "type": "agence",
	891	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~9 910 fiches (Kitchener-Waterloo). RealtyPress/DDF."
	892	+ },
	893	+ {
	894	+ "id": "rp_ag_greybruce",
	895	+ "name": "Grey Bruce Real Estate",
	896	+ "url": "https://greybrucerealestate.ca",
	897	+ "listing_url": "https://greybrucerealestate.ca/listing/",
	898	+ "coverage": "Grey-Bruce / Georgian Bay (ITSO) — ~8 300 fiches DDF",
	899	+ "connector": "realtypress",
	900	+ "status": "actif",
	901	+ "type": "agence",
	902	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~8 268 fiches (Grey-Bruce/Georgian Bay). RealtyPress/DDF. Même feed que collaborativerealestate.ca (fallback)."
	903	+ },
	904	+ {
	905	+ "id": "rp_ag_remaxfinest",
	906	+ "name": "RE/MAX Finest Realty (Kingston)",
	907	+ "url": "https://remaxfinestrealty.com",
	908	+ "listing_url": "https://remaxfinestrealty.com/listing/",
	909	+ "coverage": "Kingston et région (KAREA) — ~8 200 fiches DDF",
	910	+ "connector": "realtypress",
	911	+ "status": "actif",
	912	+ "type": "agence",
	913	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~8 217 fiches (Kingston). RealtyPress/DDF."
	914	+ },
	915	+ {
	916	+ "id": "rp_ag_riouxbaker",
	917	+ "name": "Rioux Baker Real Estate Team (Collingwood)",
	918	+ "url": "https://riouxbakerteam.com",
	919	+ "listing_url": "https://riouxbakerteam.com/listing/",
	920	+ "coverage": "Collingwood / South Georgian Bay — ~7 700 fiches DDF",
	921	+ "connector": "realtypress",
	922	+ "status": "actif",
	923	+ "type": "agence",
	924	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~7 716 fiches (Collingwood/South Georgian Bay). RealtyPress/DDF."
	925	+ },
	926	+ {
	927	+ "id": "rp_ag_countyguys",
	928	+ "name": "The County Guys (Prince Edward County)",
	929	+ "url": "https://thecountyguys.com",
	930	+ "listing_url": "https://thecountyguys.com/listing/",
	931	+ "coverage": "Prince Edward County / Quinte — ~6 900 fiches DDF",
	932	+ "connector": "realtypress",
	933	+ "status": "actif",
	934	+ "type": "agence",
	935	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~6 878 fiches (Prince Edward County/Quinte). RealtyPress/DDF."
	936	+ },
	937	+ {
	938	+ "id": "rp_ag_labrosse",
	939	+ "name": "Labrosse Real Estate (Ottawa/Orléans)",
	940	+ "url": "https://labrosserealestate.com",
	941	+ "listing_url": "https://labrosserealestate.com/listing/",
	942	+ "coverage": "Ottawa / Orléans (équipe francophone, OREB) — ~6 900 fiches DDF",
	943	+ "connector": "realtypress",
	944	+ "status": "actif",
	945	+ "type": "agence",
	946	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~6 873 fiches (Ottawa/Orléans, équipe FRANCOPHONE). RealtyPress/DDF."
	947	+ },
	948	+ {
	949	+ "id": "rp_ag_ryanpattinson",
	950	+ "name": "Ryan Pattinson (Pembroke/Renfrew)",
	951	+ "url": "https://ryanpattinson.com",
	952	+ "listing_url": "https://ryanpattinson.com/listing/",
	953	+ "coverage": "Pembroke / vallée de l'Outaouais ontarienne — ~6 600 fiches DDF",
	954	+ "connector": "realtypress",
	955	+ "status": "actif",
	956	+ "type": "agence",
	957	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~6 601 fiches (Pembroke/vallée de l'Outaouais ON). RealtyPress/DDF."
	958	+ },
	959	+ {
	960	+ "id": "rp_ag_grapevine",
	961	+ "name": "Grapevine (Ottawa)",
	962	+ "url": "https://grapevine.ca",
	963	+ "listing_url": "https://grapevine.ca/listing/",
	964	+ "coverage": "Ottawa (OREB) — ~6 600 fiches DDF",
	965	+ "connector": "realtypress",
	966	+ "status": "actif",
	967	+ "type": "agence",
	968	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~6 550 fiches (Ottawa). RealtyPress/DDF. Suivre les redirections www/apex."
	969	+ },
	970	+ {
	971	+ "id": "rp_ag_rlpheartland",
	972	+ "name": "Royal LePage Heartland Realty",
	973	+ "url": "https://rlpheartland.ca",
	974	+ "listing_url": "https://rlpheartland.ca/listing/",
	975	+ "coverage": "Midwestern Ontario (Huron-Perth) — ~5 600 fiches DDF",
	976	+ "connector": "realtypress",
	977	+ "status": "actif",
	978	+ "type": "agence",
	979	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~5 577 fiches (Midwestern Ontario — Huron/Perth). RealtyPress/DDF."
	980	+ },
	981	+ {
	982	+ "id": "rp_ag_signaturenorth",
	983	+ "name": "Signature North Realty (Thunder Bay)",
	984	+ "url": "https://signaturenorthrealty.ca",
	985	+ "listing_url": "https://signaturenorthrealty.ca/listing/",
	986	+ "coverage": "Thunder Bay — ~1 000 fiches DDF",
	987	+ "connector": "realtypress",
	988	+ "status": "actif",
	989	+ "type": "agence",
	990	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~1 023 fiches (Thunder Bay). RealtyPress/DDF."
	991	+ },
	992	+ {
	993	+ "id": "rp_ag_saultstemarie",
	994	+ "name": "Sault Ste. Marie Real Estate (Century 21 Choice)",
	995	+ "url": "https://saultstemarierealestate.com",
	996	+ "listing_url": "https://saultstemarierealestate.com/listing/",
	997	+ "coverage": "Sault Ste. Marie — ~860 fiches DDF",
	998	+ "connector": "realtypress",
	999	+ "status": "actif",
	1000	+ "type": "agence",
	1001	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~864 fiches (Sault Ste. Marie). RealtyPress/DDF."
	1002	+ },
	1003	+ {
	1004	+ "id": "rp_ag_cbnorthbay",
	1005	+ "name": "Coldwell Banker Peter Minogue (North Bay)",
	1006	+ "url": "https://cbnorthbay.com",
	1007	+ "listing_url": "https://cbnorthbay.com/listing/",
	1008	+ "coverage": "North Bay / Nipissing — ~300 fiches DDF",
	1009	+ "connector": "realtypress",
	1010	+ "status": "actif",
	1011	+ "type": "agence",
	1012	+ "note": "Extension ONTARIO 2026-08 — connecteur générique realtypress.py (registre data/ontario_agencies.json). Plugin WordPress RealtyPress branché sur le flux CREA DDF : chaque site expose l'inventaire complet de son board. Infixe _ag_ : external_id ddf<id> partagé entre sites → la dédup masque les doublons inter-sites. Recensement ON 2026-08-27 : ~307 fiches (North Bay). RealtyPress/DDF."
870	1013	}
871	1014	]
872	1015	}

added immoka/connectors/realtypress.py +292 −0

		`@@ -0,0 +1,292 @@`
	1	+# -----------------------------------------------------------------------------
	2	+# Immo-Ka — Agrégateur de maisons à vendre (Québec + Ontario)
	3	+# Auteur : Simon-Pierre Boucher — contact@spboucher.ai
	4	+# connectors/realtypress.py : connecteur GÉNÉRIQUE RealtyPress (Ontario)
	5	+# RealtyPress = plugin WordPress branché sur le flux CREA DDF ; ~35 sites
	6	+# d'agences/équipes ontariennes confirmés (recensement 2026-08-27, voir
	7	+# docs/ontario-agences-connecteurs.md). Chaque site expose l'IDX/DDF complet
	8	+# de son board (OREB, ITSO, KAREA…) en HTML server-rendered, sans anti-bot :
	9	+# un seul parseur couvre quasi toute la province.
	10	+#
	11	+# - liste : archive /listing/page/N/?posts_per_page=100 (100 cartes/page ;
	12	+# ⚠ ?posts_per_page directement sur /listing = 301/vide) ; cartes
	13	+# class="rps-property-result" (ruban For sale/For rent, prix, adresse,
	14	+# ville, caractéristiques) ;
	15	+# - fiche : mur CREA « I Accept The Terms » contourné par le cookie
	16	+# `disclaimer=accepted` ; tableaux <strong>Label</strong>/valeur (MLS®
	17	+# Number, Property Type, Bedrooms…), description « … (id:NNNNN) »,
	18	+# lat/lng JSON-LD, photos ddfcdn.realtor.ca ;
	19	+# - external_id = ddf<id> (préfixe : jamais de collision avec les n° Centris
	20	+# QC) ; sources avec infixe _ag_ → la dédup par external_id masque les
	21	+# doublons inter-sites (le même bien DDF publié sur plusieurs sites).
	22	+# Sites générés depuis data/ontario_agencies.json (un source_id par site).
	23	+# -----------------------------------------------------------------------------
	24	+from __future__ import annotations
	25	+
	26	+import html as _html
	27	+import json
	28	+import os
	29	+import re
	30	+import urllib.parse
	31	+from pathlib import Path
	32	+
	33	+from .base import BaseConnector
	34	+from . import _detailutil as du
	35	+from ..schema import PropertyListing
	36	+
	37	+REGISTRY = Path(__file__).resolve().parent.parent.parent / "data" / "ontario_agencies.json"
	38	+DETAIL_LIMIT = int(os.environ.get("IMMOKA_RP_DETAIL_LIMIT",
	39	+ os.environ.get("IMMOKA_DETAIL_LIMIT", "150")))
	40	+
	41	+_CARD_RE = re.compile(r'<div class="rps-property-result">')
	42	+_LINK_RE = re.compile(r'href="(https?://[^"]+/listings?/[^"]+?-(\d{6,10})/?)"')
	43	+_RIBBON_RE = re.compile(r'rps-ribbon[^>]>\s([^<]+?)\s*<')
	44	+_PRICE_RE = re.compile(r'rps-price[^>]>\s\$\s*([\d,]+)')
	45	+_H4_RE = re.compile(r"<h4>\s(.?)\s*</h4>", re.S)
	46	+# avec ou sans <strong> selon le thème du site
	47	+_CITY_RE = re.compile(r'city-province-postalcode[^>]>\s(?:<strong>\s)?([^<]+?)\s<', re.S)
	48	+_FEAT_RE = re.compile(r'rps-result-feature-label[^>]>\s([^<]+?)\s*<')
	49	+_CARD_BROKER_RE = re.compile(r'text-muted[^>]>\s<small>\s([^<]+?)\s(?:<br\|</small>)', re.S)
	50	+_DDFIMG_RE = re.compile(r'https://ddfcdn\.realtor\.ca/[^")\'\s\\]+')
	51	+_ROW_RE = re.compile(r"<td[^>]>\s<strong>([^<]{2,45})</strong>\s</td>\s"
	52	+ r"<td[^>]>(.?)</td>", re.S)
	53	+_DESC_RE = re.compile(r'<!--\sDescription\s-->\s<p[^>]>(.*?)</p>', re.S)
	54	+_DESC_RE2 = re.compile(r'<p itemprop="description"[^>]>(.?)</p>', re.S)
	55	+# ville depuis <title> « 3383 Romeo Street, Greater Sudbury (Valley East), Ontario … »
	56	+_TITLE_CITY_RE = re.compile(r",\s([^,<>\|]{2,60}),\sOntario\b")
	57	+_PRICING_RE = re.compile(r'rps-pricing[^>]>\s\$\s*([\d,]+)')
	58	+_ID_TAIL_RE = re.compile(r"\s$id:\d{4,9}$\s$")
	59	+_TAG_RE = re.compile(r"<[^>]+>")
	60	+_NUM_RE = re.compile(r"[\d,]+(?:\.\d+)?")
	61	+_YEAR_RE = re.compile(r"\b(1[6-9]\d{2}\|20\d{2})\b")
	62	+# territoire couvert (Québec + Ontario) — même boîte que schema.finalize()
	63	+_BBOX = (41.6, 63.0, -95.5, -56.0)
	64	+
	65	+
	66	+def _num(s: str) -> float \| None:
	67	+ m = _NUM_RE.search(s or "")
	68	+ if not m:
	69	+ return None
	70	+ try:
	71	+ return float(m.group(0).replace(",", ""))
	72	+ except ValueError:
	73	+ return None
	74	+
	75	+
	76	+class _RealtyPress(BaseConnector):
	77	+ """Connecteur générique de site RealtyPress (voir data/ontario_agencies.json)."""
	78	+
	79	+ agency_name = ""
	80	+ site_url = ""
	81	+ max_pages = 150 # 100 cartes/page → jusqu'à 15 000 fiches par site
	82	+ request_delay = 0.6
	83	+
	84	+ def fetch(self) -> list[PropertyListing]:
	85	+ # mur CREA des fiches détail : le cookie suffit (posé pour tout domaine,
	86	+ # les redirections www/apex restent couvertes)
	87	+ self.session.cookies.set("disclaimer", "accepted")
	88	+ by_id: dict[str, PropertyListing] = {}
	89	+ base = self.site_url.rstrip("/")
	90	+ dry = 0
	91	+ for page in range(1, self.max_pages + 1):
	92	+ url = f"{base}/listing/page/{page}/?posts_per_page=100"
	93	+ try:
	94	+ body = self.get(url).text
	95	+ except Exception:
	96	+ break
	97	+ cards = self._cards(body)
	98	+ if not cards:
	99	+ break
	100	+ before = len(by_id)
	101	+ for card in cards:
	102	+ self._parse_card(card, by_id)
	103	+ dry = dry + 1 if len(by_id) == before else 0
	104	+ if dry >= 2:
	105	+ break
	106	+ listings = list(by_id.values())
	107	+ du.enrich(self, listings, DETAIL_LIMIT, parse_rp_detail, key="v1")
	108	+ for lst in listings:
	109	+ # n° MLS du board (fiche détail) — utile à la dédup inter-plateformes
	110	+ if not lst.mls and lst.details.get("MLS® Number"):
	111	+ lst.mls = str(lst.details["MLS® Number"])
	112	+ if not lst.title:
	113	+ lst.title = ", ".join(filter(None, (lst.address, lst.city))) \
	114	+ or "Propriété à vendre"
	115	+ return listings
	116	+
	117	+ def _cards(self, body: str) -> list[str]:
	118	+ marks = list(_CARD_RE.finditer(body))
	119	+ return [body[m.start():(marks[i + 1].start() if i + 1 < len(marks)
	120	+ else m.start() + 6000)]
	121	+ for i, m in enumerate(marks)]
	122	+
	123	+ def _parse_card(self, card: str, by_id: dict) -> None:
	124	+ ml = _LINK_RE.search(card)
	125	+ if not ml:
	126	+ return
	127	+ url, ddf = ml.group(1), ml.group(2)
	128	+ eid = f"ddf{ddf}"
	129	+ if eid in by_id:
	130	+ return
	131	+ mr = _RIBBON_RE.search(card)
	132	+ ribbon = (mr.group(1) if mr else "").strip().lower()
	133	+ if "rent" in ribbon or "lease" in ribbon:
	134	+ return # locations : hors périmètre
	135	+ lst = PropertyListing(source=self.source_id, external_id=eid, url=url,
	136	+ region="Ontario", agency=self.agency_name,
	137	+ broker_name=self.agency_name)
	138	+ ma = _H4_RE.search(card)
	139	+ if ma:
	140	+ lst.address = _html.unescape(_TAG_RE.sub(" ", ma.group(1))).strip()
	141	+ mc = _CITY_RE.search(card)
	142	+ if mc:
	143	+ city = _html.unescape(mc.group(1)).strip().rstrip(",")
	144	+ city = re.sub(r",?\sOntario\b.$", "", city, flags=re.I)
	145	+ lst.city = city.split("(")[0].strip()
	146	+ mp = _PRICE_RE.search(card)
	147	+ if mp:
	148	+ lst.price = _num(mp.group(1))
	149	+ lst.price_label = f"{mp.group(1)} $"
	150	+ for feat in _FEAT_RE.findall(card):
	151	+ f = _html.unescape(feat).strip()
	152	+ low = f.lower()
	153	+ n = _num(f)
	154	+ if not n:
	155	+ continue
	156	+ if "bedroom" in low:
	157	+ lst.bedrooms = int(n)
	158	+ elif "bathroom" in low:
	159	+ lst.bathrooms = int(n)
	160	+ elif "sqft" in low or "sq ft" in low or "ft" in low:
	161	+ lst.area_sqft = n # plage « 1,100 - 1,500 ft² » : borne basse
	162	+ mbk = _CARD_BROKER_RE.search(card)
	163	+ if mbk:
	164	+ lst.broker_name = _html.unescape(mbk.group(1)).strip()[:120]
	165	+ mi = _DDFIMG_RE.search(card)
	166	+ if mi:
	167	+ lst.images = [mi.group(0)]
	168	+ by_id[lst.external_id] = lst
	169	+
	170	+
	171	+def parse_rp_detail(html: str) -> dict:
	172	+ """Fiche RealtyPress : tableaux DDF, description, GPS, galerie, courtier."""
	173	+ out: dict = {}
	174	+ details: dict = {}
	175	+
	176	+ for lab, val in _ROW_RE.findall(html):
	177	+ label = _html.unescape(lab).strip().rstrip(":")
	178	+ value = re.sub(r"\s+", " ", _html.unescape(_TAG_RE.sub(" ", val))).strip()
	179	+ if label and value and len(value) <= 300:
	180	+ details.setdefault(label, value)
	181	+
	182	+ def dv(*labels: str) -> str:
	183	+ for lb in labels:
	184	+ if details.get(lb):
	185	+ return details[lb]
	186	+ return ""
	187	+
	188	+ b = _num(dv("Bedrooms Total", "Bedrooms", "Bedrooms Above Ground"))
	189	+ if b is not None and 0 < b <= 30:
	190	+ out["bedrooms"] = int(b)
	191	+ b = _num(dv("Bathroom Total", "Bathrooms"))
	192	+ if b is not None and 0 < b <= 30:
	193	+ out["bathrooms"] = int(b)
	194	+ b = _num(dv("Half Bath Total"))
	195	+ if b is not None and 0 < b <= 10:
	196	+ out["powder_rooms"] = int(b)
	197	+ my = _YEAR_RE.search(dv("Constructed Date", "Construction Year", "Age"))
	198	+ if my:
	199	+ out["year_built"] = int(my.group(1))
	200	+ si = dv("Size Interior")
	201	+ if si and "sqft" in si.lower().replace(" ", ""):
	202	+ a = _num(si) # « 7,901 Sqft » / « 1200 - 1399 sqft »
	203	+ if a and a >= 100:
	204	+ out["area_sqft"] = a
	205	+ pt = dv("Property Type", "Building Type", "Type")
	206	+ if pt:
	207	+ out["property_type"] = pt # anglais DDF — normalisé par finalize()
	208	+ sec = dv("Neigbourhood", "Neighbourhood", "Community Name")
	209	+ if sec:
	210	+ out["sector"] = sec
	211	+
	212	+ mp = _PRICING_RE.search(html)
	213	+ if mp:
	214	+ out["price"] = _num(mp.group(1))
	215	+ out["price_label"] = f"{mp.group(1)} $"
	216	+
	217	+ md = _DESC_RE.search(html) or _DESC_RE2.search(html)
	218	+ if md:
	219	+ desc = _html.unescape(_TAG_RE.sub(" ", md.group(1)))
	220	+ desc = re.sub(r"\s+", " ", desc).strip()
	221	+ out["description"] = _ID_TAIL_RE.sub("", desc)[:6000]
	222	+
	223	+ mt = re.search(r"<title>(.*?)</title>", html, re.S)
	224	+ if mt:
	225	+ mc = _TITLE_CITY_RE.search(_html.unescape(mt.group(1)))
	226	+ if mc:
	227	+ # « Greater Sudbury (Valley East) » : le secteur part dans sector
	228	+ city = mc.group(1).split("(")[0].strip()
	229	+ if city and not any(c.isdigit() for c in city):
	230	+ out["city"] = city
	231	+ msec = re.search(r"$([^)]{2,45})$", mc.group(1))
	232	+ if msec and "sector" not in out:
	233	+ out["sector"] = msec.group(1).strip()
	234	+
	235	+ for n in du.ld_nodes(html):
	236	+ t = n.get("@type")
	237	+ types = set(t if isinstance(t, list) else [t])
	238	+ geo = n.get("geo") or {}
	239	+ if isinstance(geo, dict) and "lat" not in out:
	240	+ try:
	241	+ lat, lng = float(geo["latitude"]), float(geo["longitude"])
	242	+ if _BBOX[0] <= lat <= _BBOX[1] and _BBOX[2] <= lng <= _BBOX[3]:
	243	+ out["lat"], out["lng"] = lat, lng
	244	+ except (KeyError, TypeError, ValueError):
	245	+ pass
	246	+ if types & {"RealEstateAgent", "Organization"}:
	247	+ name = str(n.get("name") or "").strip()
	248	+ if name and "broker_name" not in out:
	249	+ out["broker_name"] = name[:120]
	250	+ tel = str(n.get("telephone") or "").strip()
	251	+ if tel and "broker_phone" not in out:
	252	+ out["broker_phone"] = tel[:40]
	253	+ if "lat" not in out:
	254	+ m = re.search(r'"latitude"\s:\s"?(-?\d{1,2}\.\d{3,})"?\s,\s'
	255	+ r'"longitude"\s:\s"?(-?\d{2,3}\.\d{3,})"?', html)
	256	+ if m:
	257	+ lat, lng = float(m.group(1)), float(m.group(2))
	258	+ if _BBOX[0] <= lat <= _BBOX[1] and _BBOX[2] <= lng <= _BBOX[3]:
	259	+ out["lat"], out["lng"] = lat, lng
	260	+
	261	+ gal = [u for u in dict.fromkeys(_DDFIMG_RE.findall(html))
	262	+ if "/listings/" in u.lower()]
	263	+ if gal:
	264	+ out["images"] = gal[:60]
	265	+
	266	+ if details:
	267	+ out["details"] = details
	268	+ return out
	269	+
	270	+
	271	+def _load() -> list[dict]:
	272	+ try:
	273	+ return json.loads(REGISTRY.read_text(encoding="utf-8"))
	274	+ except Exception:
	275	+ return []
	276	+
	277	+
	278	+# Génère une classe par site du registre.
	279	+for _ag in _load():
	280	+ if not all(_ag.get(k) for k in ("id", "site")):
	281	+ continue
	282	+ _sid = _ag["id"]
	283	+ globals()[f"REALTYPRESS_{_sid.upper()}"] = type(
	284	+ "RealtyPress" + "".join(p.title() for p in _sid.split("_")),
	285	+ (_RealtyPress,),
	286	+ {
	287	+ "source_id": _sid,
	288	+ "site_url": _ag["site"],
	289	+ "agency_name": _ag.get("name", _sid),
	290	+ "max_pages": int(_ag.get("max_pages", 150)),
	291	+ },
	292	+ )

modified immoka/geocode.py +34 −26

		`@@ -45,8 +45,11 @@ RETRY_FAILED_AFTER = 30 * 86400 # re-tenter les échecs après 30 jours`
45	45	# Bounding boxes (lat_min, lat_max, lng_min, lng_max)
46	46	_BBOX_QUEBEC = (46.55, 47.10, -71.75, -70.85) # Québec, Lévis, St-Augustin
47	47	# garde-fou provincial : couvre Gatineau, l'Abitibi, le Saguenay, la Côte-Nord
48		−# et la Gaspésie (expansion provinciale) — même boîte que schema.finalize()
	48	+# et la Gaspésie (expansion provinciale)
49	49	_BBOX_SUD_QC = (44.50, 63.00, -80.00, -56.00)
	50	+# territoire complet Groupe Ka : Québec + Ontario (Windsor -83°, Kenora -94.5°)
	51	+# — même boîte que schema.finalize()
	52	+_BBOX_KA = (41.60, 63.00, -95.50, -56.00)
50	53
51	54	_VILLES_QUEBEC = {"quebec", "levis", "saint-augustin-de-desmaures",
52	55	"l'ancienne-lorette", "ancienne-lorette"}
		`@@ -65,7 +68,7 @@ CLUSTER_MIN = 3 # une grappe de fiches devient une référence de confiance d`
65	68
66	69	def _bbox_for(city: str) -> tuple[float, float, float, float]:
67	70	key = strip_accents((city or "").strip().lower())
68		− return _BBOX_QUEBEC if key in _VILLES_QUEBEC else _BBOX_SUD_QC
	71	+ return _BBOX_QUEBEC if key in _VILLES_QUEBEC else _BBOX_KA
69	72
70	73
71	74	def _in_bbox(lat: float, lng: float, bbox: tuple) -> bool:
		`@@ -264,31 +267,35 @@ class Geocoder:`
264	267	return None
265	268	if not allow_api:
266	269	return None # pas de mémoire : on réessaiera quand l'API sera permise
267		− self.ville_api_calls += 1
268		− wait = REQUEST_DELAY - (time.time() - self._last)
269		− if wait > 0:
270		− time.sleep(wait)
271	270	v = None
272		− try:
273		− resp = self.session.get(NOMINATIM_URL, params={
274		− "q": f"{city}, Québec, Canada", "format": "jsonv2", "limit": 1,
275		− "countrycodes": "ca", "addressdetails": 1}, timeout=20)
276		− self._last = time.time()
277		− resp.raise_for_status()
278		− hits = resp.json()
279		− except Exception:
280		− self._last = time.time()
281		− return None # erreur réseau : ne pas cacher un échec définitif
282		− if hits:
	271	+ # priorité au Québec ; à défaut, l'Ontario (extension immo-ka)
	272	+ for prov in ("Québec", "Ontario"):
	273	+ self.ville_api_calls += 1
	274	+ wait = REQUEST_DELAY - (time.time() - self._last)
	275	+ if wait > 0:
	276	+ time.sleep(wait)
283	277	try:
284		− lat, lng = float(hits[0]["lat"]), float(hits[0]["lon"])
285		− if _in_bbox(lat, lng, _BBOX_SUD_QC):
286		− a = hits[0].get("address") or {}
287		− muni = (a.get("city") or a.get("town") or a.get("village")
288		− or a.get("municipality"))
289		− v = {"lat": lat, "lng": lng, "muni": muni}
290		− except (KeyError, ValueError, TypeError):
291		− v = None
	278	+ resp = self.session.get(NOMINATIM_URL, params={
	279	+ "q": f"{city}, {prov}, Canada", "format": "jsonv2", "limit": 1,
	280	+ "countrycodes": "ca", "addressdetails": 1}, timeout=20)
	281	+ self._last = time.time()
	282	+ resp.raise_for_status()
	283	+ hits = resp.json()
	284	+ except Exception:
	285	+ self._last = time.time()
	286	+ return None # erreur réseau : ne pas cacher un échec définitif
	287	+ if hits:
	288	+ try:
	289	+ lat, lng = float(hits[0]["lat"]), float(hits[0]["lon"])
	290	+ if _in_bbox(lat, lng, _BBOX_KA):
	291	+ a = hits[0].get("address") or {}
	292	+ muni = (a.get("city") or a.get("town") or a.get("village")
	293	+ or a.get("municipality"))
	294	+ v = {"lat": lat, "lng": lng, "muni": muni}
	295	+ except (KeyError, ValueError, TypeError):
	296	+ v = None
	297	+ if v:
	298	+ break
292	299	self.con.execute(
293	300	"INSERT INTO geocode_cache (address, lat, lng, muni, provider, failed, ts)"
294	301	" VALUES (?,?,?,?,?,?,?) ON CONFLICT(address) DO UPDATE SET"
		`@@ -474,7 +481,8 @@ def run_batch(limit: int \| None = None) -> dict:`
474	481	rows = con.execute(
475	482	"""SELECT uid, address, city FROM listings
476	483	WHERE active=1 AND lat IS NULL AND address<>'' AND geocode_failed=0
477		− AND dup_hidden=0 ORDER BY address""").fetchall()
	484	+ AND dup_hidden=0 AND COALESCE(region,'')<>'Ontario'
	485	+ ORDER BY address""").fetchall()
478	486	# 1 entrée par immeuble (adresse normalisée) ; on saute les échecs en cache
479	487	uniq: dict[str, dict] = {}
480	488	for r in rows:

modified immoka/normalize.py +16 −10

		`@@ -193,21 +193,26 @@ _TYPE_MAP = [`
193	193	# (mots-clés dans le texte source normalisé, type canonique)
194	194	(("plain-pied", "bungalow",), "Maison"),
195	195	(("maison a etages", "a etage", "deux etages", "cottage",), "Maison"),
196		− (("maison mobile", "unimodulaire",), "Maison mobile"),
197		− (("jumele", "semi-detache", "semi detache",), "Jumelé"),
198		− (("maison de ville", "townhouse", "en rangee",), "Maison de ville"),
199		− (("unifamiliale", "maison", "house", "residence", "split",), "Maison"),
	196	+ (("maison mobile", "unimodulaire", "mobile home", "manufactured home",),
	197	+ "Maison mobile"),
	198	+ (("jumele", "semi-detache", "semi detache", "semi-detached", "semi detached",),
	199	+ "Jumelé"),
	200	+ (("maison de ville", "townhouse", "en rangee", "row house", "row / town",),
	201	+ "Maison de ville"),
	202	+ (("unifamiliale", "maison", "house", "residence", "split", "detached",
	203	+ "single family",), "Maison"),
200	204	(("condo", "copropriete", "appartement", "apartment", "loft", "penthouse",
201	205	"studio",), "Condo"),
202	206	(("duplex",), "Duplex"),
203	207	(("triplex",), "Triplex"),
204	208	(("quadruplex", "quintuplex", "multiplex", "multilogement", "multi-logement",
205		− "immeuble a revenus", "revenus",), "Multiplex"),
206		− (("chalet", "cottage 4 saisons", "acces au plan d'eau", "bord de l'eau",),
207		− "Chalet"),
	209	+ "immeuble a revenus", "revenus", "multi-family", "multi family",
	210	+ "fourplex",), "Multiplex"),
	211	+ (("chalet", "cottage 4 saisons", "acces au plan d'eau", "bord de l'eau",
	212	+ "recreational",), "Chalet"),
208	213	(("terre", "terrain", "lot ", "land",), "Terrain"),
209		− (("ferme", "fermette", "agricole", "agriculture", "hobby farm",),
210		− "Fermette/Agricole"),
	214	+ (("ferme", "fermette", "agricole", "agriculture", "hobby farm", "farm",
	215	+ "acreage",), "Fermette/Agricole"),
211	216	# vocabulaire des fiches enrichies : « 4 logements », « propriété à revenu »
212	217	(("logements", "logement/", "unites et +", "revenu",), "Multiplex"),
213	218	(("bi generation", "bi-generation", "bigeneration", "intergeneration",),
		`@@ -215,7 +220,8 @@ _TYPE_MAP = [`
215	220	(("domaine et villa", "villa", "domaine",), "Maison"),
216	221	(("commercial", "commerce", "industriel", "industrie", "bureau", "local",
217	222	"entreprise", "batisse", "restaurant", "depanneur", "hotel", "motel",
218		− "garage/", "concessionnaire", "coiffure", "esthetique", "camping",),
	223	+ "garage/", "concessionnaire", "coiffure", "esthetique", "camping",
	224	+ "retail", "office", "industrial", "warehouse", "business",),
219	225	"Commercial"),
220	226	]
221	227

modified immoka/schema.py +3 −3

		`@@ -134,10 +134,10 @@ class PropertyListing:`
134	134	if not self.agency:
135	135	self.agency = self.broker_name
136	136
137		− # coordonnées fournies par la source : rejeter tout point hors de la
138		− # province (lat/lng inversés, 0/0, coquilles)
	137	+ # coordonnées fournies par la source : rejeter tout point hors du
	138	+ # territoire couvert — Québec + Ontario (lat/lng inversés, 0/0, coquilles)
139	139	if self.lat is not None and self.lng is not None:
140		− if not (44.5 <= self.lat <= 63.0 and -80.0 <= self.lng <= -56.0):
	140	+ if not (41.6 <= self.lat <= 63.0 and -95.5 <= self.lng <= -56.0):
141	141	self.lat = self.lng = None
142	142
143	143	return self
144	144