feat: enrichissement fiches Kijiji (description, attributs, galerie 1600px) et LesPAC (adresse civique, caractéristiques, galerie basephoto) + champs annexes Ubee

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Simon-Pierre Boucher committed 1 mo ago (Aug 13, 2026) parent 798091f

4 changed files +141 −4

modified immoka/connectors/_detailutil.py +4 −1

		`@@ -157,7 +157,10 @@ def apply_detail(lst: PropertyListing, d: dict) -> None:`
157	157	lst.details.update(d["details"])
158	158	if d.get("broker_name"):
159	159	lst.broker_name = d["broker_name"]
160		− for f in ("description", "price_label"):
	160	+ # description : on garde la plus riche (la fiche détail bat le résumé liste)
	161	+ if d.get("description") and len(d["description"]) > len(lst.description or ""):
	162	+ lst.description = d["description"]
	163	+ for f in ("price_label", "address", "city", "sector", "property_type"):
161	164	if d.get(f) and not getattr(lst, f, ""):
162	165	setattr(lst, f, d[f])
163	166	for f in ("bedrooms", "bathrooms", "powder_rooms", "year_built",

modified immoka/connectors/kijiji.py +66 −1

		`@@ -16,6 +16,8 @@ import re`
16	16	from ..schema import PropertyListing
17	17	from .base import BaseConnector
18	18
	19	+from . import _detailutil as du
	20	+
19	21	BASE = "https://www.kijiji.ca"
20	22	# (code catégorie, segment d'URL, type canonique)
21	23	CATEGORIES = [
		`@@ -24,6 +26,67 @@ CATEGORIES = [`
24	26	(641, "b-terrain-a-vendre", "Terrain"),
25	27	]
26	28	MAX_PAGES = int(os.environ.get("IMMOKA_KIJIJI_MAX_PAGES", "100"))
	29	+DETAIL_LIMIT = int(os.environ.get("IMMOKA_KIJIJI_DETAIL_LIMIT", "400"))
	30	+
	31	+_ATTR_LABELS = {
	32	+ "numberbedrooms": "Chambres", "numberbathrooms": "Salles de bain",
	33	+ "areainfeet": "Superficie (pi²)", "forsalebyhousing": "À vendre par",
	34	+ "yearbuilt": "Année de construction", "sizesqft": "Superficie (pi²)",
	35	+}
	36	+
	37	+
	38	+def _parse_kijiji_detail(html: str) -> dict:
	39	+ """Fiche Kijiji : description complète, attributs, galerie haute résolution."""
	40	+ m = re.search(
	41	+ r'<script id="__NEXT_DATA__" type="application/json">(.*?)</script>',
	42	+ html, re.S)
	43	+ if not m:
	44	+ return {}
	45	+ try:
	46	+ data = json.loads(m.group(1))
	47	+ except ValueError:
	48	+ return {}
	49	+ apollo = data.get("props", {}).get("pageProps", {}).get("__APOLLO_STATE__", {})
	50	+ it = next((v for k, v in apollo.items()
	51	+ if k.startswith("StandardListing:") and isinstance(v, dict)
	52	+ and v.get("description")), None)
	53	+ if not it:
	54	+ return {}
	55	+ out: dict = {}
	56	+ if it.get("description"):
	57	+ out["description"] = str(it["description"]).strip()[:6000]
	58	+ imgs = [re.sub(r"rule=kijijica-\d+-\w+", "rule=kijijica-1600-jpg", u)
	59	+ for u in it.get("imageUrls") or []]
	60	+ if imgs:
	61	+ out["images"] = imgs
	62	+ features, details = [], {}
	63	+ for a in (it.get("attributes") or {}).get("all") or []:
	64	+ cn = a.get("canonicalName") or ""
	65	+ val = ", ".join(str(v) for v in a.get("values") or [])
	66	+ if not val:
	67	+ continue
	68	+ label = _ATTR_LABELS.get(cn, a.get("name") or cn)
	69	+ features.append(f"{label} : {val}")
	70	+ details[label] = val
	71	+ if cn == "numberbedrooms" and val.isdigit():
	72	+ out["bedrooms"] = int(val)
	73	+ elif cn == "numberbathrooms" and val.isdigit():
	74	+ out["bathrooms"] = int(val)
	75	+ elif cn in ("areainfeet", "sizesqft"):
	76	+ mn = re.search(r"[\d.]+", val.replace(",", ""))
	77	+ if mn:
	78	+ out["area_sqft"] = float(mn.group(0))
	79	+ elif cn == "yearbuilt" and val.isdigit():
	80	+ out["year_built"] = int(val)
	81	+ if features:
	82	+ out["features"] = features
	83	+ if details:
	84	+ out["details"] = details
	85	+ loc = it.get("location") or {}
	86	+ addr = (loc.get("address") or "").replace(", Canada", "")
	87	+ if re.match(r"\s*\d", addr):
	88	+ out["address"] = addr.split(",")[0]
	89	+ return out
27	90
28	91
29	92	class KijijiConnector(BaseConnector):
		`@@ -99,4 +162,6 @@ class KijijiConnector(BaseConnector):`
99	162	# plus rien de neuf (page de fin remplie de topAds répétés)
100	163	if fresh == 0 or len(items) < 10:
101	164	break
102		− return list(out.values())
	165	+ listings = list(out.values())
	166	+ du.enrich(self, listings, DETAIL_LIMIT, _parse_kijiji_detail, key="v1")
	167	+ return listings

modified immoka/connectors/lespac.py +63 −1

		`@@ -10,13 +10,18 @@`
10	10	# -----------------------------------------------------------------------------
11	11	from __future__ import annotations
12	12
	13	+import html as _html
13	14	import json
	15	+import os
14	16	import re
15	17
16	18	from ..schema import PropertyListing
17	19	from .base import BaseConnector
18	20
	21	+from . import _detailutil as du
	22	+
19	23	BASE = "https://www.lespac.com"
	24	+DETAIL_LIMIT = int(os.environ.get("IMMOKA_LESPAC_DETAIL_LIMIT", "400"))
20	25	CATEGORIES = [
21	26	(37, "immobilier-achat-vente-residentiel", "Maison"),
22	27	(38, "immobilier-achat-vente-terrains", "Terrain"),
		`@@ -89,4 +94,61 @@ class LesPacConnector(BaseConnector):`
89	94	if lst is not None:
90	95	out.setdefault(lst.uid, lst)
91	96	page += 1
92		− return list(out.values())
	97	+ listings = list(out.values())
	98	+ du.enrich(self, listings, DETAIL_LIMIT, _parse_lespac_detail, key="v1")
	99	+ return listings
	100	+
	101	+
	102	+def _clean(s: str) -> str:
	103	+ return re.sub(r"\s+", " ", _html.unescape(re.sub(r"<[^>]+>", " ", s))).strip()
	104	+
	105	+
	106	+def _parse_lespac_detail(html: str) -> dict:
	107	+ """Fiche LesPAC : description complète, adresse civique, caractéristiques
	108	+ (boîte « Caractéristiques » : <p><span>Label</span><span>Valeur</span></p>)
	109	+ et galerie pleine taille (binary/basephoto)."""
	110	+ out: dict = {}
	111	+ md = re.search(r'class="description"[^>]>(.?)</(?:p\|div)>', html, re.S \| re.I)
	112	+ if md:
	113	+ desc = _clean(md.group(1))
	114	+ if desc:
	115	+ out["description"] = desc[:6000]
	116	+ features, details = [], {}
	117	+ mbox = re.search(r'>Caractéristiques</p>\s<div class="box">(.?)</div>',
	118	+ html, re.S)
	119	+ if mbox:
	120	+ for lm, vm in re.findall(r"<p><span>(.?)</span>\s<span>(.*?)</span>",
	121	+ mbox.group(1), re.S):
	122	+ label, value = _clean(lm), _clean(vm)
	123	+ if not label or not value:
	124	+ continue
	125	+ features.append(f"{label} : {value}")
	126	+ details[label] = value
	127	+ if label == "Adresse" and re.match(r"\s*\d", value):
	128	+ out["address"] = value
	129	+ elif label == "Année":
	130	+ my = re.search(r"(18\|19\|20)\d{2}", value)
	131	+ if my:
	132	+ out["year_built"] = int(my.group(0))
	133	+ elif label == "Type de propriété":
	134	+ out["property_type"] = value
	135	+ elif "chambre" in label.lower():
	136	+ mn = re.search(r"\d+", value)
	137	+ if mn:
	138	+ out["bedrooms"] = int(mn.group(0))
	139	+ elif "salle" in label.lower() and "bain" in label.lower():
	140	+ mn = re.search(r"\d+", value)
	141	+ if mn:
	142	+ out["bathrooms"] = int(mn.group(0))
	143	+ if features:
	144	+ out["features"] = features
	145	+ if details:
	146	+ out["details"] = details
	147	+ imgs, seen = [], set()
	148	+ for u in re.findall(r'https://cdn\.lespac\.com/binary/basephoto/\d+\.jpg', html):
	149	+ if u not in seen:
	150	+ seen.add(u)
	151	+ imgs.append(u)
	152	+ if imgs:
	153	+ out["images"] = imgs
	154	+ return out

modified immoka/connectors/ubee.py +8 −1

		`@@ -72,7 +72,14 @@ class UbeeConnector(BaseConnector):`
72	72	lot_sqft=round(land * M2_TO_SQFT) if land else None,
73	73	year_built=it.get("yearBuilt"),
74	74	details={k: it.get(k) for k in
75		− ("propertyType", "buildingType", "toBuild") if it.get(k)},
	75	+ ("propertyType", "buildingType", "toBuild", "taxable",
	76	+ "openHouseDetail", "isOnlineSince") if it.get(k)},
	77	+ features=[f for f in (
	78	+ f"Type de bâtiment : {it['buildingType']}" if it.get("buildingType") else "",
	79	+ f"Sous-type : {it['propertyType']}" if it.get("propertyType") else "",
	80	+ "Neuf / à construire" if it.get("toBuild") else "",
	81	+ "Prix taxable (+tx)" if it.get("taxable") else "",
	82	+ ) if f],
76	83	images=images,
77	84	lat=it.get("latitude"),
78	85	lng=it.get("longitude"),
79	86