Python 68.8%
TypeScript 18.6%
CSS 8.7%
JavaScript 3.3%
HTML 0.6%
1# -----------------------------------------------------------------------------2# Rent-Ka — Rental listings aggregator (Canada, outside Québec)3# Author: Simon-Pierre Boucher — contact@spboucher.ai4# connectors/kijiji.py : Kijiji (kijiji.ca) — FOR-RENT classifieds ONLY,5# every province EXCEPT Québec (province location ids verified live6# 2026-08-27):7# c37 apartments & condos for rent · c36 room rentals & roommates8# Pages list 40+ ads in __NEXT_DATA__ (Apollo state) with title, price,9# GPS, address, availability date and attributes (furnished, pets,10# inclusions…) — no private API needed. Adapted from Immo-Ka's for-sale11# connector.12# «Please contact» prices (price.type=CONTACT, ~3% of ads): the amount is13# often written in the description («Rent: $1,550/month») — recovered14# conservatively (amount glued to $ + a monthly word), otherwise the label15# shows «On request» rather than being empty.16# -----------------------------------------------------------------------------17from __future__ import annotations1819import json20import os21import re2223from ..schema import Listing24from .base import BaseConnector2526from . import _detailutil as du2728BASE = "https://www.kijiji.ca"29# (category code, URL segment, default unit type)30CATEGORIES = [31 (37, "b-apartments-condos", ""), # unit derived from attributes32 (36, "b-room-rental-roommate", "Room"),33]34# provinces outside Québec: (URL slug, Kijiji location id, province code)35PROVINCES = [36 ("ontario", "9004", "ON"),37 ("british-columbia", "9007", "BC"),38 ("alberta", "9003", "AB"),39 ("manitoba", "9006", "MB"),40 ("saskatchewan", "9009", "SK"),41 ("nova-scotia", "9002", "NS"),42 ("new-brunswick", "9005", "NB"),43 ("newfoundland", "9008", "NL"),44 ("prince-edward-island", "9011", "PE"),45]46MAX_PAGES = int(os.environ.get("RENTKA_KIJIJI_MAX_PAGES", "12")) # per prov/cat47DETAIL_LIMIT = int(os.environ.get("RENTKA_KIJIJI_DETAIL_LIMIT", "400"))4849# les annonces vivent sous des clés Apollo « RealEstateListing:123 » (c37)50# ou « StandardListing:123 » (c36)51_LISTING_KEY_RE = re.compile(r"^(?:RealEstate|Standard)Listing:\d+$")52_NEXT_RE = re.compile(53 r'<script id="__NEXT_DATA__" type="application/json">(.*?)</script>', re.S)5455# binary attributes -> displayable amenity (only when the value is true)56_AMENITY_LABELS = {57 "heat": "Heat included", "hydro": "Electricity included",58 "water": "Water included", "internet": "Internet included",59 "cabletv": "Cable TV included", "laundryinunit": "In-unit laundry",60 "laundryinbuilding": "Laundry in building", "dishwasher": "Dishwasher",61 "fridgefreezer": "Fridge/freezer", "airconditioning": "Air conditioning",62 "balcony": "Balcony", "elevator": "Elevator", "gym": "Gym",63 "pool": "Pool", "concierge": "Concierge",64 "twentyfourhoursecurity": "24-hour security",65 "storagelocker": "Storage locker",66 "bicycleparking": "Bicycle parking", "yard": "Yard",67 "wheelchairaccessible": "Wheelchair accessible",68}69_UNIT_TYPES = {70 "apartment": "Apartment", "condo": "Condo",71 "basement-apartment": "Basement apartment", "house": "House",72 "townhouse": "Townhouse", "duplex-triplex": "Duplex/Triplex",73}74_AGREEMENTS = {"one-year": "1-year lease", "month-to-month": "Month-to-month",75 "not-available": ""}76# common city fixes in Kijiji addresses77_CITY_FIX = {78 "st. johns": "St. John's", "st johns": "St. John's",79}808182# loyer mensuel écrit dans le texte (annonces « Sur demande ») : un montant83# DOIT toucher un « $ » ET un mot mensuel (mois/month) ou un libellé loyer/prix84_NUM = r"(\d{1,2}[\s,.]?\d{3}|\d{3,4})"85_PRICE_TXT_RE = re.compile(86 r"\b(?:loyer|prix|rent|price)\s*:?\s*(?:est\s+de\s+|de\s+|à partir de\s+)?"87 + _NUM + r"(?:[.,]\d{2})?\s*\$"88 r"|" + _NUM + r"(?:[.,]\d{2})?\s*\$\s*"89 r"(?:/|par\s+|per\s+)\s*(?:mois|month)"90 r"|\$\s*" + _NUM + r"(?:\.\d{2})?\s*(?:/|per\s+|a\s+)\s*month"91 # fourchette « $1,100 to $1,300/month » : capter la borne BASSE aussi92 r"|\$\s*" + _NUM + r"(?:\.\d{2})?\s*(?:to|à|[-–])\s*\$\s*[\d ,.]+"93 r"\s*(?:/|per\s+)\s*month",94 re.I)959697def _price_from_text(text: str) -> float | None:98 """Loyer mensuel plausible (300–12 000 $) déduit du texte de l'annonce.99100 Conservateur : montant collé à un « $ » et à un contexte mensuel101 (loyer/prix/rent ou /mois, /month). Le plus BAS des montants trouvés102 (« à partir de… ») ; None si rien de plausible — jamais inventé.103 """104 vals = []105 for m in _PRICE_TXT_RE.finditer(text or ""):106 raw = next(g for g in m.groups() if g)107 try:108 val = float(re.sub(r"[\s,.]", "", raw))109 except ValueError:110 continue111 if 300 <= val <= 12000:112 vals.append(val)113 return min(vals) if vals else None114115116def _fix_city(raw: str) -> str:117 key = (raw or "").strip().lower()118 if key in _CITY_FIX:119 return _CITY_FIX[key]120 return " ".join(w.capitalize() for w in key.replace("-", " ").split())121122123def _attr_value(a: dict) -> str:124 """Première valeur d'un attribut Apollo (canonique, sinon affichée)."""125 for k in ("canonicalValues", "values"):126 vals = a.get(k) or []127 if vals:128 return str(vals[0])129 return ""130131132def _apply_attrs(attrs: list[dict], out: dict) -> None:133 """Interprète les attributs Kijiji (mêmes clés en liste et en fiche)."""134 amenities = out.setdefault("amenities", [])135 details = out.setdefault("details", {})136 for a in attrs or []:137 cn = a.get("canonicalName") or ""138 val = _attr_value(a)139 if not val:140 continue141 if cn in _AMENITY_LABELS:142 if val == "1":143 amenities.append(_AMENITY_LABELS[cn])144 elif cn == "furnished":145 out["furnished"] = val == "1"146 elif cn == "petsallowed":147 out["pets"] = "yes" if val == "1" else "no"148 elif cn == "numberbedrooms":149 out["bedrooms"] = val # '0' = studio, else bedroom count150 elif cn == "numberbathrooms":151 try: # canonical in tenths: '15' = 1.5152 n = int(val) / 10153 details["bathrooms"] = f"{n:g}"154 except ValueError:155 pass156 elif cn in ("areainfeet", "sizesqft"):157 m = re.search(r"[\d.]+", val.replace(",", ""))158 if m and float(m.group(0)) > 0:159 out["area_sqft"] = float(m.group(0))160 elif cn == "dateavailable":161 m = re.match(r"(\d{4}-\d{2}-\d{2})", val)162 if m:163 out["availability_date"] = m.group(1)164 elif cn == "unittype":165 details["Unit type"] = _UNIT_TYPES.get(val, val)166 elif cn == "agreementtype":167 lease = _AGREEMENTS.get(val, val)168 if lease:169 details["Lease"] = lease170 elif cn == "numberparkingspots" and val.isdigit() and int(val) > 0:171 amenities.append(f"Parking ({val})")172173174def _parse_kijiji_detail(html: str) -> dict:175 """Fiche Kijiji : description complète, attributs, galerie haute résolution."""176 m = _NEXT_RE.search(html)177 if not m:178 return {}179 try:180 data = json.loads(m.group(1))181 except ValueError:182 return {}183 apollo = data.get("props", {}).get("pageProps", {}).get("__APOLLO_STATE__", {})184 it = next((v for k, v in apollo.items()185 if _LISTING_KEY_RE.match(k) and isinstance(v, dict)186 and v.get("description")), None)187 if not it:188 return {}189 out: dict = {}190 if it.get("description"):191 out["description"] = str(it["description"]).strip()[:6000]192 imgs = [re.sub(r"rule=kijijica-\d+-\w+", "rule=kijijica-1600-jpg", u)193 for u in it.get("imageUrls") or []]194 if imgs:195 out["images"] = imgs196 _apply_attrs((it.get("attributes") or {}).get("all") or [], out)197 out.pop("bedrooms", None) # le type d'unité est déjà fixé au niveau liste198 loc = it.get("location") or {}199 addr = (loc.get("address") or "").replace(", Canada", "")200 if re.match(r"\s*\d", addr):201 out["address"] = addr.split(",")[0]202 return out203204205class KijijiConnector(BaseConnector):206 source_id = "kijiji"207 request_delay = 1.2208209 def _page(self, seg: str, cat: int, page: int, prov_slug: str,210 loc_id: str) -> list[dict]:211 """Ads (Apollo state) from one category page of one province."""212 path = (f"{seg}/{prov_slug}/c{cat}l{loc_id}" if page == 1213 else f"{seg}/{prov_slug}/page-{page}/c{cat}l{loc_id}")214 html = self.get(f"{BASE}/{path}").text215 m = _NEXT_RE.search(html)216 data = json.loads(m.group(1)) if m else {}217 apollo = (data.get("props", {}).get("pageProps", {})218 .get("__APOLLO_STATE__", {}))219 return [v for k, v in apollo.items()220 if _LISTING_KEY_RE.match(k) and isinstance(v, dict)]221222 def _to_listing(self, it: dict, unit_default: str,223 province: str = "ON") -> Listing | None:224 lid = str(it.get("id") or "")225 url = it.get("url") or ""226 if not lid or not url:227 return None228 price = None229 pr = it.get("price") or {}230 if isinstance(pr, dict) and pr.get("amount"):231 price = round(pr["amount"] / 100.0, 0) # cents → $/mois232 loc = it.get("location") or {}233 coords = loc.get("coordinates") or {}234 address = (loc.get("address") or "").replace(", Canada", "")235 parts = [p.strip() for p in address.split(",") if p.strip()]236 # adresse à la française « 89, rue Dartois, Montréal » : le n° civique237 # arrive seul en tête — le recoller à la rue, sinon la rue devenait238 # la « ville » et polluait les filtres239 if len(parts) >= 2 and re.fullmatch(r"\d+[A-Za-z]?", parts[0]):240 parts = [f"{parts[0]} {parts[1]}"] + parts[2:]241 street = parts[0] if parts and re.match(r"\s*\d", parts[0]) else ""242 # the city = first element after the street that is neither the243 # province nor a postal code («street, city, ON M5V 1J1» formats)244 rest = [p for p in (parts[1:] if street else parts)245 if not re.match(r"(?i)^(on|bc|ab|sk|mb|nb|ns|pe|nl|yt|nt|nu|"246 r"ontario|british columbia|alberta|"247 r"saskatchewan|manitoba|new brunswick|"248 r"nova scotia|prince edward island|"249 r"newfoundland)\b", p)250 and not re.match(r"(?i)^[a-z]\d[a-z]", p)]251 city = _fix_city(rest[0] if rest else (loc.get("name") or ""))252 images = [re.sub(r"rule=kijijica-\d+-", "rule=kijijica-640-", u)253 for u in it.get("imageUrls") or []]254 extra: dict = {}255 _apply_attrs((it.get("attributes") or {}).get("all") or [], extra)256 unit_type = unit_default257 beds = extra.pop("bedrooms", None)258 if not unit_type and beds:259 try: # Kijiji sometimes codes «2.5» (2 bed + den)260 n = int(float(beds))261 except ValueError:262 n = 0263 unit_type = ("Studio" if n == 0264 else f"{n} bedroom" + ("s" if n > 1 else ""))265 lst = Listing(266 source=self.source_id,267 external_id=lid,268 url=url,269 title=it.get("title") or "",270 address=street,271 city=city,272 province=province,273 unit_type=unit_type,274 price=price,275 price_label=(f"${price:,.0f}/month" if price else ""),276 description=(it.get("description") or "")[:2000],277 amenities=extra.get("amenities") or [],278 details=extra.get("details") or {},279 images=images,280 lat=coords.get("latitude"),281 lng=coords.get("longitude"),282 )283 if extra.get("availability_date"):284 lst.availability_date = extra["availability_date"]285 lst.availability = f"Available {extra['availability_date']}"286 if extra.get("furnished") is not None:287 lst.furnished = extra["furnished"]288 if extra.get("pets"):289 lst.pets = extra["pets"]290 if extra.get("area_sqft"):291 lst.area_sqft = extra["area_sqft"]292 return lst293294 def fetch(self) -> list[Listing]:295 out: dict[str, Listing] = {}296 for prov_slug, loc_id, prov in PROVINCES:297 for cat, seg, unit_default in CATEGORIES:298 for page in range(1, MAX_PAGES + 1):299 try:300 items = self._page(seg, cat, page, prov_slug, loc_id)301 except Exception:302 break303 fresh = 0304 for it in items:305 lst = self._to_listing(it, unit_default, province=prov)306 if lst is not None and lst.uid not in out:307 out[lst.uid] = lst308 fresh += 1309 # nothing new (end page filled with repeated topAds)310 if fresh == 0 or len(items) < 10:311 break312 listings = list(out.values())313 du.enrich(self, listings, DETAIL_LIMIT, _parse_kijiji_detail, key="v1")314 # «please contact» prices: try the amount written in the ad text315 # (AFTER enrich: the full description comes from the detail page)316 for lst in listings:317 if lst.price is None:318 p = _price_from_text(f"{lst.title}\n{lst.description}")319 if p is not None:320 lst.price = p321 lst.price_label = f"${p:,.0f}/month (from description)"322 elif not lst.price_label:323 lst.price_label = "On request"324 return listings325