Kangalou Scraper — Technical Analysis

Platform Overview

Kangalou is Quebec's #1 rental listings platform, owned by CORPIQ (Corporation des propriétaires immobiliers du Québec, 30,000 landlord members). Free for landlords to post, with paid visibility packages ($40-$540).

• ~7,500 active listings (mostly Montreal)
• ~1M+ monthly page views
• Bilingual (FR/EN)
• URL: https://kangalou.com

Data Available Per Listing

Field	Available	Notes
Title	Yes	Property type + size + location
Price (rent/month)	Yes	In dollars
Property type	Yes	Apartment, condo, house, room, loft, studio
Size (rooms)	Yes	Quebec notation: 3½, 4½, 5½
Bedrooms / Bathrooms	Yes	Numeric
Floor	Yes	Floor number
Area (sq ft / m²)	Yes	Both units
Availability date	Yes	Specific date or "now"
City / Neighborhood	Yes	Montreal neighborhoods
Description	Yes	Full text, mostly French
Images	Yes	Multiple, via kangalou.com/images/glide/
Coordinates (lat/lng)	Yes	Embedded in HTML — no geocoding needed
Street address	Partial	Some in URL slug, some neighborhood only
Amenities	Yes	Pets, parking, laundry, furnished, etc.
Contact name	Yes	Landlord name on page
Phone number	Auth-gated	Requires login + reCAPTCHA v3
Email	No	Message form only
Listing ID	Yes	Numeric (e.g., 222455)

Technical Architecture

Component	Details
Server	Nginx (no CDN, no Cloudflare)
Backend	PHP (custom, not WordPress/Laravel)
Frontend	jQuery, server-side rendered HTML
Maps	Google Maps
Anti-bot	None — no WAF, no CAPTCHA on search, no rate limiting

Internal API Endpoints

Discovered from JS bundle at /dist/js/kangalou.js:

Endpoint	Method	Auth	Purpose
/:lang/api/search	POST	No	Main search — returns JSON with HTML listings
/:lang/api/process	POST	No	Autocomplete/instant search
/:lang/api/listings/:id/contact-phone	GET	Yes (401)	Phone number reveal
/:lang/api/listings/:id/contact-message	POST	Yes	Send message
/:lang/api/listings/:id/visit-days	GET	?	Visit schedule
/:lang/api/postal-codes/:code/locations	GET	No	Postal code lookup

Search API

POST /:lang/api/search — the main discovery endpoint.

Parameters (form body):

• search-string — text query (e.g., "montreal")
• search-filter-city — city filter
• search-filter-sort — sort order (newest, pasc)
• page — pagination (1-indexed)
• s — size filter
• t[] — type filter (array)
• pmin / pmax — price range
• f[] — feature/amenity filter
• g[], z[] — unknown filters

Response: JSON { ok: true, html: "<div>...</div>" } — server-rendered HTML with listing cards.

~53 listings per page, ~142 pages for full inventory.

URL Structure

• Search: /:lang/search/ or /:lang/recherche/
• City: /:lang/location/:city/
• Neighborhood: /:lang/location/:city/:neighborhood/
• Detail: /:lang/listing/:slug/:id/ (EN) or /:lang/annonce/:slug/:id/ (FR)

Anti-Bot Assessment

Measure	Status
Cloudflare / WAF	None
CAPTCHA on search	None
CAPTCHA on phone	reCAPTCHA v3 (site key: 6LepJboUAAAAAIWJCyG8PzlT2kTSl6fGmVtRSZXi)
Rate limiting	None detected
Bot detection JS	None
User-Agent check	None — curl works
IP blocking	Not observed

Verdict: Very low protection. Orders of magnitude easier than Kijiji.

Recommended Scraping Approach

HTTP-only — no Playwright/browser needed.

1. Discovery: POST to /en/api/search with pagination → listing URLs + basic card data
2. Detail: GET each listing page → full data (description, coordinates, amenities, images, contact name)
3. Phone numbers: Skip unless account is created (auth + reCAPTCHA v3)

Stack: undici/fetch + cheerio for HTML parsing. Fits existing BaseScraper interface.

Estimated Scrape Time

Phase	Volume	Time (sequential)	Time (5 concurrent)
Search pages	142 pages × 18s	~43 min	~9 min
Detail pages	7,500 × 3.6s	~7.5 hours	~1.5 hours
Total			~2 hours

Phone Numbers

Phone numbers are behind auth at /:lang/api/listings/:id/contact-phone (returns 401 without login). Also protected by reCAPTCHA v3.

Options:

1. Create an account, solve reCAPTCHA v3 tokens programmatically — moderate difficulty
2. Skip phone numbers initially, only import listings with phones from other sources
3. Use Playwright for phone reveal (same approach as Kijiji) — heavier but proven

Legal Risk

Moderate — same category as Kijiji.

ToS prohibitions:

• "use, download or otherwise copy... any directory of Users"
• "gather or otherwise collect information about others, including email addresses"
• All content copyrighted by Kangalou/CORPIQ

Mitigating factors:

• Listing data is publicly accessible without login
• No aggressive anti-bot enforcement suggests they tolerate some scraping
• Phone numbers are the most sensitive data (auth-gated)

Risk owner: CORPIQ — organized Quebec landlord association with resources to enforce.

Mitigation: Avoid storing landlord names. Rate-limit respectfully. Don't scrape phone numbers without careful consideration.

Comparison to Kijiji

Factor	Kijiji	Kangalou
Anti-bot	High (Cloudflare, reCAPTCHA Enterprise)	Very low (nothing)
Browser needed	Yes (Playwright)	No (HTTP + cheerio)
Data format	Apollo GraphQL JSON	Server-rendered HTML
Phone numbers	reCAPTCHA Enterprise (hard)	Auth + reCAPTCHA v3 (moderate)
Listings (MTL)	~5,000-10,000	~7,500
Rate limiting	Aggressive	None
Implementation effort	2-3 weeks	3-5 days
Overlap with Kijiji	N/A	Low-moderate
Coordinates included	No (need geocoding)	Yes (embedded)

Implementation Priority

High priority — easiest scraper to build. Kangalou should be the next scraper after Kijiji because:

1. Zero anti-bot → simple HTTP client
2. No browser dependency → faster, less resource-intensive
3. Coordinates embedded → no geocoding API calls needed
4. Good data quality (CORPIQ-verified landlords)
5. Complementary inventory to Kijiji
6. 3-5 day implementation estimate