Skip to main content
CrAutos Scraper started as a personal itch: I wanted to buy a used car in Costa Rica, and staring at listing pages one at a time felt like a waste of a perfectly good weekend of programming. So I built an async Python scraper that pulls structured data on used cars from crautos.com and hands me back a clean dataset I can actually analyze.

Download the latest dataset

A ready-to-use CSV export, published straight from the repo’s releases.

​
What it does

It runs in two phases, both concurrent:
1

Listing phase

Concurrent POST requests gather the basics for every car — id, name, model year, and price in both colones and dollars.
2

Detail phase

Concurrent GET requests fetch the full spec sheet — mileage, engine, transmission, and equipment.
Then it cleans everything up (turning "91,500 kms" into the number 91500), normalizes fields, and applies a simple scoring formula so the best deals float to the top: 
score = year × 1 + price × (−1) + mileage × (−1)
Newer, cheaper, lower-mileage cars score higher. Crude, but surprisingly effective for a first pass.

​
The engineering

  • asyncio + httpx for concurrent requests, with semaphore-based rate limiting so I stay a polite guest on someone else’s server.
  • BeautifulSoup for parsing, pandas for the data, tqdm for a progress bar.
  • ruff, pytest and pre-commit hooks, because even a weekend project deserves to not rot.
I built this strictly for learning and research — it’s rate-limited on purpose, and the license forbids commercial use. It’s the kind of project I love: a real problem of my own, solved well enough to be genuinely useful.

Source on GitHub

miltonials/crautos-scrapper