Scraping¶

The offline demo needs none of this. casita demo renders the fixture.

Live scraping uses Playwright:

• Zillow: src/casita/zillow.py
• Craigslist: src/casita/craigslist.py
• Zumper: src/casita/zumper.py
• Redfin: src/casita/redfin.py

Zillow and Redfin can trigger PerimeterX. The human-in-the-loop path is:

uv run casita solve
uv run casita search --headed

solve opens a headed browser at Zillow. You clear the captcha manually, then the session persists in .chrome-profile/. Later headed searches reuse that profile. Craigslist, Zumper, and Redfin have their own source-specific parsing paths.

Ways This Could Go Further¶

Scraping could gain better source isolation, clearer retry reporting, richer fixtures, or tests around parser behavior. The current code preserves the human-in-the-loop reality instead of pretending the live scrape is fully unattended.