Get Places

Local personal-use scraper for finding small businesses by city and category.

Stack

• Frontend: React, Vite, HTML/CSS
• Backend: Node 20, Express
• Scraping: Playwright

Setup

pnpm install
pnpm --filter @get-places/server exec playwright install chromium
pnpm dev

Frontend: http://localhost:5173
Backend: http://localhost:3001

API

POST /api/scrape

Body:

{
  "city": "Cebu City",
  "category": "Food, Beverage & Entertainment (Discretionary Dining)",
  "keyword": "coffee shop",
  "limit": 50
}

Validation errors return a normal JSON error response (400). On success, the response streams as newline-delimited JSON (Content-Type: application/x-ndjson) so the client can render rows as they're found instead of waiting for the whole batch:

{"type":"status","message":"Searching Google Maps for \"coffee shop in Cebu City\"..."}
{"type":"status","message":"Found 12 places, checking each one..."}
{"type":"row","row":{"shopName":"Example Coffee","email":"-","number":"-","accountLink":"https://www.instagram.com/example","facebookLink":"https://www.facebook.com/example","category":"coffee shop","location":"Cebu City","mapsUrl":"https://www.google.com/maps/place/Example+Coffee/..."},"index":1,"total":50}
{"type":"done","count":12,"csv":"Shop Name,Email,Number,..."}

An {"type":"error","message":"..."} line can appear instead of done if the scrape fails partway through (the HTTP status is already 200 by then since streaming has started).

The scraper only uses public pages, runs without login, and stops at the configured result limit. Each row includes its mapsUrl (the Google Maps listing) so you can jump straight to it.

Rows are scraped concurrently (SCRAPER_CONCURRENCY, default 5) to cut wall-clock time for a 50-row run, with a randomized 1.2-2.6s delay between requests on each worker to avoid triggering rate limiting or blocking from Google Maps or target sites. Raise SCRAPER_CONCURRENCY for more speed at the cost of higher block risk; lower it if you start seeing captchas or empty results.