ozon-grabber

CLI that scans your Ozon orders/returns via a real Chrome session and submits the parsed data to a SQLite-backed service.

The CLI launches Chrome through chrome-devtools-mcp, navigates the order/return pages it has access to, dumps each page's HTML, parses it locally with jsdom, and POSTs the result to a configurable backend (the companion service in ozon-orders-history_v2/backend).

Quick start

# 1. log in once — Chrome opens, log in to Ozon, press Ctrl+C
npx ozon-grabber login

# 2. scan orders. user-id and backend-url are picked up from env;
#    --start-order is fetched from the backend (auto-resume).
npx ozon-grabber start --verbose

# 3. scan returns
npx ozon-grabber start --returns --verbose

Configuration

Set these once in your shell profile (e.g. ~/.zshrc):

export OZON_USER_ID=12345678
export OZON_BACKEND_URL=http://localhost:3000           # or your remote service

Both can also be passed as flags (--user-id, --backend-url) — flags override env.

Flags

| Flag | Default | Purpose | |---|---|---| | --user-id | $OZON_USER_ID | Ozon account id (required if env unset) | | --start-order | from GET /scan-state/<user-id> | Where to begin. Omit to resume | | --max-orders | unlimited | Cap how many to scan in this run | | --returns | false | Scan returns (/my/returnDetails) instead of orders | | --output <path> | stdout | Where to write the JSON summary | | --backend-url | $OZON_BACKEND_URL or http://localhost:3000 | Submit URL | | --backend / --no-backend | true | Submit POST per parsed result (or skip with --no-backend) | | --user-data-dir | ./chrome-profile | Chrome profile directory (cookies + login persisted here) | | --verbose | false | Per-order log lines |

Notes

• The Chrome profile (chrome-profile/ by default) carries your Ozon cookies. Don't share it.
• First navigation can hit the Ozon anti-bot challenge — the CLI auto-retries Navigation timeout up to three times.
• --no-backend outputs JSON to --output (or stdout) and skips submission. Useful for offline parsing.

License

MIT