website-cloning

Use Playwright to save the target page's raw HTML, screenshots, design tokens, and all images before building. Then create a React + Vite app, replace the scaffold CSS with plain CSS using extracted design tokens, and build each section component from the raw HTML in exact DOM order.

Navigate with wait_until set to domcontentloaded, wait 5 seconds, then save page.content() as raw.html. Use page.evaluate to collect computed styles, CSS variables, fonts, images, and SVGs, and download all assets with retry logic and browser User-Agent headers.

Sites with persistent WebSocket or analytics connections never reach networkidle, causing page.goto to hang. Use wait_until domcontentloaded with a 60-second timeout, then add page.wait_for_timeout(5000) to let client-side rendering finish.

CDNs like Shopify and Cloudinary block requests without browser headers. Send a real browser User-Agent string with each request, retry with fallback agents, and upscale resolution by rewriting URL patterns such as _small.jpg to _1200x.jpg.

Cloning is acceptable for your own sites, client work, learning, staging copies, or platform migration. It is not acceptable for impersonation, phishing, or trademark infringement, and non-owned clones should replace logos, brand names, and product data.

Check the URL for locale paths like /es-do or /fr and build all user-facing text in that language, even if the raw HTML returns English due to client-side localization. Keep brand terms, product names, and color names in their original language.