IScrape

Describe what you need.
Get the data.

Point us to a website. Tell us what you need in English. Our AI extracts the data for you as CSV, JSON or Excel.

Listings Product pages Search results Directories JavaScript sites Anything with rows

Free to try. No card.

Google sign-in is not configured on this server, so there is no way to sign in. Set GOOGLE_CLIENT_ID and GOOGLE_CLIENT_SECRET.

Say what you want

A sentence, not a selector. “Every listing with its price and link” is a complete specification.

It writes the scraper

The model reads the page once and writes the extractor. If a column comes back empty it rewrites itself and tries again.

Take the file

CSV, JSON or Excel, ready to download.

What it handles

The awkward pages,
not just the tidy ones

Most scrapers work on a clean listing and give up after that. These are the cases that come up in practice.

Pages built by JavaScript

If the rows are not in the HTML, it opens a real browser and reads the page after it renders. Infinite scroll included.

Sites that would rather you didn't

If a page turns a plain request away, it tries harder until the page loads. It never pretends a blocked page worked.

Links behind countdowns

Shorteners, redirect chains and ad gates get followed to the thing they point at, so the file holds real links.

Markup that changed overnight

When a column comes back empty the scraper is rewritten against the live page and only kept if it does better.

Values that are not really there

Every value is lifted out of the page by selector. Nothing is generated, so nothing can be invented.

More than one page

Follows pagination, and can open each row's own page to fill in what the listing leaves out.

Point it at a page.

Continue with Google

Free to try. No card.

CSV, Excel, JSON, Markdown and plain text out.

Extraction Settings

Deep Crawl Follow internal links across the site Unblocker Route through a provider for protected sites Unshorten links Resolve ouo.io, gplinks, bit.ly and other ad gates in the results to the real destination. Slow: a gate can need a browser. Rewrite the scraper Ignore the cached recipe and plan it again. Use this when the site has changed.

Pages to follow Open each item's own page Slower, but gets descriptions and specs Max item pages

Proxy optional Delay between requests Fast mode 0.2s between requests, no long break, no warm-up. For sites you own or that welcome the traffic — about 3x quicker. A delay set above still wins. Tidy the output Strip repeated labels and stray spacing, and order the columns the way you asked for them Drop empty columns Hide fields nothing was found for Unshorten: try every link Not just recognised shorteners. Use this when a site wraps links in its own redirector — one extra request per distinct link.

Pages are fetched whatever a site's robots.txt asks unless this server is configured otherwise. Make sure you have the right to the data you collect.

Recents

Nothing yet.

Best on a desktop

IScrape puts wide tables of extracted data side by side with the settings that produced them, and that needs a real screen.

Open iscrape.site on a laptop or desktop to get started.