APEX FLOWWEB SCRAPER

APEX FLOW WEB SCRAPER / REFERENCE LIBRARY

How to write a reusable extraction recipe

An extraction recipe maps field names to selectors and validation rules. Good recipes are small, explicit and tested against more than one page layout.

By Apex Flow Labs · Reviewed 2026-10-08

Prefer meaningful selectors

Use stable content classes or semantic elements when available. Avoid long chains of positional selectors that break when a banner is inserted. For a product page, .product-title is often easier to maintain than div:nth-child(4) > div > h2.

Separate extraction and interpretation

Keep the displayed price as a string unless you have a tested currency and decimal conversion rule. A comma can be a decimal mark or a grouping separator. A missing price should remain missing, not become zero.

Use required fields carefully

Make the title required when every usable record needs one. Treat optional features as optional. Apex reports missing required fields and records selector matches so a reviewer can distinguish an empty page from a changed layout.

Recipe example

A title field can use {"selector":"h1","required":true}; a price field can use {"selector":".price"}. Save the recipe in the workspace, run it on a small sample, then inspect its field evidence before a larger job. Multiple matches deserve review: the first may belong to a recommendation card.

← All reference articles · Current capabilities and limits →