wechat-html-normalizer

Run the normalize-html.mjs script with Node.js, passing the draft HTML and output path. It opens the file in headless Chromium via Puppeteer, computes the CSS cascade, and writes allowed computed styles into each element's inline style attribute.

The WeChat editor strips style tags and external stylesheets, so only inline styles survive. This normalizer moves all computed styles inline and removes style tags, scripts, and divs so the pasted article keeps its appearance.

No. Remote stylesheets are explicitly rejected and the script exits with an error listing the URLs. All CSS must be embedded in the draft's style tag or inline styles before normalization, and network requests are blocked during rendering.

Divs are replaced based on their computed display value: block-level divs become section elements and inline divs become span elements. All attributes and children are preserved during the conversion.

The script compares visible text, link count, and image count before and after processing. If they differ, it aborts with a content_integrity_mismatch error to prevent silent content loss, so check the draft for unusual markup.