@faircopy/html

Extract static prose from rendered .html and .htm documents or fragments, with locations in the original HTML. Parsing never executes scripts or loads linked resources.

npm i -D faircopy @faircopy/html

import { defineConfig } from '@faircopy/config'
import { html } from '@faircopy/html'

export default defineConfig({
  files: ['src/site/data/resources/blog/**/*.html'],
  adapters: [html()],
  rules: { 'no-weasel-words': 'error' },
})

Inline elements keep their surrounding sentence intact. Block elements separate segments. Entities decode without losing source locations. Attributes, comments, URLs/email addresses, script, style, code, pre, kbd, samp, template, svg, math, noscript, and subtrees marked data-faircopy-ignore, hidden, or aria-hidden="true" are excluded. Code and URL spans separate prose segments.

This is static text extraction: CSS visibility, JavaScript-generated copy, and attribute copy such as alt and aria-label are outside its scope. Prefer the Markdown adapter for authored Markdown; use HTML for HTML-only content, a separate rendered-copy audit, or an independently maintained served surface. Including both representations produces separate findings in their own files.