Refactor site generation into a modular content pipeline

test/html-checks.test.js

@@ -0,0 +1,23 @@
+import assert from 'node:assert/strict'
+import { test } from 'node:test'
+import { htmlFacts } from '../html-checks.js'
+
+test('extracts decoded attributes regardless of quote style or order', () => {
+  const facts = htmlFacts(`<!doctype html><meta content='width=device-width' name='viewport'>
+    <a class='button' href='#a&amp;b'>Go</a><h2 id='a&amp;b'>Title</h2>
+    <code class='extra language-js another'>code</code>`)
+  assert.equal(facts.hasViewport, true)
+  assert.deepEqual(facts.ids, ['a&b'])
+  assert.deepEqual(facts.hrefs, ['#a&b'])
+  assert.deepEqual(facts.languages, ['js'])
+})
+
+test('ignores markup-like text in scripts and comments', () => {
+  const facts = htmlFacts(`<script>const example = '<a href="/missing"><i id="fake"></i>'</script>
+    <!-- <heading id="also-fake"></heading> -->
+    <meta name='robots' content='nofollow, noindex'>`)
+  assert.deepEqual(facts.ids, [])
+  assert.deepEqual(facts.hrefs, [])
+  assert.equal(facts.hasHeadingElement, false)
+  assert.equal(facts.isDraft, true)
+})