Refactor site generation into a modular content pipeline

html-checks.js

@@ -0,0 +1,31 @@
+import { parse } from 'parse5'
+
+const attribute = (node, name) => node.attrs?.find((item) => item.name === name)?.value
+
+export const htmlFacts = (html) => {
+  const facts = {
+    ids: [], hrefs: [], languages: [],
+    hasViewport: false, hasHeadingElement: false, isDraft: false,
+  }
+  const visit = (node) => {
+    if (node.tagName) {
+      const id = attribute(node, 'id')
+      if (id !== undefined) facts.ids.push(id)
+      const href = attribute(node, 'href')
+      if (href !== undefined) facts.hrefs.push(href)
+      for (const className of (attribute(node, 'class') ?? '').split(/\s+/)) {
+        if (className.startsWith('language-')) facts.languages.push(className.slice('language-'.length))
+      }
+      if (node.tagName === 'heading') facts.hasHeadingElement = true
+      if (node.tagName === 'meta') {
+        const name = attribute(node, 'name')?.toLowerCase()
+        if (name === 'viewport') facts.hasViewport = true
+        if (name === 'robots' && /(?:^|[,\s])noindex(?:$|[,\s])/i.test(attribute(node, 'content') ?? '')) facts.isDraft = true
+      }
+    }
+    for (const child of node.childNodes ?? []) visit(child)
+    if (node.content) visit(node.content)
+  }
+  visit(parse(html))
+  return facts
+}