Refactor site generation into a modular content pipeline

check.js

@@ -1,8 +1,9 @@
 import { execFileSync } from 'node:child_process'
 import { readdir, readFile, stat } from 'node:fs/promises'
 import { join } from 'node:path'
+import { generatedRoot, siteUrl } from './site.config.js'
+import { htmlFacts } from './html-checks.js'
 
-const generatedRoot = '.generated'
 
 const walk = async (directory) => {
   const files = []
@@ -45,21 +46,22 @@ const prismLanguageAliases = { sh: 'bash', shell: 'bash' }
 
 for (const file of htmlFiles) {
   const html = await readFile(file, 'utf8')
-  const ids = [...html.matchAll(/\sid="([^"]+)"/g)].map((match) => match[1])
+  const { ids, hrefs, languages, hasViewport, hasHeadingElement, isDraft } = htmlFacts(html)
   const duplicateIds = ids.filter((id, index) => ids.indexOf(id) !== index)
   if (duplicateIds.length > 0) errors.push(`${file}: duplicate IDs: ${[...new Set(duplicateIds)].join(', ')}`)
-  if (!html.includes('<meta name="viewport"')) errors.push(`${file}: missing viewport metadata`)
-  if (html.includes('<heading')) errors.push(`${file}: contains a nonstandard <heading> element`)
-  if (html.includes('name="robots"') && html.includes('noindex')) {
+  if (!hasViewport) errors.push(`${file}: missing viewport metadata`)
+  if (hasHeadingElement) errors.push(`${file}: contains a nonstandard <heading> element`)
+  if (isDraft) {
     noindexFiles.add(file)
     errors.push(`${file}: draft page must not be in production output`)
   }
 
-  for (const [, fragment] of html.matchAll(/href="#([^"]+)"/g)) {
+  for (const href of hrefs.filter((value) => value.startsWith('#') && value.length > 1)) {
+    const fragment = decodeURIComponent(href.slice(1))
     if (!ids.includes(fragment)) errors.push(`${file}: missing fragment target #${fragment}`)
   }
 
-  for (const [, href] of html.matchAll(/href="(\/[^"]*)"/g)) {
+  for (const href of hrefs.filter((value) => value.startsWith('/'))) {
     if (href === '/') continue
     const pathname = href.split(/[?#]/, 1)[0]
     const relative = pathname.replace(/^\//, '')
@@ -70,7 +72,7 @@ for (const file of htmlFiles) {
     if (!(await exists(target))) errors.push(`${file}: broken internal link ${href}`)
   }
 
-  for (const [, language] of html.matchAll(/class="language-([\w-]+)"/g)) {
+  for (const language of languages) {
     if (builtInPrismLanguages.has(language)) continue
     const component = prismLanguageAliases[language] ?? language
     if (!clientSource.includes(`prism-${component}`))
@@ -78,7 +80,6 @@ for (const file of htmlFiles) {
   }
 }
 
-const siteUrl = 'https://paulw.xyz'
 const generatedFile = (pathname) => {
   const relative = pathname.replace(/^\//, '')
   return pathname.endsWith('/') ? join(generatedRoot, relative, 'index.html') : join(generatedRoot, relative)