feat: add AI-friendly Markdown access

docusaurus.config.ts

@@ -22,6 +22,16 @@ const config: Config = {
     mermaid: true,
   },
   themes: ['@docusaurus/theme-mermaid'],
+  plugins: [
+    [
+      'docusaurus-plugin-copy-page-button',
+      {
+        placement: 'article',
+        enabledActions: ['copy'],
+        generateMarkdownRoutes: true,
+      },
+    ],
+  ],
 
   future: {
     v4: true,

package.json

@@ -5,7 +5,7 @@
   "scripts": {
     "docusaurus": "docusaurus",
     "start": "docusaurus start",
-    "build": "docusaurus build",
+    "build": "docusaurus build && node scripts/generate-llms-index.mjs",
     "check:frontmatter": "node scripts/validate-frontmatter.mjs",
     "check:encoding": "node scripts/check-encoding.mjs",
     "check:docs": "npm run check:frontmatter && npm run check:encoding && npm run build",
@@ -24,6 +24,7 @@
     "@docusaurus/theme-mermaid": "^3.10.0",
     "@mdx-js/react": "^3.0.0",
     "clsx": "^2.0.0",
+    "docusaurus-plugin-copy-page-button": "^0.8.4",
     "katex": "^0.16.45",
     "prism-react-renderer": "^2.3.0",
     "react": "^19.0.0",

scripts/generate-llms-index.mjs

@@ -0,0 +1,118 @@
+import {readdir, readFile, unlink, writeFile} from 'node:fs/promises';
+import path from 'node:path';
+import process from 'node:process';
+
+const outputDirectory = path.resolve(process.argv[2] ?? 'build');
+const excludedLocaleDirectories = new Set(['en']);
+
+async function findMarkdownFiles(directory) {
+  const entries = await readdir(directory, {withFileTypes: true});
+  const files = await Promise.all(
+    entries.map(async (entry) => {
+      const entryPath = path.join(directory, entry.name);
+      return entry.isDirectory() ? findMarkdownFiles(entryPath) : [entryPath];
+    }),
+  );
+
+  return files.flat().filter((file) => file.endsWith('.md'));
+}
+
+function getTitle(markdown, relativePath) {
+  const heading = markdown.match(/^#\s+(.+)$/m)?.[1]?.trim();
+  if (heading) {
+    return heading;
+  }
+
+  return path.basename(relativePath, '.md').replaceAll('-', ' ');
+}
+
+function isExcludedLocale(relativePath) {
+  const [topLevelDirectory] = relativePath.split(path.sep);
+  return excludedLocaleDirectories.has(topLevelDirectory);
+}
+
+function getHtmlPath(markdownFile) {
+  const relativePath = path.relative(outputDirectory, markdownFile);
+  if (relativePath === 'index.md') {
+    return path.join(outputDirectory, 'index.html');
+  }
+
+  const routePath = relativePath.slice(0, -path.extname(relativePath).length);
+  return path.join(outputDirectory, routePath, 'index.html');
+}
+
+async function isDocumentationPage(markdownFile) {
+  try {
+    const html = await readFile(getHtmlPath(markdownFile), 'utf8');
+    return html.includes('docs-doc-page');
+  } catch (error) {
+    if (error?.code === 'ENOENT') {
+      return false;
+    }
+    throw error;
+  }
+}
+
+function getSiteOrigin(sitemap) {
+  const firstLocation = sitemap.match(/<loc>(https?:\/\/[^<]+)<\/loc>/)?.[1];
+  if (!firstLocation) {
+    throw new Error('Unable to determine the site URL from build/sitemap.xml');
+  }
+
+  return new URL(firstLocation).origin;
+}
+
+const allMarkdownFiles = await findMarkdownFiles(outputDirectory);
+const documentationChecks = await Promise.all(
+  allMarkdownFiles.map((file) => isDocumentationPage(file)),
+);
+const defaultLocaleFiles = allMarkdownFiles.filter(
+  (file, index) =>
+    !isExcludedLocale(path.relative(outputDirectory, file)) &&
+    documentationChecks[index],
+);
+const excludedMarkdownFiles = allMarkdownFiles.filter(
+  (file) => !defaultLocaleFiles.includes(file),
+);
+
+await Promise.all(excludedMarkdownFiles.map((file) => unlink(file)));
+
+if (defaultLocaleFiles.length === 0) {
+  throw new Error('No default-locale Markdown pages were generated');
+}
+
+const sitemap = await readFile(path.join(outputDirectory, 'sitemap.xml'), 'utf8');
+const siteOrigin = getSiteOrigin(sitemap);
+const documents = await Promise.all(
+  defaultLocaleFiles.map(async (file) => {
+    const relativePath = path.relative(outputDirectory, file);
+    const publicPath = relativePath.split(path.sep).join('/');
+    const markdown = await readFile(file, 'utf8');
+
+    return {
+      title: getTitle(markdown, relativePath),
+      url: new URL(`/${publicPath}`, siteOrigin).toString(),
+    };
+  }),
+);
+
+documents.sort((left, right) =>
+  left.title.localeCompare(right.title, 'zh-CN'),
+);
+
+const llmsIndex = [
+  "# Prism's Docs",
+  '',
+  '> Prism-Committee 中文文档索引，为 AI 工具提供可直接读取的 Markdown 内容。',
+  '',
+  '## 中文文档',
+  '',
+  ...documents.map(({title, url}) => `- [${title}](${url})`),
+  '',
+].join('\n');
+
+await writeFile(path.join(outputDirectory, 'llms.txt'), llmsIndex, 'utf8');
+
+console.log(
+  `Generated llms.txt with ${documents.length} Chinese documents; removed ${excludedMarkdownFiles.length} non-Chinese or non-document Markdown routes.`,
+);

static/robots.txt

@@ -0,0 +1,4 @@
+User-agent: *
+Allow: /
+
+Sitemap: https://docs.akihito.dpdns.org/sitemap.xml