docs/verify-links: validate links to docs from other parts of the microsite

docs/verify-links.js

@@ -15,13 +15,13 @@
  * limitations under the License.
  */
 
-const { resolve: resolvePath, dirname } = require('path');
+const { resolve: resolvePath, join: joinPath, dirname } = require('path');
 const fs = require('fs-extra');
 const recursive = require('recursive-readdir');
 
 const projectRoot = resolvePath(__dirname, '..');
 
-async function verifyUrl(basePath, url) {
+async function verifyUrl(basePath, url, docPages) {
   // Avoid having absolute URL links within docs/, so that links work on the site
   if (
     url.match(
@@ -29,7 +29,7 @@ async function verifyUrl(basePath, url) {
     ) &&
     basePath.match(/^(?:docs|microsite)\//)
   ) {
-    return { url, basePath, problem: 'absolute' };
+    return { url, basePath, problem: 'github' };
   }
 
   url = url.replace(/#.*$/, '');
@@ -49,8 +49,16 @@ async function verifyUrl(basePath, url) {
   let path = '';
 
   if (url.startsWith('/')) {
-    if (url.startsWith('/docs/') && basePath.match(/^(?:docs|microsite)\//)) {
-      return { url, basePath, problem: 'not-relative' };
+    if (url.startsWith('/docs/')) {
+      if (basePath.match(/^(?:docs)\//)) {
+        return { url, basePath, problem: 'not-relative' };
+      }
+      if (basePath.startsWith('microsite/')) {
+        if (docPages.has(url)) {
+          return;
+        }
+        return { url, basePath, problem: 'doc-missing' };
+      }
     }
 
     const staticPath = resolvePath(projectRoot, 'microsite/static', `.${url}`);
@@ -71,14 +79,14 @@ async function verifyUrl(basePath, url) {
   return;
 }
 
-async function verifyFile(filePath) {
+async function verifyFile(filePath, docPages) {
   const content = await fs.readFile(filePath, 'utf8');
   const mdLinks = content.match(/\[.+?\]\(.+?\)/g) || [];
   const badUrls = [];
 
   for (const mdLink of mdLinks) {
     const url = mdLink.match(/\[.+\]\((.+)\)/)[1].trim();
-    const badUrl = await verifyUrl(filePath, url);
+    const badUrl = await verifyUrl(filePath, url, docPages);
     if (badUrl) {
       badUrls.push(badUrl);
     }
@@ -87,20 +95,46 @@ async function verifyFile(filePath) {
   return badUrls;
 }
 
+// This discovers the doc paths as they will be available on the microsite.
+// It is used to validate microsite links from outside /docs/, as those
+// are not transformed from the markdown file representation by docusaurus.
+async function findExternalDocsLinks(dir) {
+  const allFiles = await recursive(dir);
+  const mdFiles = allFiles.filter(p => p.endsWith('.md'));
+
+  const paths = new Map();
+
+  for (const file of mdFiles) {
+    const content = await fs.readFile(file, 'utf8');
+    const url = `/${file}`;
+    const match = content.match(/---(?:\r|\n|.)*^id: (.*)$/m);
+
+    // Both docs with an id and without should remove trailing /index
+    const realPath = (match
+      ? joinPath(dirname(url), match[1])
+      : url.replace(/\.md$/, '')
+    ).replace(/\/index$/, '');
+
+    paths.set(url, realPath);
+    if (url.endsWith('/index.md')) {
+      paths.set(url.replace(/\/index\.md$/, ''), realPath);
+    }
+  }
+
+  return paths;
+}
+
 async function main() {
   process.chdir(projectRoot);
 
-  const files = await recursive('.', [
-    'node_modules',
-    'dist',
-    'bin',
-    'microsite',
-  ]);
+  const files = await recursive('.', ['node_modules', 'dist', 'bin']);
   const mdFiles = files.filter(f => f.endsWith('.md'));
   const badUrls = [];
 
+  const docPages = await findExternalDocsLinks('docs');
+
   for (const mdFile of mdFiles) {
-    const badFileUrls = await verifyFile(mdFile);
+    const badFileUrls = await verifyFile(mdFile, new Set(docPages.values()));
     badUrls.push(...badFileUrls);
   }
 
@@ -111,12 +145,24 @@ async function main() {
         console.error(
           `Unable to reach ${url} from root or microsite/static/, linked from ${basePath}`,
         );
-      } else if (problem === 'not-relative') {
-        console.error('Links to /docs/ must be relative');
+      } else if (problem === 'doc-missing') {
+        const suggestion =
+          docPages.get(url) ||
+          docPages.get(new URL(url, 'http://localhost').pathname);
+        console.error('Links into /docs/ must use an externally reachable ID');
         console.error(`  From: ${basePath}`);
         console.error(`  To: ${url}`);
-      } else if (problem === 'absolute') {
-        console.error(`Link to docs/ should be replaced by a relative URL`);
+        if (suggestion) {
+          console.error(`  Replace With: ${suggestion}`);
+        }
+      } else if (problem === 'not-relative') {
+        console.error('Links within /docs/ must be relative');
+        console.error(`  From: ${basePath}`);
+        console.error(`  To: ${url}`);
+      } else if (problem === 'github') {
+        console.error(
+          `Link to docs/ should not use a GitHub URL, use a relative URL instead`,
+        );
         console.error(`  From: ${basePath}`);
         console.error(`  To: ${url}`);
       }