Fix redirect checker (#44417)

The checker assumed that values of the `source` field in each redirect
configuration begin with `/docs/`, but they do not. Remove `/docs/` from
docs site URLs in the target directory before checking them against
configured redirects.

Fix an issue with the first item in the output of the program
always being `undefined` by correctly concatenating the result arrays.

Finally, fix a bug in which the checker would not accept redirects for
docs URLs with fragments.

docs/check-redirects/check-redirects.js

@@ -1,7 +1,7 @@
 const yaml = require('yaml');
 const path = require('path');
 
-const teleportDomain = 'https://goteleport.com';
+const teleportDocsURL = 'https://goteleport.com/docs';
 // RedirectChecker checks for Teleport docs site domains and paths within a
 // given file tree and determines whether a given docs configuration requires
 // redirects.
@@ -38,21 +38,14 @@ class RedirectChecker {
   // URLs that do not correspond to an existing docs page or redirect. It
   // returns a list of problematic URLs.
   check() {
-    const results = this.checkDir(this.otherRepoRoot);
-    let deduped = {};
-    if (results != undefined) {
-      results.forEach(r => {
-        deduped[r] = true;
-      });
-      return Object.keys(deduped);
-    }
+    return this.checkDir(this.otherRepoRoot);
   }
 
   // checkDir recursively checks for docs URLs with missing docs paths or
   // redirects at dirPath. It returns an array of missing URLs.
   checkDir(dirPath) {
     const files = this.fs.readdirSync(dirPath, 'utf8');
-    let result = [];
+    let result = new Set();
     files.forEach(f => {
       for (let e = 0; e < this.exclude.length; e++) {
         if (f.endsWith(this.exclude[e])) {
@@ -62,26 +55,30 @@ class RedirectChecker {
       const fullPath = path.join(dirPath, f);
       const info = this.fs.statSync(fullPath);
       if (!info.isDirectory()) {
-        result = result.concat(this.checkFile(fullPath));
+        for (const r of this.checkFile(fullPath)) {
+          result.add(r);
+        }
         return;
       }
-      result = result.concat(this.checkDir(fullPath));
+      for (const r of this.checkDir(fullPath)) {
+        result.add(r);
+      }
     });
-    return result;
+    return [...result];
   }
 
   // checkFile determines whether docs URLs found in the file
   // match either an actual docs file path or a redirect source.
   // Returns an array of URLs with missing files or redirects.
   checkFile(filePath) {
     const docsPattern = new RegExp(
-      /https:\/\/goteleport.com\/docs\/(ver\/[0-9]+\.x\/)?[\w\/_#-]+/,
+      /https:\/\/goteleport.com\/docs(\/ver\/[0-9]+\.x)?(\/[\w\/_-]+)\/?#?/,
       'gm'
     );
     const text = this.fs.readFileSync(filePath, 'utf8');
     const docsURLs = [...text.matchAll(docsPattern)];
     if (docsURLs.length === 0) {
-      return;
+      return [];
     }
     let result = [];
     docsURLs.forEach(url => {
@@ -92,10 +89,10 @@ class RedirectChecker {
         }) == undefined;
 
       if (!missingEntry) {
-        return;
+        return [];
       }
 
-      let pathPart = url[0].slice(teleportDomain.length);
+      let pathPart = url[2];
       if (pathPart[pathPart.length - 1] != '/') {
         pathPart += '/';
       }
@@ -109,7 +106,7 @@ class RedirectChecker {
 
   urlToDocsPath(url) {
     let nofrag = url.split('#')[0]; // Remove the fragment
-    let rest = nofrag.slice((teleportDomain + '/docs/').length);
+    let rest = nofrag.slice(teleportDocsURL.length);
     if (rest.length == 0) {
       return path.join(this.docsRoot, 'docs', 'pages', 'index.mdx');
     }

docs/check-redirects/check-redirects.test.js

@@ -24,6 +24,8 @@ This is a link to a [documentation page](https://goteleport.com/docs/subdirector
 
 Here is a link to a [missing page](https://goteleport.com/docs/page3).
 
+Here is a link to a [missing page](https://goteleport.com/docs/page3).
+
 `,
     '/docs/content/1.x/docs/pages/page1.mdx': `---
 title: "Sample Page 1"
@@ -71,8 +73,80 @@ title: "Desktop Access Getting Started"
     const fs = createFsFromVolume(vol);
     const checker = new RedirectChecker(fs, '/blog', '/docs/content/1.x', [
       {
-        source: '/docs/page3/',
-        destination: '/docs/page1',
+        source: '/page3/',
+        destination: '/page1/',
+        permanent: true,
+      },
+    ]);
+    const results = checker.check();
+    expect(results).toEqual([]);
+  });
+
+  test(`allows missing docs pages for links with fragments if there is a redirect`, () => {
+    const vol = Volume.fromJSON({
+      '/blog/content2.mdx': `---
+title: "Sample Page 2"
+---
+
+Here is a link to a [missing page](https://goteleport.com/docs/page3/#my-fragment).
+
+`,
+      '/docs/content/1.x/docs/pages/page1.mdx': `---
+title: "Sample Page 1"
+---
+`,
+    });
+    const fs = createFsFromVolume(vol);
+    const checker = new RedirectChecker(fs, '/blog', '/docs/content/1.x', [
+      {
+        source: '/page3/',
+        destination: '/page1/',
+        permanent: true,
+      },
+    ]);
+    const results = checker.check();
+    expect(results).toEqual([]);
+  });
+
+  test(`catches missing redirects for URLs with query strings`, () => {
+    const vol = Volume.fromJSON({
+      '/blog/content2.mdx': `---
+title: "Sample Page 2"
+---
+
+Here is a link to a [missing page](https://goteleport.com/docs/access-controls/device-trust/guide/?scope=enterprise#step-12-register-a-trusted-device").
+
+`,
+      '/docs/content/1.x/docs/pages/page1.mdx': `---
+title: "Sample Page 1"
+---
+`,
+    });
+    const fs = createFsFromVolume(vol);
+    const checker = new RedirectChecker(fs, '/blog', '/docs/content/1.x', []);
+    const results = checker.check();
+    expect(results).toEqual(['https://goteleport.com/docs/access-controls/device-trust/guide/']);
+  });
+
+  test(`allows a docs URL with a query string if there is a redirect`, () => {
+    const vol = Volume.fromJSON({
+      '/blog/content2.mdx': `---
+title: "Sample Page 2"
+---
+
+Here is a link to a [missing page](https://goteleport.com/docs/access-controls/device-trust/guide/?scope=enterprise#step-12-register-a-trusted-device").
+
+`,
+      '/docs/content/1.x/docs/pages/page1.mdx': `---
+title: "Sample Page 1"
+---
+`,
+    });
+    const fs = createFsFromVolume(vol);
+    const checker = new RedirectChecker(fs, '/blog', '/docs/content/1.x', [
+      {
+        source: '/access-controls/device-trust/guide/',
+        destination: '/access-controls/device-trust/',
         permanent: true,
       },
     ]);