build(aio): blacklist unwanted URLs from the generated sitemap.xml (#22061)

Closes #22017

PR Close #22061

aio/tools/transforms/angular-base-package/processors/createSitemap.js

@@ -1,5 +1,15 @@
 module.exports = function createSitemap() {
   return {
+    blacklistedDocTypes: [
+      'navigation-json',
+      'contributors-json',
+      'resources-json',
+    ],
+    blacklistedPaths: [
+      'test',
+      'file-not-found',
+      'overview-dump'
+    ],
     $runAfter: ['paths-computed'],
     $runBefore: ['rendering-docs'],
     $process(docs) {
@@ -8,7 +18,16 @@ module.exports = function createSitemap() {
         path: 'sitemap.xml',
         outputPath: '../sitemap.xml',
         template: 'sitemap.template.xml',
-        urls: docs.filter(doc => doc.outputPath).map(doc => doc.path)
+        urls: docs
+          // Filter out docs that are not outputted
+          .filter(doc => doc.outputPath)
+          // Filter out unwanted docs
+          .filter(doc => this.blacklistedDocTypes.indexOf(doc.docType) === -1)
+          .filter(doc => this.blacklistedPaths.indexOf(doc.path) === -1)
+          // Capture the path of each doc
+          .map(doc => doc.path)
+          // Convert the homepage: `index` to `/`
+          .map(path => path === 'index' ? '' : path)
       });
     }
   };

aio/tools/transforms/angular-base-package/processors/createSitemap.spec.js

@@ -46,6 +46,38 @@ describe('createSitemap processor', () => {
         processor.$process(docs);
         expect(docs.pop().urls).toEqual(['abc', 'fgh']);
       });
+
+      it('ignoring blacklisted doc types', () => {
+        const docs = [
+          { path: 'abc', outputPath: 'abc', docType: 'good' },
+          { path: 'cde', outputPath: 'cde', docType: 'bad' },
+          { path: 'fgh', outputPath: 'fgh', docType: 'good' },
+        ];
+        processor.blacklistedDocTypes = ['bad'];
+        processor.$process(docs);
+        expect(docs.pop().urls).toEqual(['abc', 'fgh']);
+      });
+
+      it('ignoring blacklisted paths', () => {
+        const docs = [
+          { path: 'abc', outputPath: 'abc' },
+          { path: 'cde', outputPath: 'cde' },
+          { path: 'fgh', outputPath: 'fgh' },
+        ];
+        processor.blacklistedPaths = ['cde'];
+        processor.$process(docs);
+        expect(docs.pop().urls).toEqual(['abc', 'fgh']);
+      });
+
+      it('mapping the home page\'s path to `/`', () => {
+        const docs = [
+          { path: 'abc', outputPath: 'abc' },
+          { path: 'index', outputPath: 'index.json' },
+          { path: 'fgh', outputPath: 'fgh' },
+        ];
+        processor.$process(docs);
+        expect(docs.pop().urls).toEqual(['abc', '', 'fgh']);
+      });
     });
   });
 });