feat(sitemap): add ignorePatterns option (#6979)

Co-authored-by: Joshua Chen <sidachen2003@gmail.com>

packages/docusaurus-plugin-sitemap/src/__tests__/createSitemap.test.ts

@@ -19,6 +19,7 @@ describe('createSitemap', () => {
       {
         changefreq: EnumChangefreq.DAILY,
         priority: 0.7,
+        ignorePatterns: [],
       },
     );
     expect(sitemap).toContain(
@@ -42,11 +43,34 @@ describe('createSitemap', () => {
       {
         changefreq: EnumChangefreq.DAILY,
         priority: 0.7,
+        ignorePatterns: [],
       },
     );
     expect(sitemap).not.toContain('404');
   });
 
+  it('excludes patterns configured to be ignored', async () => {
+    const sitemap = await createSitemap(
+      {
+        url: 'https://example.com',
+      } as DocusaurusConfig,
+      ['/', '/search/', '/tags/', '/search/foo', '/tags/foo/bar'],
+      {
+        changefreq: EnumChangefreq.DAILY,
+        priority: 0.7,
+        ignorePatterns: [
+          // Shallow ignore
+          '/search/',
+          // Deep ignore
+          '/tags/**',
+        ],
+      },
+    );
+    expect(sitemap).not.toContain('/search/</loc>');
+    expect(sitemap).toContain('/search/foo');
+    expect(sitemap).not.toContain('/tags');
+  });
+
   it('keep trailing slash unchanged', async () => {
     const sitemap = await createSitemap(
       {
@@ -57,6 +81,7 @@ describe('createSitemap', () => {
       {
         changefreq: EnumChangefreq.DAILY,
         priority: 0.7,
+        ignorePatterns: [],
       },
     );
 
@@ -76,6 +101,7 @@ describe('createSitemap', () => {
       {
         changefreq: EnumChangefreq.DAILY,
         priority: 0.7,
+        ignorePatterns: [],
       },
     );
 
@@ -95,6 +121,7 @@ describe('createSitemap', () => {
       {
         changefreq: EnumChangefreq.DAILY,
         priority: 0.7,
+        ignorePatterns: [],
       },
     );
 

packages/docusaurus-plugin-sitemap/src/__tests__/options.test.ts

@@ -27,6 +27,7 @@ describe('validateOptions', () => {
     const userOptions = {
       changefreq: 'yearly',
       priority: 0.9,
+      ignorePatterns: ['/search/**'],
     };
     expect(testValidate(userOptions)).toEqual({
       ...defaultOptions,
@@ -49,4 +50,17 @@ describe('validateOptions', () => {
       `"\\"changefreq\\" must be one of [daily, monthly, always, hourly, weekly, yearly, never]"`,
     );
   });
+
+  it('rejects bad ignorePatterns inputs', () => {
+    expect(() =>
+      testValidate({ignorePatterns: '/search'}),
+    ).toThrowErrorMatchingInlineSnapshot(
+      `"\\"ignorePatterns\\" must be an array"`,
+    );
+    expect(() =>
+      testValidate({ignorePatterns: [/^\/search/]}),
+    ).toThrowErrorMatchingInlineSnapshot(
+      `"\\"ignorePatterns[0]\\" must be a string"`,
+    );
+  });
 });

packages/docusaurus-plugin-sitemap/src/createSitemap.ts

@@ -6,25 +6,28 @@
  */
 
 import {SitemapStream, streamToPromise} from 'sitemap';
-import type {Options} from '@docusaurus/plugin-sitemap';
+import type {PluginOptions} from '@docusaurus/plugin-sitemap';
 import type {DocusaurusConfig} from '@docusaurus/types';
 import {applyTrailingSlash} from '@docusaurus/utils-common';
+import {createMatcher} from '@docusaurus/utils';
 
 export default async function createSitemap(
   siteConfig: DocusaurusConfig,
   routesPaths: string[],
-  options: Options,
+  options: PluginOptions,
 ): Promise<string> {
   const {url: hostname} = siteConfig;
   if (!hostname) {
     throw new Error('URL in docusaurus.config.js cannot be empty/undefined.');
   }
-  const {changefreq, priority} = options;
+  const {changefreq, priority, ignorePatterns} = options;
+
+  const ignoreMatcher = createMatcher(ignorePatterns);
 
   const sitemapStream = new SitemapStream({hostname});
 
   routesPaths
-    .filter((route) => !route.endsWith('404.html'))
+    .filter((route) => !route.endsWith('404.html') && !ignoreMatcher(route))
     .forEach((routePath) =>
       sitemapStream.write({
         url: applyTrailingSlash(routePath, {

packages/docusaurus-plugin-sitemap/src/index.ts

@@ -7,13 +7,13 @@
 
 import fs from 'fs-extra';
 import path from 'path';
-import type {Options} from '@docusaurus/plugin-sitemap';
+import type {PluginOptions} from '@docusaurus/plugin-sitemap';
 import createSitemap from './createSitemap';
 import type {LoadContext, Plugin} from '@docusaurus/types';
 
 export default function pluginSitemap(
   context: LoadContext,
-  options: Options,
+  options: PluginOptions,
 ): Plugin<void> {
   return {
     name: 'docusaurus-plugin-sitemap',

packages/docusaurus-plugin-sitemap/src/options.ts

@@ -7,12 +7,13 @@
 
 import {Joi} from '@docusaurus/utils-validation';
 import {EnumChangefreq} from 'sitemap';
-import type {Options} from '@docusaurus/plugin-sitemap';
+import type {Options, PluginOptions} from '@docusaurus/plugin-sitemap';
 import type {OptionValidationContext} from '@docusaurus/types';
 
-export const DEFAULT_OPTIONS: Options = {
+export const DEFAULT_OPTIONS: PluginOptions = {
   changefreq: EnumChangefreq.WEEKLY,
   priority: 0.5,
+  ignorePatterns: [],
 };
 
 const PluginOptionSchema = Joi.object({
@@ -24,6 +25,9 @@ const PluginOptionSchema = Joi.object({
     .valid(...Object.values(EnumChangefreq))
     .default(DEFAULT_OPTIONS.changefreq),
   priority: Joi.number().min(0).max(1).default(DEFAULT_OPTIONS.priority),
+  ignorePatterns: Joi.array()
+    .items(Joi.string())
+    .default(DEFAULT_OPTIONS.ignorePatterns),
   trailingSlash: Joi.forbidden().messages({
     'any.unknown':
       'Please use the new Docusaurus global trailingSlash config instead, and the sitemaps plugin will use it.',
@@ -33,7 +37,7 @@ const PluginOptionSchema = Joi.object({
 export function validateOptions({
   validate,
   options,
-}: OptionValidationContext<Options, Options>): Options {
+}: OptionValidationContext<Options, PluginOptions>): PluginOptions {
   const validatedOptions = validate(PluginOptionSchema, options);
   return validatedOptions;
 }

packages/docusaurus-plugin-sitemap/src/plugin-sitemap.d.ts

@@ -7,10 +7,16 @@
 
 import type {EnumChangefreq} from 'sitemap';
 
-export type Options = {
-  id?: string;
+export type PluginOptions = {
   /** @see https://www.sitemaps.org/protocol.html#xmlTagDefinitions */
-  changefreq?: EnumChangefreq;
+  changefreq: EnumChangefreq;
   /** @see https://www.sitemaps.org/protocol.html#xmlTagDefinitions */
-  priority?: number;
+  priority: number;
+  /**
+   * A list of glob patterns; matching route paths will be filtered from the
+   * sitemap. Note that you may need to include the base URL in here.
+   */
+  ignorePatterns: string[];
 };
+
+export type Options = Partial<PluginOptions>;