Add sitemap filter option (#2755)

* Add filter for sitemap

* chore: add changeset

* fix: types

* docs: add sitemapFilter docs

Co-authored-by: David Rothmann <david@rothmann.it>

.changeset/cool-ravens-occur.md

@@ -0,0 +1,18 @@
+---
+'astro': patch
+---
+
+Add user-configurable `sitemapFilter` option.
+
+This option can be used to ensure certain pages are excluded from your final sitemap.
+
+```ts
+// astro.config.ts
+import type { AstroUserConfig } from 'astro'
+
+const config: AstroUserConfig = {
+  sitemap: true,
+  sitemapFilter: (page: string) => !page.includes('secret-page')
+}
+export default config
+```

packages/astro/src/@types/astro.ts

@@ -226,6 +226,25 @@ export interface AstroUserConfig {
 		 */
 		sitemap?: boolean;
 
+		/**
+		 * @docs
+		 * @name buildOptions.sitemapFilter
+		 * @type {undefined|((page: string) => boolean)}
+		 * @default `undefined`
+		 * @description
+		 * Customize sitemap generation for your build by excluding certain pages.
+		 *
+		 * ```js
+		 * {
+		 *   buildOptions: {
+		 * 	   sitemap: true
+		 * 	   sitemapFilter: (page) => !page.includes('secret-page')
+		 *   }
+		 * }
+		 * ```
+		 */
+		sitemapFilter?: (page: string) => boolean
+
 		/**
 		 * @docs
 		 * @name buildOptions.pageUrlFormat

packages/astro/src/core/build/index.ts

@@ -149,7 +149,8 @@ class AstroBuilder {
 		// Build your final sitemap.
 		if (this.config.buildOptions.sitemap && this.config.buildOptions.site) {
 			timer.sitemapStart = performance.now();
-			const sitemap = generateSitemap(pageNames.map((pageName) => new URL(pageName, this.config.buildOptions.site).href));
+			const sitemapFilter = this.config.buildOptions.sitemapFilter ? (this.config.buildOptions.sitemapFilter as (page: string) => boolean) : undefined;
+			const sitemap = generateSitemap(pageNames.map((pageName) => new URL(pageName, this.config.buildOptions.site).href), sitemapFilter);
 			const sitemapPath = new URL('./sitemap.xml', this.config.dist);
 			await fs.promises.mkdir(new URL('./', sitemapPath), { recursive: true });
 			await fs.promises.writeFile(sitemapPath, sitemap, 'utf8');

packages/astro/src/core/config.ts

@@ -50,6 +50,7 @@ export const AstroConfigSchema = z.object({
 				.string()
 				.optional()
 				.transform((val) => (val ? addTrailingSlash(val) : val)),
+			sitemapFilter: z.function().optional(),
 			sitemap: z.boolean().optional().default(true),
 			pageUrlFormat: z
 				.union([z.literal('file'), z.literal('directory')])

packages/astro/src/core/render/sitemap.ts

@@ -1,14 +1,19 @@
+
 const STATUS_CODE_PAGE_REGEXP = /\/[0-9]{3}\/?$/;
 
 /** Construct sitemap.xml given a set of URLs */
-export function generateSitemap(pages: string[]): string {
+export function generateSitemap(pages: string[], filter?: (page: string) => boolean): string {
 	// TODO: find way to respect <link rel="canonical"> URLs here
-	// TODO: find way to exclude pages from sitemap
 
 	// copy just in case original copy is needed
 	// make sure that 404 page is excluded
 	// also works for other error pages
-	const urls = [...pages].filter((url) => !STATUS_CODE_PAGE_REGEXP.test(url));
+	let urls = [...pages].filter((url) => !STATUS_CODE_PAGE_REGEXP.test(url));
+
+	if (filter) {
+		urls = urls.filter(url => filter(url));
+	}
+
 	urls.sort((a, b) => a.localeCompare(b, 'en', { numeric: true })); // sort alphabetically so sitemap is same each time
 	let sitemap = `<?xml version="1.0" encoding="UTF-8"?><urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">`;
 	for (const url of urls) {