fix(#7472): sitemap should only exclude 404 and 500 pages

withastro · Jul 14, 2023 · 4a518ae · 4a518ae
1 parent 795d598
commit 4a518ae
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 2 deletions.
diff --git a/.changeset/chatty-pigs-film.md b/.changeset/chatty-pigs-film.md
@@ -0,0 +1,5 @@
+---
+'@astrojs/sitemap': minor
+---
+
+Ensure sitemap only excludes numerical pages matching `/404` and `/500` exactly
diff --git a/packages/integrations/sitemap/src/generate-sitemap.ts b/packages/integrations/sitemap/src/generate-sitemap.ts
@@ -2,13 +2,13 @@ import type { EnumChangefreq } from 'sitemap';
 import type { SitemapItem, SitemapOptions } from './index.js';
 import { parseUrl } from './utils/parse-url.js';
 
-const STATUS_CODE_PAGE_REGEXP = /\/[0-9]{3}\/?$/;
+const STATUS_CODE_PAGES = new Set(['/404', '/500']);
 
 /** Construct sitemap.xml given a set of URLs */
 export function generateSitemap(pages: string[], finalSiteUrl: string, opts: SitemapOptions) {
 	const { changefreq, priority, lastmod: lastmodSrc, i18n } = opts!;
 	// TODO: find way to respect <link rel="canonical"> URLs here
-	const urls = [...pages].filter((url) => !STATUS_CODE_PAGE_REGEXP.test(url));
+	const urls = [...pages].filter((url) => !STATUS_CODE_PAGES.has(url.endsWith('/') ? url.slice(0, -1) : url));
 	urls.sort((a, b) => a.localeCompare(b, 'en', { numeric: true })); // sort alphabetically so sitemap is same each time
 
 	const lastmod = lastmodSrc?.toISOString();