From 37e68c24fb4b24db9360086efba56f54ea565191 Mon Sep 17 00:00:00 2001
From: Jim Bumgardner <jbum@jbum.com>
Date: Fri, 9 Aug 2024 16:43:51 -0700
Subject: [PATCH] Fixed sitemap to exclude pages with do-not-crawl tag. Fixed
 such pages to include correct robots directive.

---
 pages/_data/eleventyComputed.js  |  6 ++----
 pages/manual-content/sitemap.njk | 14 +++++++-------
 2 files changed, 9 insertions(+), 11 deletions(-)

diff --git a/pages/_data/eleventyComputed.js b/pages/_data/eleventyComputed.js
index ff378728..b39523ad 100644
--- a/pages/_data/eleventyComputed.js
+++ b/pages/_data/eleventyComputed.js
@@ -182,17 +182,15 @@ module.exports = async () => {
   if (!perfAudits) {
     perfAudits = await fetch(perfApiUrl).then((res) => res.json());
   }
-
   return {
     title: (article) => article?.data?.title || article.title,
     id: (article) => article?.data?.id,
     date: (article) => article?.data?.date || article.date,
-    publishdate: (article) =>
-      article?.data?.date.split("T")[0] || article.publishdate,
+    publishdate: (article) => article?.data?.date.split("T")[0] || article.publishdate,
     meta: (article) => article?.data?.excerpt || article.meta,
     description: (article) => article?.data?.excerpt || article.meta,
     author: (article) => article?.data?.author || article.author,
-    tags: (article) => (isPost(article) ? ["news"] : article.tags),
+    tags: (article) => article?.data?.tags || (isPost(article) ? ["news"] : article.tags),
     permalink: (article) => getPermalink(article),
     layout: (article) => getLayout(article),
     eleventyNavigation: (article) => getNavigation(article),
diff --git a/pages/manual-content/sitemap.njk b/pages/manual-content/sitemap.njk
index 50542911..4270a9b4 100644
--- a/pages/manual-content/sitemap.njk
+++ b/pages/manual-content/sitemap.njk
@@ -5,12 +5,12 @@ eleventyExcludeFromCollections: true
 <?xml version="1.0" encoding="utf-8"?>
 <urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
   {% for page in collections.all %}
-    {% if not sitemap.urls[page.url | url ].exclude %}
-      <url>
-        <loc>https://innovation.ca.gov{{ page.url | url }}</loc>
-        <lastmod>{{ page.date.toISOString() }}</lastmod>
-        <changefreq>daily</changefreq>
-      </url>
-    {% endif %}
+      {% if (page.data.tags and not page.data.tags.includes('do-not-crawl')) and not sitemap.urls[page.url | url ].exclude %}
+        <url>
+          <loc>https://innovation.ca.gov{{ page.url | url }}</loc>
+          <lastmod>{{ page.date.toISOString() }}</lastmod>
+          <changefreq>daily</changefreq>
+        </url>
+      {% endif %}
   {% endfor %}
 </urlset>
\ No newline at end of file