From e4e271995c8c56c322b9e7134ee8ec0b77a4d8a8 Mon Sep 17 00:00:00 2001
From: Tim Cosgrove <timcosgrove@users.noreply.github.com>
Date: Wed, 11 Dec 2024 13:24:35 -0800
Subject: [PATCH] Remove Event page queries from Content Build (#2368)

* Remove Event page queries.

* Add debug to broken link check.

* Add ignore patterns to broken link checker.

* Fix botched for loop.

* Refine events regex.

* Remove debug logging.
---
 .../constants/brokenLinkIgnorePatterns.js     | 26 +++++++++++++++++++
 .../graphql/CountEntityTypes.graphql.js       | 20 --------------
 .../drupal/graphql/GetAllPages.graphql.js     |  3 +--
 .../stages/build/drupal/individual-queries.js |  8 ------
 .../helpers/isBrokenLink.js                   | 11 +++++++-
 5 files changed, 37 insertions(+), 31 deletions(-)
 create mode 100644 src/site/constants/brokenLinkIgnorePatterns.js

diff --git a/src/site/constants/brokenLinkIgnorePatterns.js b/src/site/constants/brokenLinkIgnorePatterns.js
new file mode 100644
index 0000000000..756386cf65
--- /dev/null
+++ b/src/site/constants/brokenLinkIgnorePatterns.js
@@ -0,0 +1,26 @@
+/*
+  This is a set of patterns which the broken link checker should ignore when making a check.
+
+  Given a page dog.html that contains the following three links:
+    - /animals/cat.html
+    - /animals/horse.html
+    - /animals/capybara.html
+
+  If an item in IGNORE_PATTERNS contains 'capybara', it will ignore the third link.
+  If an item in IGNORE_PATTERNS contains 'animals', it will ignore ALL the links.
+
+  Note that it's the target URL that checks the ignore pattern. If you included
+  'dog' in the IGNORE_PATTERNS in the example above, it would still test dog.html
+  for broken links it contains, but block testing dog.html as a destination.
+
+  Be very careful of unintended consequences when adding these patterns. Be
+  sure you are targeting what you want to ignore precisely.
+
+  */
+const IGNORE_PATTERNS = [
+  /\/events($|\/)?/, // This ignores all links to Event and Event Listing pages.
+];
+
+module.exports = {
+  IGNORE_PATTERNS,
+};
diff --git a/src/site/stages/build/drupal/graphql/CountEntityTypes.graphql.js b/src/site/stages/build/drupal/graphql/CountEntityTypes.graphql.js
index d3807596ba..2848cc0a1f 100644
--- a/src/site/stages/build/drupal/graphql/CountEntityTypes.graphql.js
+++ b/src/site/stages/build/drupal/graphql/CountEntityTypes.graphql.js
@@ -91,26 +91,6 @@ const CountEntityTypes = `
     count
   }
 
-  eventListing: nodeQuery(
-    filter: {
-      conditions: [
-        {field: "status", value: ["1"]},
-        {field: "type", value: ["event_listing"]}
-      ]}
-  	) {
-    count
-  }
-
-  event: nodeQuery(
-    filter: {
-      conditions: [
-        {field: "status", value: ["1"]},
-        {field: "type", value: ["event"]}
-      ]}
-  	) {
-    count
-  }
-
   healthCareRegionDetailPage: nodeQuery(
     filter: {
       conditions: [
diff --git a/src/site/stages/build/drupal/graphql/GetAllPages.graphql.js b/src/site/stages/build/drupal/graphql/GetAllPages.graphql.js
index db16a71a82..4945843eeb 100644
--- a/src/site/stages/build/drupal/graphql/GetAllPages.graphql.js
+++ b/src/site/stages/build/drupal/graphql/GetAllPages.graphql.js
@@ -115,8 +115,7 @@ const buildQuery = () => {
         ... nodeOffice
         ... bioPage
         ... benefitListingPage
-        ... nodeEventListing
-        ... nodeEvent
+
         ... storyListingPage
         ... leadershipListingPage
         ... pressReleasesListingPage
diff --git a/src/site/stages/build/drupal/individual-queries.js b/src/site/stages/build/drupal/individual-queries.js
index 3006c15b28..e8084f0929 100644
--- a/src/site/stages/build/drupal/individual-queries.js
+++ b/src/site/stages/build/drupal/individual-queries.js
@@ -29,12 +29,6 @@ const {
   GetNodePressReleaseListingPages,
 } = require('./graphql/pressReleasesListingPage.graphql');
 
-const {
-  getNodeEventListingQueries,
-} = require('./graphql/nodeEventListing.graphql');
-
-const { getNodeEventQueries } = require('./graphql/nodeEvent.graphql');
-
 const {
   GetNodeStoryListingPages,
 } = require('./graphql/storyListingPage.graphql');
@@ -118,8 +112,6 @@ function getNodeQueries(entityCounts) {
     ...getNewsStoryQueries(entityCounts),
     ...getPressReleaseQueries(entityCounts),
     GetNodePressReleaseListingPages,
-    ...getNodeEventListingQueries(entityCounts),
-    ...getNodeEventQueries(entityCounts),
     ...getVaPoliceQueries(entityCounts),
     GetNodeStoryListingPages,
     GetNodeLocationsListingPages,
diff --git a/src/site/stages/build/plugins/modify-dom/check-broken-links/helpers/isBrokenLink.js b/src/site/stages/build/plugins/modify-dom/check-broken-links/helpers/isBrokenLink.js
index b88204e4e3..753c309841 100644
--- a/src/site/stages/build/plugins/modify-dom/check-broken-links/helpers/isBrokenLink.js
+++ b/src/site/stages/build/plugins/modify-dom/check-broken-links/helpers/isBrokenLink.js
@@ -1,5 +1,8 @@
 const path = require('path');
 const url = require('url');
+const {
+  IGNORE_PATTERNS,
+} = require('../../../../../../constants/brokenLinkIgnorePatterns');
 
 /**
  * Validates an HREF/SRC value
@@ -24,6 +27,13 @@ function isBrokenLink(link, pagePath, allPaths) {
 
   let filePath = decodeURIComponent(parsed.pathname);
 
+  // Check for link destinations we are not testing.
+  for (let i = 0; i < IGNORE_PATTERNS.length; i += 1) {
+    if (filePath.match(IGNORE_PATTERNS[i])) {
+      return false;
+    }
+  }
+
   if (path.isAbsolute(filePath)) {
     filePath = path.join('.', filePath);
   } else {
@@ -33,7 +43,6 @@ function isBrokenLink(link, pagePath, allPaths) {
   if (!path.extname(filePath)) {
     filePath = path.join(filePath, 'index.html');
   }
-
   return !allPaths.has(filePath);
 }