From e122a90674e1e36222fd62f46538020d77d2a88f Mon Sep 17 00:00:00 2001
From: Jens Vannerum <jens.vannerum@atmire.com>
Date: Tue, 11 Mar 2025 10:58:08 +0100
Subject: [PATCH] Implement a SEOHealthIndicator which verifies all relevant
 parameters for SEO are ok

(cherry picked from commit 4bd8a24ca75f6d2e6384e850b45c96c4f1229f02)
---
 .../configuration/ActuatorConfiguration.java  |  7 ++
 .../app/rest/health/SEOHealthIndicator.java   | 77 +++++++++++++++++++
 2 files changed, 84 insertions(+)
 create mode 100644 dspace-server-webapp/src/main/java/org/dspace/app/rest/health/SEOHealthIndicator.java

diff --git a/dspace-server-webapp/src/main/java/org/dspace/app/rest/configuration/ActuatorConfiguration.java b/dspace-server-webapp/src/main/java/org/dspace/app/rest/configuration/ActuatorConfiguration.java
index ad78fe2db4..15b9bd9506 100644
--- a/dspace-server-webapp/src/main/java/org/dspace/app/rest/configuration/ActuatorConfiguration.java
+++ b/dspace-server-webapp/src/main/java/org/dspace/app/rest/configuration/ActuatorConfiguration.java
@@ -14,6 +14,7 @@ import java.util.Arrays;
 import org.apache.solr.client.solrj.SolrServerException;
 import org.dspace.app.rest.DiscoverableEndpointsService;
 import org.dspace.app.rest.health.GeoIpHealthIndicator;
+import org.dspace.app.rest.health.SEOHealthIndicator;
 import org.dspace.authority.AuthoritySolrServiceImpl;
 import org.dspace.discovery.SolrSearchCore;
 import org.dspace.statistics.SolrStatisticsCore;
@@ -82,6 +83,12 @@ public class ActuatorConfiguration {
         return new SolrHealthIndicator(solrServerResolver.getServer());
     }
 
+    @Bean
+    @ConditionalOnEnabledHealthIndicator("seo")
+    public SEOHealthIndicator seoHealthIndicator() {
+        return new SEOHealthIndicator();
+    }
+
     @Bean
     @ConditionalOnEnabledHealthIndicator("geoIp")
     public GeoIpHealthIndicator geoIpHealthIndicator() {
diff --git a/dspace-server-webapp/src/main/java/org/dspace/app/rest/health/SEOHealthIndicator.java b/dspace-server-webapp/src/main/java/org/dspace/app/rest/health/SEOHealthIndicator.java
new file mode 100644
index 0000000000..d936fce635
--- /dev/null
+++ b/dspace-server-webapp/src/main/java/org/dspace/app/rest/health/SEOHealthIndicator.java
@@ -0,0 +1,77 @@
+/**
+ * The contents of this file are subject to the license and copyright
+ * detailed in the LICENSE and NOTICE files at the root of the source
+ * tree and available online at
+ *
+ * http://www.dspace.org/license/
+ */
+package org.dspace.app.rest.health;
+
+import org.apache.commons.lang3.StringUtils;
+import org.dspace.services.ConfigurationService;
+import org.dspace.services.factory.DSpaceServicesFactory;
+import org.springframework.boot.actuate.health.AbstractHealthIndicator;
+import org.springframework.boot.actuate.health.Health;
+import org.springframework.web.client.RestTemplate;
+
+/**
+ * Implementation of {@link org.springframework.boot.actuate.health.HealthIndicator} that verifies if the SEO of the
+ * DSpace instance is configured correctly.
+ *
+ * This is only relevant in a production environment, where the DSpace instance is exposed to the public.
+ */
+public class SEOHealthIndicator extends AbstractHealthIndicator {
+
+    ConfigurationService configurationService = DSpaceServicesFactory.getInstance().getConfigurationService();
+
+    private final RestTemplate restTemplate = new RestTemplate();
+
+    @Override
+    protected void doHealthCheck(Health.Builder builder) {
+        String baseUrl = configurationService.getProperty("dspace.ui.url");
+
+        boolean sitemapOk = checkUrl(baseUrl + "/sitemap_index.xml") || checkUrl(baseUrl + "/sitemap_index.html");
+        boolean robotsTxtOk = checkRobotsTxt(baseUrl + "/robots.txt");
+        boolean ssrOk = checkSSR(baseUrl);
+
+        if (sitemapOk && robotsTxtOk && ssrOk) {
+            builder.up()
+                   .withDetail("sitemap", "OK")
+                   .withDetail("robots.txt", "OK")
+                   .withDetail("ssr", "OK");
+        } else {
+            builder.down()
+                   .withDetail("sitemap", sitemapOk ? "OK" : "Missing or inaccessible")
+                   .withDetail("robots.txt", robotsTxtOk ? "OK" : "Empty or contains local URLs")
+                   .withDetail("ssr", ssrOk ? "OK" : "Server-side rendering might be disabled");
+        }
+    }
+
+    private boolean checkUrl(String url) {
+        try {
+            restTemplate.getForEntity(url, String.class);
+            return true;
+        } catch (Exception e) {
+            return false;
+        }
+    }
+
+    private boolean checkRobotsTxt(String url) {
+        try {
+            String content = restTemplate.getForObject(url, String.class);
+            return StringUtils.isNotBlank(content) && !content.contains("localhost");
+        } catch (Exception e) {
+            return false;
+        }
+    }
+
+    private boolean checkSSR(String url) {
+        try {
+            String content = restTemplate.getForObject(url, String.class);
+            return content != null && !content.contains("<ds-app></ds-app>");
+        } catch (Exception e) {
+            return false;
+        }
+    }
+}
+