Disallow fetching from Wikidata disambiguation pages

Close #23

src/NodaWikidataFetcher.php

@@ -178,7 +178,20 @@ final class NodaWikidataFetcher {
         if (empty($data['entities'][$wikidata_id])) {
             throw new MDhttpFailedException("Failed fetching from Wikidata. Try again later.");
         }
-        return $data['entities'][$wikidata_id];
+
+        $output = $data['entities'][$wikidata_id];
+
+        // Throw exception if this page is a dedicated disambigation item.-
+        // P31: Instance of; Q4167410: Wikimedia disambiguation page
+        if (isset($output['claims']) && isset($output['claims']['P31'])) {
+            foreach ($output['claims']['P31'] as $is_instance_of) {
+                if (isset($is_instance_of['mainsnak']['datavalue']['value']['id']) && $is_instance_of['mainsnak']['datavalue']['value']['id'] === 'Q4167410') {
+                    throw new NodaWikidataFetcherDisambiguationIsDisallowedException("Loading wikidata disambiguation pages is disallowed");
+                }
+            }
+        }
+
+        return $output;
 
     }
 

src/NodaWikidataFetcherDisambiguationIsDisallowedException.php

@@ -0,0 +1,27 @@
+<?PHP
+/**
+ * This file contains an exception class to be thrown if a user attempts to load
+ * data from a Wikidata item specifically established for a disambiguation page.
+ *
+ * @file
+ *
+ * @author Joshua Ramon Enslin <joshua@museum-digital.de>
+ */
+declare(strict_types = 1);
+
+/**
+ * Exception class to be thrown if a user attempts to load
+ * data from a Wikidata item specifically established for a disambiguation page.
+ */
+final class NodaWikidataFetcherDisambiguationIsDisallowedException extends MDgenericInvalidInputsException {
+    /**
+     * Error message.
+     *
+     * @return string
+     */
+    public function errorMessage() {
+        //error message
+        return 'Attempted to load a disambiguation page. Please select the specific item you want to fetch to enrich the given entry: ' . $this->getMessage();
+
+    }
+}

tests/NodaWikidataFetcherTest.php

@@ -11,6 +11,7 @@ use PHPUnit\Framework\Attributes\Medium;
 use PHPUnit\Framework\Attributes\DataProvider;
 
 require_once __DIR__ . '/../../MDMysqli/test_connections.conf.php';
+require_once __DIR__ . '/../src/NodaWikidataFetcherDisambiguationIsDisallowedException.php';
 
 /**
  * This script contains tests for the Wikidata fetcher.
@@ -37,6 +38,34 @@ final class NodaWikidataFetcherTest extends TestCase {
 
     }
 
+    /**
+     * Data provider providing a Wikidata ID for a dedicated wikidata item for disambiguation pages.
+     *
+     * @return array<string: array{0: string}>
+     */
+    public static function disambiguationPageProvider():array {
+
+        return [
+            'Disambiguation page for "Mochi" - Q6916210' => ['Q6916210'],
+        ];
+
+    }
+
+    /**
+     * Throw error when attempting to load a dedicated wikidata entry for a disambiguation page.
+     *
+     * @param string $wikidata_id Wikidata ID.
+     *
+     * @return void
+     */
+    #[DataProvider('disambiguationPageProvider')]
+    public function testWikidataIdFromLinkFailsForDisambiguationPages(string $wikidata_id):void {
+
+        self::expectException(NodaWikidataFetcherDisambiguationIsDisallowedException::class);
+        NodaWikidataFetcher::getWikidataEntity($wikidata_id);
+
+    }
+
     /**
      * Test to check whether the HTML page is correctly generated.
      *