From 2e157f0cbbe253206fb75371c7c0831f818e7e55 Mon Sep 17 00:00:00 2001
From: j3nsch <j3nsch@gmail.com>
Date: Wed, 22 Nov 2023 09:04:18 +0100
Subject: [PATCH] #338 Functions for finding (duplicate) DOIs

---
 library/Opus/Doi/DoiManager.php   | 39 +++++++++++++
 tests/Opus/Doi/DoiManagerTest.php | 94 +++++++++++++++++++++++++++++++
 2 files changed, 133 insertions(+)

diff --git a/library/Opus/Doi/DoiManager.php b/library/Opus/Doi/DoiManager.php
index 56a05fba..1320c145 100644
--- a/library/Opus/Doi/DoiManager.php
+++ b/library/Opus/Doi/DoiManager.php
@@ -42,6 +42,8 @@
 use Opus\Common\Log;
 use Opus\Common\Log\LogService;
 use Opus\Common\Model\NotFoundException;
+use Opus\Db\DocumentIdentifiers;
+use Opus\Db\TableGateway;
 use Opus\Document;
 use Opus\DocumentFinder;
 use Opus\Doi\Generator\DoiGeneratorException;
@@ -851,4 +853,41 @@ public function storeRegistrationXml($doc, $xml)
 
         file_put_contents($filePath, $xml);
     }
+
+    /**
+     * @return string[]
+     */
+    public function getAllDoiValues()
+    {
+        $table = TableGateway::getInstance(DocumentIdentifiers::class);
+
+        $database = $table->getAdapter();
+
+        $select = $table->select()
+            ->from('document_identifiers', 'value')
+            ->distinct(true)
+            ->where('type = ?', 'doi');
+
+        return $database->fetchCol($select);
+    }
+
+    /**
+     * Returns DOIs that are linked to multiple documents.
+     *
+     * @return string[]
+     */
+    public function getDuplicateDoiValues()
+    {
+        $table = TableGateway::getInstance(DocumentIdentifiers::class);
+
+        $database = $table->getAdapter();
+
+        $select = $table->select()
+            ->from('document_identifiers', 'value')
+            ->group('value')
+            ->having('count(value) > 1')
+            ->where('type = ?', 'doi');
+
+        return $database->fetchCol($select);
+    }
 }
diff --git a/tests/Opus/Doi/DoiManagerTest.php b/tests/Opus/Doi/DoiManagerTest.php
index d04f05ea..3ca5a5b0 100644
--- a/tests/Opus/Doi/DoiManagerTest.php
+++ b/tests/Opus/Doi/DoiManagerTest.php
@@ -908,4 +908,98 @@ public function testGetLandingPageUrlOfDocForId()
             $manager->getLandingPageUrlOfDoc($docId)
         );
     }
+
+    public function testGetAllDoiValues()
+    {
+        $doiValues = [
+            '10.0000/1111',
+            '10.0000/2222',
+            '10.0000/3333',
+        ];
+
+        foreach ($doiValues as $doi) {
+            $doc        = Document::new();
+            $identifier = $doc->addIdentifier();
+            $identifier->setType('doi');
+            $identifier->setValue($doi);
+            $doc->store();
+        }
+
+        // other identifier type than DOI
+        $identifier = $doc->addIdentifier();
+        $identifier->setType('isbn');
+        $identifier->setValue('isbn-value');
+
+        // duplicate DOI identifier
+        $identifier = $doc->addIdentifier();
+        $identifier->setType('doi');
+        $identifier->setValue('10.0000/3333');
+
+        $doc->store();
+
+        $manager = new DoiManager();
+
+        $values = $manager->getAllDoiValues();
+
+        $this->assertCount(3, $values);
+        $this->assertEquals($doiValues, $values);
+    }
+
+    public function testGetAllDoiValuesNonFound()
+    {
+        $manager = new DoiManager();
+
+        $values = $manager->getAllDoiValues();
+
+        $this->assertIsArray($values);
+        $this->assertEmpty($values);
+    }
+
+    public function testGetDuplicateDoiValues()
+    {
+        $doi1 = '10.1000/1111';
+        $doi2 = '10.1000/2222';
+
+        $doc = Document::new();
+        $doi = $doc->addIdentifier();
+        $doi->setType('doi');
+        $doi->setValue($doi1);
+        $doc->store();
+
+        $doc = Document::new();
+        $doi = $doc->addIdentifier();
+        $doi->setType('doi');
+        $doi->setValue($doi1);
+        $doc->store();
+
+        $doc = Document::new();
+        $doi = $doc->addIdentifier();
+        $doi->setType('doi');
+        $doi->setValue($doi2);
+        $doc->store();
+
+        // other identifier type than DOI
+        $doc        = Document::new();
+        $identifier = $doc->addIdentifier();
+        $identifier->setType('isbn');
+        $identifier->setValue('isbn-value');
+        $doc->store();
+
+        $manager = new DoiManager();
+
+        $values = $manager->getDuplicateDoiValues();
+
+        $this->assertCount(1, $values);
+        $this->assertEquals($doi1, $values[0]);
+    }
+
+    public function testGetDuplicateDoiValuesNonFound()
+    {
+        $manager = new DoiManager();
+
+        $values = $manager->getDuplicateDoiValues();
+
+        $this->assertIsArray($values);
+        $this->assertEmpty($values);
+    }
 }