From 68014633250e965a66748cad0ef7b7523892c96a Mon Sep 17 00:00:00 2001
From: mshroom <32199029+mshroom@users.noreply.github.com>
Date: Mon, 10 Jun 2024 14:05:35 +0300
Subject: [PATCH] [FINNA-2110] QDC: Add option to set preferred dc.type fields
 when indexing format (#158)

---
 src/RecordManager/Base/Record/Qdc.php         | 32 ++++++++++++++++-
 .../RecordManagerTest/Base/Record/QdcTest.php | 35 +++++++++++++++++++
 tests/fixtures/Base/record/qdc1.xml           |  3 ++
 3 files changed, 69 insertions(+), 1 deletion(-)

diff --git a/src/RecordManager/Base/Record/Qdc.php b/src/RecordManager/Base/Record/Qdc.php
index f56e308a7..6a4d00cee 100644
--- a/src/RecordManager/Base/Record/Qdc.php
+++ b/src/RecordManager/Base/Record/Qdc.php
@@ -34,6 +34,8 @@
 use RecordManager\Base\Utils\Logger;
 use RecordManager\Base\Utils\MetadataUtils;
 
+use function in_array;
+
 /**
  * Qdc record class
  *
@@ -73,6 +75,13 @@ class Qdc extends AbstractRecord
      */
     protected $recordNs = 'http://www.openarchives.org/OAI/2.0/oai_dc/';
 
+    /**
+     * Type fields which should be excluded when defining format.
+     *
+     * @var array
+     */
+    protected $excludedFormatTypes = [];
+
     /**
      * Constructor
      *
@@ -328,7 +337,28 @@ public function getSeriesNumbering()
      */
     public function getFormat()
     {
-        return $this->doc->type ? trim((string)$this->doc->type) : 'Unknown';
+        $param = $this->getDriverParam('preferredFormatTypes', '');
+        $preferredTypes = $param ? explode(',', $param) : [];
+        $collectedTypes = [];
+        $first = '';
+        foreach ($this->doc->type ?? [] as $node) {
+            if ($value = trim((string)$node)) {
+                $typeAttr = trim((string)($node->attributes()->type ?? '')) ?: 'no_type';
+                if (!in_array($typeAttr, $this->excludedFormatTypes) && !($collectedTypes[$typeAttr] ?? '')) {
+                    $collectedTypes[$typeAttr] = $value;
+                    $first = $first ?: $typeAttr;
+                }
+            }
+        }
+        if ($collectedTypes) {
+            foreach ($preferredTypes as $pref) {
+                if ($collectedTypes[$pref] ?? '') {
+                    return $collectedTypes[$pref];
+                }
+            }
+            return $collectedTypes[$first];
+        }
+        return 'Unknown';
     }
 
     /**
diff --git a/tests/RecordManagerTest/Base/Record/QdcTest.php b/tests/RecordManagerTest/Base/Record/QdcTest.php
index 07b2390ae..b8be685e0 100644
--- a/tests/RecordManagerTest/Base/Record/QdcTest.php
+++ b/tests/RecordManagerTest/Base/Record/QdcTest.php
@@ -80,6 +80,9 @@ public function testQdc1()
                 '2021-06-16T06:31:44Z',
                 '2021',
                 'Article',
+                'okm_type',
+                'okm_type_2',
+                'other_type',
                 'Eeva-Liisa Viskari, Suvi Lehtoranta, Riikka Malila. Urine : The'
                     . ' potential, value chain and its sustainable management. '
                     . 'Sanitation Value Chain (2021) 5, 1, pages 10-12. '
@@ -192,4 +195,36 @@ public function testQdc1()
 
         $this->compareArray($expected, $keys, 'getWorkIdentificationData');
     }
+
+    /**
+     * Test format
+     *
+     * @return void
+     */
+    public function testFormat()
+    {
+        $expected = [
+            'okm' => 'okm_type',
+            'okm,other' => 'okm_type',
+            'finna,other' => 'other_type',
+            'finna' => 'Article',
+        ];
+        foreach ($expected as $preferredTypes => $format) {
+            $record = $this->createRecord(
+                Qdc::class,
+                'qdc1.xml',
+                [
+                    '__unit_test_no_source__' => [
+                        'driverParams' => [
+                            "preferredFormatTypes=$preferredTypes",
+                        ],
+                    ],
+                ],
+                'Base',
+                [$this->createMock(\RecordManager\Base\Http\ClientManager::class)]
+            );
+            $fields = $record->toSolrArray();
+            $this->assertEquals($format, $fields['format']);
+        }
+    }
 }
diff --git a/tests/fixtures/Base/record/qdc1.xml b/tests/fixtures/Base/record/qdc1.xml
index 95de39764..095232f16 100644
--- a/tests/fixtures/Base/record/qdc1.xml
+++ b/tests/fixtures/Base/record/qdc1.xml
@@ -17,6 +17,9 @@
   <available>2021-06-16T06:31:44Z</available>
   <issued>2021</issued>
   <type lang="fi">Article</type>
+  <type lang="fi" type="okm">okm_type</type>
+  <type lang="fi" type="okm">okm_type_2</type>
+  <type lang="fi" type="other">other_type</type>
   <identifier type="citation">Eeva-Liisa Viskari, Suvi Lehtoranta, Riikka Malila. Urine : The potential, value chain and its sustainable management. Sanitation Value Chain (2021) 5, 1, pages 10-12. https://doi.org/10.34416/svc.00029</identifier>
   <identifier type="issn">2432-5058</identifier>
   <identifier type="uri">http://hdl.handle.net/10138/331330</identifier>