[FINNA-1020] LIDO: index repository locations (#47)

NatLibFi · Oct 17, 2023 · 01b0e5a · 01b0e5a
1 parent a897189
commit 01b0e5a
Show file tree

Hide file tree

Showing 4 changed files with 98 additions and 11 deletions.
diff --git a/conf/recordmanager.ini.sample.finna b/conf/recordmanager.ini.sample.finna
@@ -5,3 +5,9 @@
 ;excluded_file_extensions[] = "mp3"
 ;excluded_file_extensions[] = "css"
 ;excluded_file_extensions[] = "sib"
+
+; Lido only settings
+[LidoRecord]
+; The following section excludes certain location places
+; Compares to: ...namePlaceSet->appellationValue->attributes()->label
+;excluded_location_labels[] = "Torin nurkkaus etäällä"
diff --git a/src/RecordManager/Finna/Record/Lido.php b/src/RecordManager/Finna/Record/Lido.php
@@ -114,6 +114,13 @@ class Lido extends \RecordManager\Base\Record\Lido
      */
     protected $descriptionTypesExcludedFromTitle = ['provenance', 'provenienssi'];
 
+    /**
+     * Location labels which should be excluded when getting location information.
+     *
+     * @var array
+     */
+    protected $excludedLocationLabels;
+
     /**
      * Constructor
      *
@@ -134,6 +141,11 @@ public function __construct(
             $logger,
             $metadataUtils
         );
+        $this->excludedLocationLabels
+            = $this->config['LidoRecord']['excluded_location_labels'] ?? [
+                'tarkempi paikka',
+            ];
+
         $this->initMediaTypeTrait($config);
     }
 
@@ -271,8 +283,10 @@ public function toSolrArray(Database $db = null)
                 $data['hires_image_str_mv'] = $this->source;
             }
         }
-
-        $data['location_geo'] = $this->getEventPlaceLocations();
+        $data['location_geo'] = [
+            ...$this->getEventPlaceLocations(),
+            ...$this->getRepositoryLocations(),
+        ];
         $data['center_coords']
             = $this->metadataUtils->getCenterCoordinates($data['location_geo']);
 
@@ -362,18 +376,36 @@ function ($s) {
             }
         }
 
+        $displayLocations = [];
+        foreach (
+            $this->doc->lido->descriptiveMetadata->objectIdentificationWrap
+            ->repositoryWrap->repositorySet ?? [] as $set
+        ) {
+            if (empty($set->repositoryLocation)) {
+                continue;
+            }
+            if (!empty($set->repositoryLocation->gml)) {
+                return [];
+            }
+            if ($result = $this->getHierarchicalLocations($set->repositoryLocation)) {
+                foreach ($result as $location) {
+                    $displayLocations[] = implode(', ', $location);
+                }
+            }
+        }
         $accepted = [];
-        foreach ($locations as $location) {
-            if (preg_match_all("/[\pL']+/u", trim($location)) === 1) {
-                foreach ($subjectLocations as $subjectLocation) {
-                    if (str_starts_with($subjectLocation, $location)) {
-                        continue 2;
+        foreach ([$locations, $displayLocations] as $results) {
+            foreach ($results as $location) {
+                if (preg_match_all("/[\pL']+/u", trim($location)) === 1) {
+                    foreach ($subjectLocations as $compare) {
+                        if (str_starts_with($compare, $location)) {
+                            continue 2;
+                        }
                     }
                 }
+                $accepted[] = $location;
             }
-            $accepted[] = $location;
         }
-
         return [
             'primary' => $this->processLocations($subjectLocations),
             'secondary' => $this->processLocations($accepted),
@@ -464,6 +496,11 @@ protected function getHierarchicalLocations(\SimpleXMLElement $elem): array
                         // appellationvalues under same parent
                         // If this is not the case, then things are not going ok
                         $currentLabel = trim((string)$elemValue->attributes()->label);
+                        if (in_array(strtolower($currentLabel), $this->excludedLocationLabels)) {
+                            // Locations with labels like "Torin kulman takaosa" should be skipped
+                            continue;
+                        }
+
                         if ($label && $label !== $currentLabel) {
                             // There seems to be different types of appellationValues so skip the new ones
                             continue;
@@ -506,6 +543,29 @@ protected function getHierarchicalLocations(\SimpleXMLElement $elem): array
         return $results;
     }
 
+    /**
+     * Get repository locations
+     *
+     * @return array<int, string>
+     */
+    protected function getRepositoryLocations(): array
+    {
+        $results = [];
+        foreach (
+            $this->doc->lido->descriptiveMetadata->objectIdentificationWrap
+            ->repositoryWrap->repositorySet ?? [] as $set
+        ) {
+            if (empty($set->repositoryLocation->gml)) {
+                continue;
+            }
+            if ($wkt = $this->convertGmlToWkt($set->repositoryLocation->gml)) {
+                $results[] = $wkt;
+            }
+        }
+
+        return $results;
+    }
+
     /**
      * Split a location found from displayPlace element. Excludes all values
      * found after splitting which has redundancy or has only a single value.
@@ -594,9 +654,8 @@ function ($s) {
             },
             $result
         );
-        $result = array_unique($result);
 
-        return $result;
+        return array_values(array_unique($result));
     }
 
     /**

diff --git a/tests/RecordManagerTest/Finna/Record/LidoTest.php b/tests/RecordManagerTest/Finna/Record/LidoTest.php
@@ -557,6 +557,7 @@ public function testHierarchicalLocations(): void
                 'Håkansbölen kartano, Hakunila, Vantaa, Suomi',
                 'Vaasa',
                 'Ristimäenkatu 5, Mikkeli, Etelä-Savo, Suomi',
+                'Ahmatie 1, Helsinki',
             ],
         ];
         $this->compareArray($expected, $result, 'Locations');

diff --git a/tests/fixtures/Finna/record/lido_locations.xml b/tests/fixtures/Finna/record/lido_locations.xml
@@ -3,6 +3,27 @@
   <lido>
     <lidoRecID type="ITEM">M123</lidoRecID>
     <descriptiveMetadata>
+      <objectIdentificationWrap>
+        <repositoryWrap>
+          <repositorySet type="Current location">
+            <repositoryLocation>
+              <namePlaceSet>
+                <appellationValue label="tarkempi paikka">Should not display</appellationValue>
+              </namePlaceSet>
+              <partOfPlace>
+                <namePlaceSet>
+                  <appellationValue label="katuosoite">Ahmatie 1</appellationValue>
+                </namePlaceSet>
+                <partOfPlace>
+                  <namePlaceSet>
+                    <appellationValue label="kunta/kaupunki">Helsinki</appellationValue>
+                  </namePlaceSet>
+                </partOfPlace>
+              </partOfPlace>
+            </repositoryLocation>
+          </repositorySet>
+        </repositoryWrap>
+      </objectIdentificationWrap>
       <objectRelationWrap>
         <subjectWrap>
           <subjectSet>