Merge pull request #84 from stopfstedt/add_registry_number_list_and_registry_number

Updates MeSH parser for 2025 release.

Author: jrjohnson

README.md

@@ -3,7 +3,7 @@
 This PHP code library provides tools for extracting [Medical Subject Headings](https://www.nlm.nih.gov/mesh/meshhome.html) 
 (MeSH) descriptors and associated data from a given XML file into an object representation.
 
-It expects its input to be compliant with the 2023 or 2024 [MeSH DTDs](https://www.nlm.nih.gov/databases/download/mesh.html).
+It expects its input to be compliant with the 2024 or 2025 [MeSH DTDs](https://www.nlm.nih.gov/databases/download/mesh.html).
 
 ## Installation
 

src/Ilios/MeSH/Model/Concept.php

@@ -14,8 +14,6 @@ class Concept extends Reference
 
     protected ?string $casn1Name = null;
 
-    protected ?string $registryNumber = null;
-
     protected ?string $scopeNote = null;
 
     protected ?string $translatorsEnglishScopeNote = null;
@@ -28,6 +26,8 @@ class Concept extends Reference
 
     protected array $terms = [];
 
+    protected array $registryNumbers = [];
+
     public function isPreferred(): bool
     {
         return $this->preferred;
@@ -48,16 +48,6 @@ public function setCasn1Name(?string $casn1Name): void
         $this->casn1Name = $casn1Name;
     }
 
-    public function getRegistryNumber(): ?string
-    {
-        return $this->registryNumber;
-    }
-
-    public function setRegistryNumber(?string $registryNumber): void
-    {
-        $this->registryNumber = $registryNumber;
-    }
-
     public function getScopeNote(): ?string
     {
         return $this->scopeNote;
@@ -117,4 +107,14 @@ public function addTerm(Term $term): void
     {
         $this->terms[] = $term;
     }
+
+    public function getRegistryNumbers(): array
+    {
+        return $this->registryNumbers;
+    }
+
+    public function addRegistryNumber(string $registryNumber): void
+    {
+        $this->registryNumbers[] = $registryNumber;
+    }
 }

src/Ilios/MeSH/Parser.php

@@ -270,7 +270,7 @@ public function parse(string $uri): DescriptorSet
                         break;
                     case self::REGISTRY_NUMBER:
                         $number = $this->getNodeContents($reader);
-                        $currentConcept->setRegistryNumber($number);
+                        $currentConcept->addRegistryNumber($number);
                         break;
                     case self::RELATED_REGISTRY_NUMBER:
                         $number = $this->getNodeContents($reader);

tests/Ilios/MeSH/Model/ConceptTest.php

@@ -44,9 +44,9 @@ public function testGetSetCasn1Name(): void
         $this->basicSetTest($this->object, 'casn1Name', 'string', true);
     }
 
-    public function testGetSetRegistryNumber(): void
+    public function testAddGetRegistryNumbers(): void
     {
-        $this->basicSetTest($this->object, 'registryNumber', 'string', true);
+        $this->addTextToListTest($this->object, 'registryNumber');
     }
 
     public function testGetSetScopeNote(): void

tests/Ilios/MeSH/ParserTest.php

@@ -165,14 +165,19 @@ public function testParse(): void
         $this->assertTrue($concept->isPreferred());
         $this->assertEquals('M0000001', $concept->getUi());
         $this->assertEquals('a casn1 name', $concept->getCasn1Name());
-        $this->assertEquals('00000AAAAA', $concept->getRegistryNumber());
+
+        $registryNumbers = $concept->getRegistryNumbers();
+        $this->assertEquals(2, count($registryNumbers));
+        $this->assertEquals('00000AAAAA', $registryNumbers[0]);
+        $this->assertEquals('11111BBBBB', $registryNumbers[1]);
+
         $this->assertEquals('a scope note', $concept->getScopeNote());
         $this->assertEquals('something in English.', $concept->getTranslatorsEnglishScopeNote());
         $this->assertEquals('i got nothing', $concept->getTranslatorsScopeNote());
 
-        $registryNumbers = $concept->getRelatedRegistryNumbers();
-        $this->assertEquals(1, count($registryNumbers));
-        $this->assertEquals('a related registry number', $registryNumbers[0]);
+        $relatedRegistryNumbers = $concept->getRelatedRegistryNumbers();
+        $this->assertEquals(1, count($relatedRegistryNumbers));
+        $this->assertEquals('a related registry number', $relatedRegistryNumbers[0]);
 
         $relations = $concept->getConceptRelations();
         $this->assertEquals(1, count($relations));
@@ -202,7 +207,7 @@ public function testParse(): void
         $this->assertFalse($concept->isPreferred());
         $this->assertEquals('M0000002', $concept->getUi());
         $this->assertEquals('another concept', $concept->getName());
-        $this->assertEquals('0', $concept->getRegistryNumber());
+        $this->assertCount(0, $concept->getRegistryNumbers());
 
         $relations = $concept->getConceptRelations();
         $this->assertEquals(1, count($relations));

tests/Ilios/MeSH/desc.xml

@@ -3,7 +3,7 @@
 Fake MeSH descriptors set, for testing purposes.
 It contains all possible elements defined by the schema.
 -->
-<!DOCTYPE DescriptorRecordSet SYSTEM "https://www.nlm.nih.gov/databases/dtd/nlmdescriptorrecordset_20240101.dtd">
+<!DOCTYPE DescriptorRecordSet SYSTEM "https://www.nlm.nih.gov/databases/dtd/nlmdescriptorrecordset_20250101.dtd">
 <DescriptorRecordSet LanguageCode="eng">
     <DescriptorRecord DescriptorClass="1">
         <DescriptorUI>D000000</DescriptorUI>
@@ -133,7 +133,10 @@ It contains all possible elements defined by the schema.
                     <String>a concept</String>
                 </ConceptName>
                 <CASN1Name>a casn1 name</CASN1Name>
-                <RegistryNumber>00000AAAAA</RegistryNumber>
+                <RegistryNumberList>
+                    <RegistryNumber>00000AAAAA</RegistryNumber>
+                    <RegistryNumber>11111BBBBB</RegistryNumber>
+                </RegistryNumberList>
                 <ScopeNote>a scope note</ScopeNote>
                 <TranslatorsEnglishScopeNote>something in English.</TranslatorsEnglishScopeNote>
                 <TranslatorsScopeNote>i got nothing</TranslatorsScopeNote>
@@ -170,7 +173,6 @@ It contains all possible elements defined by the schema.
                 <ConceptName>
                     <String>another concept</String>
                 </ConceptName>
-                <RegistryNumber>0</RegistryNumber>
                 <ConceptRelationList>
                     <ConceptRelation RelationName="NRW">
                         <Concept1UI>M0000002</Concept1UI>