Capture PubMed article CommentsCorrectionsList

src/server/routes/api/document/pubmed/fetchPubmed.js

@@ -244,6 +244,42 @@ const getChemicalList = MedlineCitation => {
   return getElementsByName( ChemicalList, 'Chemical' ).map( getChemical );
 };
 
+
+const getCommentsCorrections = CommentsCorrections => {
+  const extractDOI = str => {
+    const doiRegex = /^doi: (?<DOI>10\.\d{4,9}\/[-._;()/:A-Z0-9]+)$/i;
+    const matches = str.match( doiRegex );
+    const DOI = _.get( matches, ['groups', 'DOI'], null );
+    return DOI;
+  };
+// <!ELEMENT	CommentsCorrections (RefSource,PMID?,Note?) >
+// <!ATTLIST	CommentsCorrections
+// 		     RefType (AssociatedDataset | AssociatedPublication |
+// 		             CommentIn | CommentOn |
+// 		             CorrectedandRepublishedIn | CorrectedandRepublishedFrom |
+// 		             ErratumIn | ErratumFor |
+// 		             ExpressionOfConcernIn | ExpressionOfConcernFor |
+// 		             RepublishedIn | RepublishedFrom |
+// 		             RetractedandRepublishedIn | RetractedandRepublishedFrom |
+// 		             RetractionIn | RetractionOf |
+// 		             UpdateIn | UpdateOf |
+// 		             SummaryForPatientsIn |
+// 		             OriginalReportIn |
+// 		             ReprintIn | ReprintOf |
+// 		             Cites)      #REQUIRED    >
+  const RefType = getElementAttribute( CommentsCorrections, 'RefType' );
+  const PMID = getElementByName( CommentsCorrections, 'PMID' ) ? getElementText( getElementByName( CommentsCorrections, 'PMID' ) ): null;
+  const RefSource = getElementText( getElementByName( CommentsCorrections, 'RefSource' ) );
+  const DOI = extractDOI( RefSource );
+  return { RefType, PMID, RefSource, DOI };
+};
+
+// <!ELEMENT	CommentsCorrectionsList (CommentsCorrections+) >
+const getCommentsCorrectionsList = MedlineCitation => {
+  const CommentsCorrectionsList = getElementByName( MedlineCitation, 'CommentsCorrectionsList' );
+  return getElementsByName( CommentsCorrectionsList, 'CommentsCorrections' ).map( getCommentsCorrections );
+};
+
 // <!ELEMENT	KeywordList (Keyword+) >
 // <!ATTLIST	KeywordList
 // 		    Owner (NLM | NLM-AUTO | NASA | PIP | KIE | NOTNLM | HHS) "NLM" >
@@ -313,13 +349,15 @@ const getMedlineCitation = PubmedArticle => {
 
   const Article = getArticle( MedlineCitation );
   const ChemicalList = getElementByName( MedlineCitation, 'ChemicalList' ) ? getChemicalList( MedlineCitation ): [];
+  const CommentsCorrectionsList = getElementByName( MedlineCitation, 'CommentsCorrectionsList' ) ? getCommentsCorrectionsList( MedlineCitation ): [];
   const KeywordList = getElementByName( MedlineCitation, 'KeywordList' ) ? getKeywordList( MedlineCitation ): [];
   const MeshheadingList = getElementByName( MedlineCitation, 'MeshHeadingList' ) ? getMeshheadingList( MedlineCitation ): [];
   const InvestigatorList = getElementByName( MedlineCitation, 'InvestigatorList' ) ? getInvestigatorList( MedlineCitation ): [];
 
   return {
     Article,
     ChemicalList,
+    CommentsCorrectionsList,
     KeywordList,
     MeshheadingList,
     InvestigatorList

src/util/pubmed.js

@@ -238,8 +238,9 @@ const getPubmedCitation = PubmedArticle => {
   const doi = getArticleId( PubmedArticle, 'doi' );
   const pubTypes = _.get( Article, 'PublicationTypeList' ); //required
   const { ISODate } = getPubDate( _.get( Article, ['Journal', 'JournalIssue'] ) );
+  const relations = _.get( PubmedArticle, [ 'MedlineCitation', 'CommentsCorrectionsList'], [] );
 
-  return { title, authors, reference, abstract, pmid, doi, pubTypes, ISODate };
+  return { title, authors, reference, abstract, pmid, doi, pubTypes, ISODate, relations };
 };
 
 /**
@@ -278,6 +279,7 @@ const createPubmedArticle = ({ articleTitle = null, journalName = null, publicat
         PublicationTypeList: []
       },
       ChemicalList: [],
+      CommentsCorrectionsList: [],
       InvestigatorList: [],
       KeywordList: [],
       MeshheadingList: []

test/pubmed/fetchPubmed.js

@@ -5,14 +5,17 @@ import { expect } from 'chai';
 import convert from 'xml-js';
 
 import { pubmedDataConverter } from '../../src/server/routes/api/document/pubmed/fetchPubmed';
+import { isDigits, isDoi } from '../../src/util';
 
 const TEST_PUBMED_DATA = new Map([
   ['151222', 'pmid_151222.xml'], // MedlineDate
   ['9417067', 'pmid_9417067.xml'],
   ['29440426', 'pmid_29440426.xml'], // PubMedData
   ['30078747', 'pmid_30078747.xml'], // InvestigatorList
   ['30115697', 'pmid_30115697.xml'], // Markup; Identifier
-  ['31511694', 'pmid_31511694.xml']
+  ['31511694', 'pmid_31511694.xml'],
+  ['9500320', 'pmid_9500320.xml'], //CommentsCorrections - RetractionIn etc
+  ['38289659', 'pmid_38289659.xml'] //CommentsCorrections RefSource: doi string
 ]);
 
 const xml2jsOpts = {};
@@ -215,8 +218,9 @@ describe('fetchPubmed', function(){
 
                   it('Should have item with top-level attributes', () => {
                     if( !_.isEmpty( Identifier ) ){
-                      expect( Identifier ).to.have.property( 'Identifier[0].Source' );
-                      expect( Identifier ).to.have.property( 'Identifier[0].id' );
+                      let first = _.first( Identifier );
+                      expect( first ).to.have.property( 'Source' );
+                      expect( first ).to.have.property( 'id' );
                     }
                   });
 
@@ -277,6 +281,46 @@ describe('fetchPubmed', function(){
 
           });//ChemicalList
 
+          describe( 'CommentsCorrectionsList', () => {
+
+            let CommentsCorrectionsList;
+            before( () => {
+              CommentsCorrectionsList = _.get( MedlineCitation, 'CommentsCorrectionsList' );
+            });
+
+            it('Should be a list, when exists', () => {
+              if( CommentsCorrectionsList ){  // Optional
+                expect( CommentsCorrectionsList ).to.be.an.instanceof(Array);
+              }
+            });
+
+            it('Should have required attributes, when exists', () => {
+              if( CommentsCorrectionsList ){  // Optional
+                for (const CommentsCorrections of CommentsCorrectionsList) {
+                  expect( CommentsCorrections ).to.have.property( 'RefSource' );
+                  expect( CommentsCorrections ).to.have.property( 'PMID' );
+                  expect( CommentsCorrections ).to.have.property( 'DOI' );
+                  expect( CommentsCorrections ).to.have.property( 'RefType' );
+                }
+              }
+            });
+
+            it('Should have correctly formatted PMID and DOI, when not null', () => {
+              if( CommentsCorrectionsList ){  // Optional
+                for (const CommentsCorrections of CommentsCorrectionsList) {
+                  if( CommentsCorrections.PMID !== null ){
+                    expect( isDigits( CommentsCorrections.PMID ) ).to.be.true;
+                  }
+                  if( CommentsCorrections.DOI !== null ){
+                    expect( isDoi( CommentsCorrections.DOI ) ).to.be.true;
+                  }
+                }
+              }
+            });
+
+          });//CommentsCorrectionsList
+
+
           describe( 'KeywordList', () => {
 
             let KeywordList;