Capture the Pubmed article type

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

@@ -180,6 +180,16 @@ const getJournal = Article => {
   };
 };
 
+const getPublicationType = Article => {
+  const PublicationTypeList = getElementByName( Article, 'PublicationTypeList' );
+  const PublicationType = getElementsByName( PublicationTypeList, 'PublicationType' );
+  return PublicationType.map( elt => {
+    const UI = getElementAttribute( elt, 'UI' );
+    const value = getElementText( elt );
+    return { UI, value };
+  });
+};
+
 const getArticle = MedlineCitation => {
   // <!ELEMENT	Article (Journal,ArticleTitle,((Pagination, ELocationID*) | ELocationID+),
   //                    Abstract?,AuthorList?, Language+, DataBankList?, GrantList?,
@@ -196,11 +206,18 @@ const getArticle = MedlineCitation => {
   const AuthorList =  getElementByName( Article, 'AuthorList' ) ? getAuthorList( Article ): null;
   const Journal = getJournal( Article );
 
+  // <!ELEMENT	PublicationType (#PCDATA) >
+  // <!ATTLIST	PublicationType
+  // 		    UI CDATA #REQUIRED >
+  // <!ELEMENT	PublicationTypeList (PublicationType+) >
+  const PublicationTypeList = getPublicationType( Article );
+
   return {
     Journal,
     ArticleTitle,
     Abstract,
-    AuthorList
+    AuthorList,
+    PublicationTypeList
   };
 };
 

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

@@ -14,6 +14,10 @@ const ID_TYPE = Object.freeze({
   TITLE: 'title'
 });
 
+const pubTypeToExclude = [
+  { UI: 'D016425', value: 'Published Erratum' }
+];
+
 const paperId2Type = paperId => {
   // 99.3% of CrossRef DOIs (https://www.crossref.org/blog/dois-and-matching-regular-expressions/)
   const doiRegex = /^10\.\d{4,9}\/[-._;()/:A-Z0-9]+$/i;
@@ -44,18 +48,23 @@ const findMatchingPubmedArticle = async ( paperId, IdType, uids ) => {
 
   // Try to match: Could be responsibility of user in future
   const articleMatchesPaperId = article => {
-    const { title, pmid, doi } = getPubmedCitation( article );
+    const { title, pmid, doi, pubTypes } = getPubmedCitation( article );
     let hasMatch = false;
-    switch ( IdType ) {
-      case ID_TYPE.DOI:
-        if( doi === paperId ) hasMatch = true;
-        break;
-      case ID_TYPE.PMID:
-        if( pmid === paperId ) hasMatch = true;
-        break;
-      case ID_TYPE.TITLE:
-        if( santitize( title ).includes( santitize( paperId ) ) ) hasMatch = true;
-        break;
+
+    // Ignore if this article contains an invalid PublicationType entry
+    const isValidType = _.isEmpty( _.intersectionBy( pubTypes, pubTypeToExclude, 'UI' ) );
+    if( isValidType ){
+      switch ( IdType ) {
+        case ID_TYPE.DOI:
+          if( doi === paperId ) hasMatch = true;
+          break;
+        case ID_TYPE.PMID:
+          if( pmid === paperId ) hasMatch = true;
+          break;
+        case ID_TYPE.TITLE:
+          if( santitize( title ).includes( santitize( paperId ) ) ) hasMatch = true;
+          break;
+      }
     }
     return hasMatch;
   };

src/util/pubmed.js

@@ -150,6 +150,7 @@ const getArticleId = ( PubmedArticle, IdType ) => _.get( _.find( _.get( PubmedAr
  * @return {String} result.reference (<ISOAbbreviation> | <Title>) <Year>; <Volume>
  * @return {String} result.pmid
  * @return {String} result.doi
+ * @return {String} result.pubTypes
  */
 const getPubmedCitation = PubmedArticle => {
   const Article = _.get( PubmedArticle, ['MedlineCitation','Article'] ); //required
@@ -162,8 +163,9 @@ const getPubmedCitation = PubmedArticle => {
   const abstract = _.get( Article, 'Abstract', null );
   const pmid = getArticleId( PubmedArticle, 'pubmed' );
   const doi = getArticleId( PubmedArticle, 'doi' );
+  const pubTypes = _.get( Article, 'PublicationTypeList' ); //required
 
-  return { title, authors, reference, abstract, pmid, doi };
+  return { title, authors, reference, abstract, pmid, doi, pubTypes };
 };
 
 /**

test/pubmed/fetchPubmed.js

@@ -226,6 +226,22 @@ describe('fetchPubmed', function(){
 
             }); //AuthorList
 
+            describe( 'PublicationTypeList', () => {
+
+              let PublicationTypeList;
+              before( () => {
+                PublicationTypeList = _.get( Article, 'PublicationTypeList' );
+              });
+
+              it('Should be a list of UI/text objects', () => {
+                expect( PublicationTypeList ).to.be.an.instanceof(Array);
+                PublicationTypeList.forEach( PublicationType => {
+                  expect( PublicationType ).to.have.property('UI');
+                  expect( PublicationType ).to.have.property('value');
+                });
+              });
+            });
+
           }); //Article
 
           describe( 'ChemicalList', () => {