From 3ec7203a905c2207bbac751e69a36859869d31b1 Mon Sep 17 00:00:00 2001
From: Shadab Ahmad <sahmad@ebi.ac.uk>
Date: Fri, 15 Nov 2024 15:46:56 +0000
Subject: [PATCH] add common taxon id as a return field for uniparc light

---
 .../tsv/uniparc/UniParcEntryLightValueMapper.java  | 14 +++++++++++++-
 .../uniparc/UniParcEntryLightValueMapperTest.java  | 11 ++++++++---
 2 files changed, 21 insertions(+), 4 deletions(-)

diff --git a/core-parser/src/main/java/org/uniprot/core/parser/tsv/uniparc/UniParcEntryLightValueMapper.java b/core-parser/src/main/java/org/uniprot/core/parser/tsv/uniparc/UniParcEntryLightValueMapper.java
index c149912a9..2a9bd213f 100644
--- a/core-parser/src/main/java/org/uniprot/core/parser/tsv/uniparc/UniParcEntryLightValueMapper.java
+++ b/core-parser/src/main/java/org/uniprot/core/parser/tsv/uniparc/UniParcEntryLightValueMapper.java
@@ -21,7 +21,8 @@ public class UniParcEntryLightValueMapper implements EntityValueMapper<UniParcEn
                     "accession",
                     "first_seen",
                     "last_seen",
-                    "common_taxons");
+                    "common_taxons",
+                    "common_taxon_ids");
     private static final String DELIMITER2 = "; ";
 
     @Override
@@ -102,6 +103,13 @@ private Map<String, String> getSimpleAttributeValues(UniParcEntryLight entry, Li
                                     .map(this::getCommonTaxonString)
                                     .orElse(""));
                     break;
+                case "common_taxon_ids":
+                    map.put(
+                            UNIPARC_FIELDS.get(10),
+                            Optional.of(entry.getCommonTaxons())
+                                    .map(this::getCommonTaxonIdString)
+                                    .orElse(""));
+                    break;
                 default:
                     // do nothing
             }
@@ -112,4 +120,8 @@ private Map<String, String> getSimpleAttributeValues(UniParcEntryLight entry, Li
     private String getCommonTaxonString(List<CommonOrganism> commonTaxons) {
         return commonTaxons.stream().map(CommonOrganism::getCommonTaxon).collect(Collectors.joining("; "));
     }
+
+    private String getCommonTaxonIdString(List<CommonOrganism> commonTaxons) {
+        return commonTaxons.stream().map(commonOrganism -> String.valueOf(commonOrganism.getCommonTaxonId())).collect(Collectors.joining("; "));
+    }
 }
diff --git a/core-parser/src/test/java/org/uniprot/core/parser/tsv/uniparc/UniParcEntryLightValueMapperTest.java b/core-parser/src/test/java/org/uniprot/core/parser/tsv/uniparc/UniParcEntryLightValueMapperTest.java
index e613c83b0..41bd36fc0 100644
--- a/core-parser/src/test/java/org/uniprot/core/parser/tsv/uniparc/UniParcEntryLightValueMapperTest.java
+++ b/core-parser/src/test/java/org/uniprot/core/parser/tsv/uniparc/UniParcEntryLightValueMapperTest.java
@@ -18,7 +18,7 @@ class UniParcEntryLightValueMapperTest {
     @Test
     void testGetDataOrganism() {
         UniParcEntryLight entry = create();
-        List<String> fields = Arrays.asList("upi", "organism", "organism_id", "proteome", "common_taxons");
+        List<String> fields = Arrays.asList("upi", "organism", "organism_id", "proteome", "common_taxons", "common_taxon_ids");
 
         Map<String, String> result = new UniParcEntryLightValueMapper().mapEntity(entry, fields);
 
@@ -28,13 +28,14 @@ void testGetDataOrganism() {
         verify("9606; 10090", "organism_id", result);
         verify("UP000005640:C1; UP000002494:C2", "proteome", result);
         verify("Bacteroides; Enterococcus", "common_taxons", result);
+        verify("12345; 9876", "common_taxon_ids", result);
     }
 
     @Test
     void testGetDataOrganism_withEmptyCommonTaxons() {
         UniParcEntryLight entry = create();
         UniParcEntryLight entryWithEmptyCommonTaxons = UniParcEntryLightBuilder.from(entry).commonTaxonsSet(List.of()).build();
-        List<String> fields = Arrays.asList("upi", "organism", "organism_id", "proteome", "common_taxons");
+        List<String> fields = Arrays.asList("upi", "organism", "organism_id", "proteome", "common_taxons", "common_taxon_ids");
 
         Map<String, String> result = new UniParcEntryLightValueMapper().mapEntity(entryWithEmptyCommonTaxons, fields);
 
@@ -44,6 +45,8 @@ void testGetDataOrganism_withEmptyCommonTaxons() {
         verify("9606; 10090", "organism_id", result);
         verify("UP000005640:C1; UP000002494:C2", "proteome", result);
         verify("", "common_taxons", result);
+        verify("", "common_taxons", result);
+        verify("", "common_taxon_ids", result);
     }
 
     @Test
@@ -94,6 +97,8 @@ void testGetSequenceFeature() {
         verify("sigId2", "PROSITE", result);
     }
 
+
+
     private void verify(String expected, String field, Map<String, String> result) {
         assertEquals(expected, result.get(field));
     }
@@ -109,7 +114,7 @@ private UniParcEntryLight create() {
 
         LinkedHashSet<String> proteinNames = new LinkedHashSet<>(List.of("protein1", "protein2"));
         LinkedHashSet<String> geneNames = new LinkedHashSet<>(List.of("gene1", "gene2"));
-        List<CommonOrganism> commonTaxons = List.of(new CommonOrganismBuilder().commonTaxon("Bacteroides").build(),new CommonOrganismBuilder().commonTaxon("Enterococcus").build());
+        List<CommonOrganism> commonTaxons = List.of(new CommonOrganismBuilder().commonTaxon("Bacteroides").commonTaxonId(12345L).build(),new CommonOrganismBuilder().commonTaxon("Enterococcus").commonTaxonId(9876L).build());
         LinkedHashSet<Proteome> proteomes = new LinkedHashSet<>(List.of(new ProteomeBuilder().id("UP000005640").component("C1").build(), new ProteomeBuilder().id("UP000002494").component("C2").build()));
 
         return new UniParcEntryLightBuilder()