attempt to ignore greedy ncbi matches from globalnames. related to Gl…

…obalNamesArchitecture/gni#48
globalbioticinteractions · Dec 18, 2018 · 2008d28 · 2008d28
1 parent 6d5ee65
commit 2008d28
Show file tree

Hide file tree

Showing 2 changed files with 37 additions and 1 deletion.
diff --git a/nomer-taxon-resolver/src/main/java/org/eol/globi/taxon/GlobalNamesService2.java b/nomer-taxon-resolver/src/main/java/org/eol/globi/taxon/GlobalNamesService2.java
@@ -165,6 +165,7 @@ private void parseResult(TermMatchListener termMatchListener, String result) thr
     }
 
     private void parseResultNode(TermMatchListener termMatchListener, JsonNode jsonNode) {
+
         JsonNode dataList = jsonNode.get("data");
         if (dataList != null && dataList.isArray()) {
             for (JsonNode data : dataList) {
@@ -287,7 +288,7 @@ protected void parseClassification(TermMatchListener termMatchListener, JsonNode
             }
 
             // related to https://github.com/GlobalNamesArchitecture/gni/issues/48
-            if (!pathTailRepetitions(taxon)) {
+            if (!pathTailRepetitions(taxon) && !speciesNoGenus(taxon)) {
                 termMatchListener.foundTaxonForName(requestId(data), suppliedNameString, taxon, nameType);
             }
         }
@@ -310,6 +311,11 @@ protected void parseClassification(TermMatchListener termMatchListener, JsonNode
         }
     }
 
+    private boolean speciesNoGenus(Taxon taxon) {
+        return taxon.getPathNames().contains("species")
+                && !taxon.getPathNames().contains("genus");
+    }
+
     private String getSuppliedNameString(JsonNode data) {
         return data.get("supplied_name_string").getTextValue();
     }

diff --git a/nomer-taxon-resolver/src/test/java/org/eol/globi/service/GlobalNamesService2Test.java b/nomer-taxon-resolver/src/test/java/org/eol/globi/service/GlobalNamesService2Test.java
@@ -438,4 +438,34 @@ public void foundTaxonForName(Long nodeId, String name, Taxon taxon, NameType na
 
     }
 
+    @Test
+    public void lookupNCBIIgnoreCanonicalMatchOnExactMatch() throws PropertyEnricherException {
+        GlobalNamesService2 service = new GlobalNamesService2(Arrays.asList(GlobalNamesSources2.NCBI));
+        final List<Taxon> taxa = new ArrayList<>();
+        service.findTermsForNames(Collections.singletonList("Amphipoda"), (nodeId, name, taxon, nameType) -> {
+            taxa.add(taxon);
+            assertThat(nameType, is(NameType.SAME_AS));
+        });
+
+        assertThat(taxa.size(), is(1));
+        assertThat(taxa.get(0).getPath(), endsWith("Amphipoda"));
+        assertThat(taxa.get(0).getPathNames(), endsWith("order"));
+
+    }
+
+    @Test
+    public void lookupNCBIIgnoreCanonicalMatchSingleWordSpecies() throws PropertyEnricherException {
+        GlobalNamesService2 service = new GlobalNamesService2(Collections.singletonList(GlobalNamesSources2.NCBI));
+        final List<Taxon> taxa = new ArrayList<>();
+        service.findTermsForNames(Collections.singletonList("Procladius sp1 M_PL_014"), (nodeId, name, taxon, nameType) -> {
+            taxa.add(taxon);
+            assertThat(nameType, is(NameType.SAME_AS));
+        });
+
+        assertThat(taxa.size(), is(1));
+        assertThat(taxa.get(0).getPath(), endsWith("Procladius"));
+        assertThat(taxa.get(0).getPathNames(), endsWith("genus"));
+
+    }
+
 }