ignore whitespaces for ncbi taxon id resolving

globalbioticinteractions · Oct 23, 2018 · fa0aeb4 · fa0aeb4
1 parent ac2c999
commit fa0aeb4
Show file tree

Hide file tree

Showing 2 changed files with 16 additions and 2 deletions.
diff --git a/nomer-taxon-resolver/src/main/java/org/eol/globi/taxon/NCBIService.java b/nomer-taxon-resolver/src/main/java/org/eol/globi/taxon/NCBIService.java
@@ -40,8 +40,11 @@ public Map<String, String> enrich(Map<String, String> properties) throws Propert
         String externalId = properties.get(PropertyAndValueDictionary.EXTERNAL_ID);
 
         if (PREFIXES.stream().anyMatch(x -> StringUtils.startsWith(externalId, x))) {
-            String tsn = PREFIXES.stream().reduce(externalId, (x, y) -> StringUtils.replace(x, y, ""));
-            if (tsn.matches("\\d+")) {
+            String tsn = PREFIXES
+                    .stream()
+                    .reduce(externalId, (x, y) -> StringUtils.trim(StringUtils.replace(x, y, "")));
+
+            if (tsn.matches("^\\d+$")) {
                 String fullHierarchy = getResponse("db=taxonomy&id=" + tsn);
                 if (fullHierarchy.contains("<Taxon>")) {
                     parseAndPopulate(enriched, tsn, fullHierarchy);

diff --git a/nomer-taxon-resolver/src/test/java/org/eol/globi/taxon/NCBIServiceTest.java b/nomer-taxon-resolver/src/test/java/org/eol/globi/taxon/NCBIServiceTest.java
@@ -59,6 +59,17 @@ public void lookupPathByTaxonIdNCBIPurl() throws PropertyEnricherException {
         assertThat(enrich.get(PropertyAndValueDictionary.EXTERNAL_ID), is("NCBI:9606"));
     }
 
+    @Test
+    public void lookupPathByTaxonIdNCBIPurlWhitespaces() throws PropertyEnricherException {
+        PropertyEnricher enricher = new NCBIService();
+        HashMap<String, String> props = new HashMap<String, String>() {{
+            put(PropertyAndValueDictionary.EXTERNAL_ID, "http://purl.obolibrary.org/obo/NCBITaxon_9606 ");
+        }};
+        Map<String, String> enrich = enricher.enrich(props);
+        assertThat(enrich.get(PropertyAndValueDictionary.NAME), is("Homo sapiens"));
+        assertThat(enrich.get(PropertyAndValueDictionary.EXTERNAL_ID), is("NCBI:9606"));
+    }
+
 
     @Test
     public void parseInconsistentWithAltNCBI191217() throws PropertyEnricherException {