add subspecies patterns; related to #161

globalbioticinteractions · Apr 5, 2024 · 498aadb · 498aadb
1 parent d04d166
commit 498aadb
Show file tree

Hide file tree

Showing 3 changed files with 38 additions and 13 deletions.
diff --git a/nomer-taxon-resolver/src/main/java/org/eol/globi/taxon/DiscoverLifeUtil2.java b/nomer-taxon-resolver/src/main/java/org/eol/globi/taxon/DiscoverLifeUtil2.java
@@ -42,11 +42,33 @@
 public class DiscoverLifeUtil2 {
 
     public static final List<String> RANKS = Arrays.asList("Family", "Subfamily", "Tribe", "Subtribe", "Genus", "Subgenus");
-    public static final String NAME_PATTERN_AUTHORSHIP_PARENTHESES = "(?<name>[A-Z][a-z]+[ ][a-z]+)[ ]+(?<authorship>[(][^,]+[,][ ][0-9]{4}[)])";
-    public static final String NAME_PATTERN_WITH_NOTE = "(?<name>[A-Z][a-z]+[ ][a-z]+)(?<note>[_][_a-z]+)[ ]+(?<authorship>[^,]+[,][ ][0-9]{4})";
-    public static final String NAME_PATTERN_WITH_PARENTHESIS = "(?<name>[A-Z][a-z]+[ ][(][A-Z][a-z]+[)][ ][a-z]+)[ ]+(?<authorship>[^,]+[,][ ][0-9]{4})";
-    public static final String NAME_PATTERN_AUTHORSHIP_MULTIPLE_AUTHORS = "(?<name>[A-Z][a-z]+[ ][a-z]+)[ ]+(?<authorship>([A-Z][a-z]+)([ ]and[ ])([A-Z][a-z]+)[,][ ][0-9]{4})";
-    public static final List<String> NAME_PATTERNS = Arrays.asList(NAME_PATTERN_AUTHORSHIP_PARENTHESES, NAME_PATTERN_WITH_NOTE, NAME_PATTERN_WITH_PARENTHESIS, NAME_PATTERN_AUTHORSHIP_MULTIPLE_AUTHORS);
+
+    public static final String NAME = "(?<name>([A-Z][a-z]+)([ ][a-z]+)([ ][a-z]+){0,1})";
+    public static final String NAME_PARENTHESIS = "(?<name>[A-Z][a-z]+[ ][(][A-Z][a-z]+[)][ ][a-z]+)";
+    public static final String AUTHORSHIP = "(?<authorship>[^,]+[,][ ][0-9]{4})";
+    public static final String AUTHORSHIP_PARENTHESIS = "(?<authorship>[(][^,]+[,][ ][0-9]{4}[)])";
+    public static final String AUTHORSHIP_AND = "(?<authorship>([A-Z][a-z]+)([ ]and[ ])([A-Z][a-z]+)[,][ ][0-9]{4})";
+    public static final String SPACE = "[ ]+";
+    public static final String NOTE = "(?<note>[_][_a-z]+)";
+
+    public static final String NAME_AUTHORSHIP_PARENTHESES
+            = NAME + SPACE + AUTHORSHIP_PARENTHESIS;
+    public static final String NAME_AUTHORSHIP
+            = NAME + SPACE + AUTHORSHIP;
+    public static final String NAME_WITH_NOTE
+            = NAME + NOTE + SPACE + AUTHORSHIP;
+    public static final String NAME_WITH_PARENTHESIS
+            = NAME_PARENTHESIS + SPACE + AUTHORSHIP;
+    public static final String NAME_AUTHORSHIP_MULTIPLE_AUTHORS
+            = NAME + SPACE + AUTHORSHIP_AND;
+
+
+    public static final List<String> NAME_PATTERNS = Arrays.asList(
+            NAME_AUTHORSHIP,
+            NAME_AUTHORSHIP_PARENTHESES,
+            NAME_WITH_NOTE,
+            NAME_WITH_PARENTHESIS,
+            NAME_AUTHORSHIP_MULTIPLE_AUTHORS);
 
     public static void splitRecords(InputStream is, Consumer<String> lineConsumer) {
         Scanner scanner = new Scanner(is, StandardCharsets.UTF_8.name());

diff --git a/nomer-taxon-resolver/src/test/java/org/eol/globi/taxon/DiscoverLifeUtil2IntegrationTest.java b/nomer-taxon-resolver/src/test/java/org/eol/globi/taxon/DiscoverLifeUtil2IntegrationTest.java
@@ -33,14 +33,9 @@ public void parseNames() throws ParserConfigurationException, IOException {
             @Override
             public void foundTaxonForTerm(Long requestId, Term providedTerm, NameType nameType, Taxon resolvedTaxon) {
                 counter.incrementAndGet();
-                Taxon providedTaxon = (Taxon) providedTerm;
-                if (!NameType.HAS_ACCEPTED_NAME.equals(nameType)) {
-                    System.out.println(providedTaxon.getName());
-                    System.out.println(providedTaxon.getAuthorship());
-                }
             }
         });
-        assertThat(counter.get(), Is.is(51164));
+        assertThat(counter.get(), Is.is(31278));
     }
 
 }
diff --git a/nomer-taxon-resolver/src/test/java/org/eol/globi/taxon/DiscoverLifeUtil2Test.java b/nomer-taxon-resolver/src/test/java/org/eol/globi/taxon/DiscoverLifeUtil2Test.java
@@ -48,9 +48,9 @@ public void parseRelatedNames() throws ParserConfigurationException, XPathExpres
 
         List<Taxon> taxons = DiscoverLifeUtil2.parseRelatedNames(doc);
 
-        assertThat(taxons.size(), is(13));
+        assertThat(taxons.size(), is(17));
 
-        Taxon lastTaxon = taxons.get(12);
+        Taxon lastTaxon = taxons.get(taxons.size() - 1);
         assertThat(lastTaxon.getName(), is("Andrena (Holandrena) cressonii"));
         assertThat(lastTaxon.getAuthorship(), is("Robertson, 1891"));
     }
@@ -168,6 +168,14 @@ public void parseNameAlt4() {
 
     }
 
+    @Test
+    public void parseNameAlt5() {
+        Taxon matched = DiscoverLifeUtil2.parse("Agapostemon texanus subtilior Cockerell, 1898");
+        assertThat(matched.getName(), is("Agapostemon texanus subtilior"));
+        assertThat(matched.getAuthorship(), is("Cockerell, 1898"));
+
+    }
+
     @Test
     public void patchCommonNames() {
         String name = "Protandrena bachue Gonzalez and Ruz, 2007 Rhophitulus bachue (Gonzalez and Ruz, 2007)";