manage different identities

debug option

build.gradle

@@ -3,7 +3,7 @@ plugins {
 }
 
 group 'org.doremus'
-version '1.3.1'
+version '1.3.2'
 
 sourceCompatibility = 1.8
 

src/main/java/org/doremus/isnimatcher/ISNI.java

@@ -8,6 +8,7 @@
 import javax.xml.bind.JAXBException;
 import java.io.IOException;
 import java.util.ArrayList;
+import java.util.Comparator;
 import java.util.List;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
@@ -20,6 +21,9 @@ public class ISNI {
 
   private final static Pattern FLEX_DATE_PATTERN = Pattern.compile("[\\d?.]{0,4}");
 
+  private static boolean debug = false;
+  private static boolean bestViafBehavior = false;
+
   public static ISNIRecord get(String id) throws IOException {
     if (id.startsWith(ISNI_BASE))
       id = id.replace(ISNI_BASE, "");
@@ -55,12 +59,34 @@ public static ISNIRecord search(String forename, String surname, String date) th
     if (date == null) return records.get(0);
     String _date = cleanDate(date);
 
-    // for (ISNIRecord r : records) r.save("test/" + r.id + ".xml");
-    return records.stream()
+    if (debug) {
+      System.out.println(records.size() + " records");
+      for (ISNIRecord r : records) r.save("test/" + r.id + ".xml");
+    }
+
+    ISNIRecord match = records.stream()
             .filter(r -> r.hasName(forename, surname, true))
             .filter(r -> dateMatch(_date, r.getBirthYear()))
             .findFirst()
             .orElse(null);
+
+    if (bestViafBehavior)
+      return getBestViaf(match, records);
+
+    return match;
+  }
+
+  private static ISNIRecord getBestViaf(ISNIRecord base, List<ISNIRecord> records) {
+    // Among the records with the same VIAF id, return the one with more links
+    if (base == null) return null;
+    String viaf = base.getViafURI();
+    if (viaf == null) return base;
+
+
+    return records.stream()
+            .filter(r -> viaf.equals(r.getViafURI()))
+            .max(Comparator.comparingInt(ISNIRecord::getLinksNumber))
+            .orElse(base);
   }
 
   private static boolean dateMatch(String expected, String actual) {
@@ -86,7 +112,7 @@ private static List<ISNIRecord> performQuery(String query, int n) throws IOExcep
               .queryString("recordSchema", "isni-b")
               .queryString("maximumRecords", n);
 
-      // System.out.println(request.getUrl());
+      if (debug) System.out.println(request.getUrl());
 
       HttpResponse<String> response = request.asString();
 
@@ -131,4 +157,11 @@ static List<String> splitBody(String body) {
   }
 
 
+  public static void setDebug(boolean debug) {
+    ISNI.debug = debug;
+  }
+
+  public static void setBestViafBehavior(boolean bestViafBehavior) {
+    ISNI.bestViafBehavior = bestViafBehavior;
+  }
 }

src/main/java/org/doremus/isnimatcher/ISNIRecord.java

@@ -47,9 +47,9 @@ public class ISNIRecord {
 
   public List<Source> getSources() {
     if (source == null) {
-      source = sources1;
-      if (sources1 == null) source = sources2;
-      else source.addAll(sources2);
+      source = new ArrayList<>();
+      if (sources1 != null) source.addAll(sources1);
+      if (sources2 != null) source.addAll(sources2);
     }
     return source;
   }
@@ -111,6 +111,10 @@ public String getViafURI() {
     for (Source s : getSources())
       if ("VIAF".equals(s.codeOfSource)) return s.asViafURI();
 
+    for (Source s : getSources())
+      if (s.sourceIdentifier.matches("VIAF \\d+"))
+        return Source.makeViafUri(s.sourceIdentifier.replace("VIAF ", ""));
+
     return null;
   }
 
@@ -218,4 +222,7 @@ public boolean hasName(String forename, String surname, boolean _default) {
     return false;
   }
 
+  public int getLinksNumber() {
+    return this.getSources().size() + this.getExternalInformations().size();
+  }
 }

src/main/java/org/doremus/isnimatcher/Source.java

@@ -27,6 +27,11 @@ public String asBNFUri() {
   }
 
   public String asViafURI() {
-    return VIAF_BASE + sourceIdentifier;
+    return makeViafUri(sourceIdentifier);
+  }
+
+  public static String makeViafUri(String viaf) {
+    return VIAF_BASE + viaf;
+
   }
 }

src/test/java/ISNITest.java

@@ -1,5 +1,6 @@
 import org.doremus.isnimatcher.ISNI;
 import org.doremus.isnimatcher.ISNIRecord;
+import org.junit.Before;
 import org.junit.Test;
 
 import javax.xml.bind.JAXBException;
@@ -13,6 +14,11 @@ public class ISNITest {
   private final static String MOZART_URI = "http://isni.org/isni/0000000121269154";
   private final static String BEETHOVEN_ID = "0000000121268987";
 
+  @Before
+  public void initialize() {
+    ISNI.setDebug(true);
+  }
+
   @Test
   public void searchWithFullName() throws IOException {
     String str = "Mozart, Wolfgang Amadeus";
@@ -52,6 +58,21 @@ public void searchWithNameAndDate() throws IOException {
 
   @Test
   public void searchWithNameAndIncompleteDate() throws IOException {
+    ISNIRecord r = ISNI.search("Robert Markham", "19..");
+    assertEquals("0000000107873128", r.id);
+  }
+
+  @Test
+  public void differentIdentities() throws IOException {
+    ISNI.setBestViafBehavior(true);
+    ISNIRecord r = ISNI.search("Guillaume Apollinaire", "1880");
+    assertEquals("0000000121371423", r.id);
+    ISNI.setBestViafBehavior(false);
+  }
+
+
+  @Test
+  public void searchWithNameSurname() throws IOException {
     ISNIRecord r1 = ISNI.search("Martin", "Walter", "19..");
     ISNIRecord r2 = ISNI.search("Walter", "Martin", "19..");
     assertNotEquals(r1.id, r2.id);