improve parsing of names and sources; update XML output from viaf in test suite; bump version

Author: codeforkjeff

README.md

@@ -45,7 +45,7 @@ build the .jar file using maven.
 Run this command:
 
 ```
-java -jar refine_viaf-1.0.jar
+java -jar refine_viaf-1.1.jar
 ```
 
 That's it! You should see some messages as the application starts

pom.xml

@@ -4,7 +4,7 @@
 
     <groupId>org.refine_viaf</groupId>
     <artifactId>refine_viaf</artifactId>
-    <version>1.0</version>
+    <version>1.1</version>
     <packaging>jar</packaging>
 
     <name>refine_viaf</name>
@@ -17,7 +17,7 @@
     <parent>
         <groupId>org.springframework.boot</groupId>
         <artifactId>spring-boot-starter-parent</artifactId>
-        <version>1.3.0.RELEASE</version>
+        <version>1.3.2.RELEASE</version>
     </parent>
 
     <dependencies>

run.sh

@@ -14,4 +14,4 @@
 # the JVM from having to dynamically allocate memory, which takes time.
 # -Xms128m -Xmx128m
 
-java -Xms128m -Xmx128m -Dlogging.level.com.codefork.refine=DEBUG -jar target/refine_viaf-1.0.jar
+java -Xms128m -Xmx128m -Dlogging.level.com.codefork.refine=DEBUG -jar target/refine_viaf-1.1.jar

src/main/java/com/codefork/refine/viaf/VIAFParser.java

@@ -17,6 +17,8 @@ public class VIAFParser extends DefaultHandler {
     private boolean captureChars = false;
     private boolean insideHeadings = false;
     private boolean insideSources = false;
+    private int depth = 0;
+    private int headingsDepth = -1;
 
     // viaf's weird indexed namespacing
     private int nsIndex = 2;
@@ -52,6 +54,7 @@ public void startElement(String uri, String localName, String qName, Attributes
             captureChars = true;
         } else if (getElementNameWithNS("mainHeadings").equals(qName)) {
             insideHeadings = true;
+            headingsDepth = depth;
         } else if (insideHeadings && getElementNameWithNS("data").equals(qName)) {
             getLastResult().addNameEntry();
         } else if (insideHeadings && getElementNameWithNS("text").equals(qName)) {
@@ -60,13 +63,14 @@ public void startElement(String uri, String localName, String qName, Attributes
             insideSources = true;
         } else if (insideSources && getElementNameWithNS("s").equals(qName)) {
             captureChars = true;
-        } else if (insideHeadings) {
+        } else if (headingsDepth != -1 && depth == headingsDepth + 1) {
             // if we got here, we encountered some other child of mainHeadings
             // so we want to effectively end the section, otherwise we'll end up
             // erroneously picking up other "text" and "sources" elements nested
             // under other elements in mainHeadings
             insideHeadings = false;
         }
+        depth ++;
     }
 
     @Override
@@ -83,6 +87,7 @@ public void endElement(String uri, String localName, String qName) throws SAXExc
             captureChars = false;
         } else if (getElementNameWithNS("mainHeadings").equals(qName)) {
             insideHeadings = false;
+            headingsDepth = -1;
         } else if (insideHeadings && getElementNameWithNS("text").equals(qName)) {
             getLastResult().getLastNameEntry().setName(buf.toString());
             buf = new StringBuilder();
@@ -94,6 +99,7 @@ public void endElement(String uri, String localName, String qName) throws SAXExc
             buf = new StringBuilder();
             captureChars = false;
         }
+        depth --;
     }
 
     @Override

src/test/java/com/codefork/refine/viaf/VIAFParserTest.java

@@ -0,0 +1,77 @@
+package com.codefork.refine.viaf;
+
+import org.junit.Test;
+
+import javax.xml.parsers.SAXParser;
+import javax.xml.parsers.SAXParserFactory;
+import java.io.InputStream;
+import java.util.List;
+
+import static org.junit.Assert.assertEquals;
+
+public class VIAFParserTest {
+
+    private static String joinStrings(List<String> strings, String delimiter) {
+        StringBuilder b = new StringBuilder();
+        for(String s : strings) {
+            if(b.length() > 0) {
+                b.append(delimiter);
+            }
+            b.append(s);
+        }
+        return b.toString();
+    }
+
+    @Test
+    public void testParseNames() throws Exception {
+        SAXParserFactory spf = SAXParserFactory.newInstance();
+        SAXParser parser = spf.newSAXParser();
+        VIAFParser viafParser = new VIAFParser();
+
+        InputStream is = getClass().getResourceAsStream("/steinbeck_no_type.xml");
+        parser.parse(is, viafParser);
+
+        List<VIAFResult> results = viafParser.getResults();
+
+        VIAFResult firstResult = results.get(0);
+        VIAFResult secondResult = results.get(1);
+
+        assertEquals(10, firstResult.getNameEntries().size());
+
+        assertEquals("Steinbeck, John, 1902-1968",
+                firstResult.getNameEntries().get(0).getName());
+        assertEquals("LC,BIBSYS,BNF,KRNLK,N6I,LAC,BNE,SUDOC,BAV,BNC,NLI,B2Q,PTBNP,NLP,LNB,SELIBR,NLA,ICCU,NDL,DNB,NUKAT,NKC",
+                joinStrings(firstResult.getNameEntries().get(0).getSources(), ","));
+
+        assertEquals("Steinbeck, John (John Ernst), 1902-1968",
+                firstResult.getNameEntries().get(1).getName());
+        assertEquals("NTA",
+                joinStrings(firstResult.getNameEntries().get(1).getSources(), ","));
+
+        assertEquals("NSK,SWNL",
+                joinStrings(firstResult.getNameEntries().get(2).getSources(), ","));
+        assertEquals("WKP",
+                joinStrings(firstResult.getNameEntries().get(3).getSources(), ","));
+        assertEquals("LNL,EGAXA",
+                joinStrings(firstResult.getNameEntries().get(4).getSources(), ","));
+        assertEquals("NLI",
+                joinStrings(firstResult.getNameEntries().get(5).getSources(), ","));
+        assertEquals("NLI",
+                joinStrings(firstResult.getNameEntries().get(6).getSources(), ","));
+        assertEquals("NLI",
+                joinStrings(firstResult.getNameEntries().get(7).getSources(), ","));
+        assertEquals("NLR",
+                joinStrings(firstResult.getNameEntries().get(8).getSources(), ","));
+        assertEquals("JPG",
+                joinStrings(firstResult.getNameEntries().get(9).getSources(), ","));
+
+        assertEquals(5, secondResult.getNameEntries().size());
+
+        assertEquals("Steinbeck, John 1946-1991",
+                secondResult.getNameEntries().get(0).getName());
+        assertEquals("NLP,ICCU,DNB,BNF",
+                joinStrings(secondResult.getNameEntries().get(0).getSources(), ","));
+
+    }
+
+}

src/test/java/com/codefork/refine/viaf/VIAFTest.java

@@ -70,15 +70,15 @@ public void testSearchNoParticularType() throws Exception {
         assertFalse(result1.isMatch());
 
         Result result2 = results.get(1);
-        assertEquals("Steinbeck, John, 1902-1968. | Of mice and men.", result2.getName());
-        assertEquals(NameType.Book.asVIAFNameType(), result2.getType().get(0));
-        assertEquals("180993990", result2.getId());
+        assertEquals("Steinbeck, John 1946-1991", result2.getName());
+        assertEquals(NameType.Person.asVIAFNameType(), result2.getType().get(0));
+        assertEquals("19893647", result2.getId());
         assertFalse(result2.isMatch());
 
         Result result3 = results.get(2);
-        assertEquals("Steinbeck, John 1946-1991", result3.getName());
-        assertEquals(NameType.Person.asVIAFNameType(), result3.getType().get(0));
-        assertEquals("19893647", result3.getId());
+        assertEquals("Steinbeck, John, 1902-1968. | Of mice and men.", result3.getName());
+        assertEquals(NameType.Book.asVIAFNameType(), result3.getType().get(0));
+        assertEquals("180993990", result3.getId());
         assertFalse(result3.isMatch());
     }
 
@@ -105,15 +105,15 @@ public void testSearchWithSource() throws Exception {
         assertFalse(result1.isMatch());
 
         Result result2 = results.get(1);
-        assertEquals("Nabokov, Vladimir Vladimirovič | Volšebnik", result2.getName());
+        assertEquals("Nabokov, Vladimir Vladimirovič | Lolita", result2.getName());
         assertEquals(NameType.Book.asVIAFNameType(), result2.getType().get(0));
-        assertEquals("316638111", result2.getId());
+        assertEquals("176671347", result2.getId());
         assertFalse(result2.isMatch());
 
         Result result3 = results.get(2);
         assertEquals("Nabokov, Vladimir Vladimirovič | Govori, sjećanje!", result3.getName());
         assertEquals(NameType.Book.asVIAFNameType(), result3.getType().get(0));
-        assertEquals("140144814502844904157", result3.getId());
+        assertEquals("183561595", result3.getId());
         assertFalse(result3.isMatch());
     }