Merge pull request #116 from cboehme/issue-115-null-record-ids

Improvements for the WellformednessChecker

Author: Markus M. Geipel

src/main/java/org/culturegraph/mf/framework/StreamReceiver.java

@@ -25,7 +25,7 @@
  * ENTITY_OR_LITERAL = ENTITY | literal
  * ENTITY = startEntity, ENTITY_OR_LITERAL*, endEntity)
  * 
- * The {@link WellFormednessChecker} can be used to check if a stream conforms 
+ * The {@link WellformednessChecker} can be used to check if a stream conforms 
  * to these rules.
  * 
  * @see DefaultStreamReceiver

src/main/java/org/culturegraph/mf/stream/converter/MapToStream.java

@@ -24,32 +24,32 @@
 
 /**
  * Creates a literal for each entry in a map.
- * 
+ *
  * @author Christoph Böhme
  */
 public final class MapToStream extends
 		DefaultObjectPipe<Map<?, ?>, StreamReceiver> {
 
-	
+
 	private Object idKey = StreamConstants.ID;
-	
+
 	public Object getIdKey() {
 		return idKey;
 	}
-	
+
 	public void setIdKey(final Object idKey) {
 		this.idKey = idKey;
 	}
-	
+
 	@Override
 	public void process(final Map<?, ?> map) {
 		final Object id = map.get(idKey);
 		if (id == null) {
-			getReceiver().startRecord(null);
+			getReceiver().startRecord("");
 		} else {
 			getReceiver().startRecord(id.toString());
 		}
-		for (Map.Entry<?, ?> entry: map.entrySet()) {
+		for (final Map.Entry<?, ?> entry: map.entrySet()) {
 			getReceiver().literal(entry.getKey().toString(), entry.getValue().toString());
 		}
 		getReceiver().endRecord();

src/main/java/org/culturegraph/mf/stream/converter/xml/CGXmlHandler.java

@@ -26,9 +26,9 @@
 
 /**
  * Reads CG-XML files.
- * 
+ *
  * @author Christoph Böhme
- * 
+ *
  */
 @Description("Reads CG-XML files")
 @In(XmlReceiver.class)
@@ -46,7 +46,12 @@ public final class CGXmlHandler extends DefaultXmlPipe<StreamReceiver> {
 	public void startElement(final String uri, final String localName,
 			final String qName, final Attributes attributes) {
 		if (RECORD_TAG.equals(localName)) {
-			getReceiver().startRecord(attributes.getValue("", ID_ATTR));
+			final String recordId = attributes.getValue("", ID_ATTR);
+			if (recordId == null) {
+				getReceiver().startRecord("");
+			} else {
+				getReceiver().startRecord(recordId);
+			}
 		} else if (ENTITY_TAG.equals(localName)) {
 			getReceiver().startEntity(attributes.getValue("", NAME_ATTR));
 		} else if (LITERAL_TAG.equals(localName)) {

src/main/java/org/culturegraph/mf/stream/sink/StreamValidator.java

@@ -156,7 +156,7 @@ private void appendChildren(final StringBuilder builder) {
 	private boolean strictKeyOrder;
 	private boolean strictValueOrder;
 
-	private final WellFormednessChecker wellFormednessChecker = new WellFormednessChecker();
+	private final WellformednessChecker wellformednessChecker = new WellformednessChecker();
 
 	public StreamValidator(final List<Event> expectedStream) {
 		this.eventStream = new EventNode(null, null);
@@ -207,7 +207,7 @@ public void startRecord(final String identifier) {
 			throw new ValidationException(VALIDATION_FAILED);
 		}
 
-		wellFormednessChecker.startRecord(identifier);
+		wellformednessChecker.startRecord(identifier);
 
 		validating = true;
 
@@ -224,7 +224,7 @@ public void endRecord() {
 			throw new ValidationException(VALIDATION_FAILED);
 		}
 
-		wellFormednessChecker.endRecord();
+		wellformednessChecker.endRecord();
 
 		if (!closeGroups()) {
 			validationFailed = true;
@@ -240,7 +240,7 @@ public void startEntity(final String name) {
 			throw new ValidationException(VALIDATION_FAILED);
 		}
 
-		wellFormednessChecker.startEntity(name);
+		wellformednessChecker.startEntity(name);
 
 		if (!openGroups(Event.Type.START_ENTITY, name, strictKeyOrder, strictValueOrder)) {
 			validationFailed = true;
@@ -255,7 +255,7 @@ public void endEntity() {
 			throw new ValidationException(VALIDATION_FAILED);
 		}
 
-		wellFormednessChecker.endEntity();
+		wellformednessChecker.endEntity();
 
 		if (!closeGroups()) {
 			validationFailed = true;
@@ -270,7 +270,7 @@ public void literal(final String name, final String value) {
 			throw new ValidationException(VALIDATION_FAILED);
 		}
 
-		wellFormednessChecker.literal(name, value);
+		wellformednessChecker.literal(name, value);
 
 		final List<EventNode> stackFrame = stack.peek();
 
@@ -292,7 +292,7 @@ public void literal(final String name, final String value) {
 
 	@Override
 	public void resetStream() {
-		wellFormednessChecker.resetStream();
+		wellformednessChecker.resetStream();
 
 		validating = false;
 		validationFailed = false;
@@ -308,7 +308,7 @@ public void closeStream() {
 			throw new ValidationException(VALIDATION_FAILED);
 		}
 
-		wellFormednessChecker.closeStream();
+		wellformednessChecker.closeStream();
 
 		validating = false;
 

src/main/java/org/culturegraph/mf/stream/sink/WellFormednessChecker.java renamed to src/main/java/org/culturegraph/mf/stream/sink/WellformednessChecker.java

@@ -19,29 +19,33 @@
 import org.culturegraph.mf.framework.StreamReceiver;
 
 /**
- * A stream receiver that throws an {@link WellformednessException} if 
- * the stream event methods are called in an invalid order. Additionally, 
+ * A stream receiver that throws an {@link WellformednessException} if
+ * the stream event methods are called in an invalid order. Additionally,
  * the stream receiver checks that entity and literal names are not null.
- * 
+ *
  * @see StreamValidator
  * @see WellformednessException
- * 
+ *
  * @author Christoph Böhme
- * 
+ *
  */
-public final class WellFormednessChecker implements StreamReceiver {
-	
+public final class WellformednessChecker implements StreamReceiver {
+
+	private static final String ID_MUST_NOT_BE_NULL = "id must not be null";
 	private static final String NAME_MUST_NOT_BE_NULL = "name must not be null";
-	
+
 	private static final String NOT_IN_RECORD = "Not in record";
 	private static final String NOT_IN_ENTITY = "Not in entity";
 	private static final String IN_ENTITY = "In entity";
 	private static final String IN_RECORD = "In record";
-	
+
 	private int nestingLevel;
-	
+
 	@Override
 	public void startRecord(final String identifier) {
+		if (identifier == null) {
+			throw new WellformednessException(ID_MUST_NOT_BE_NULL);
+		}
 		if (nestingLevel > 0) {
 			throw new WellformednessException(IN_RECORD);
 		}
@@ -50,10 +54,10 @@ public void startRecord(final String identifier) {
 
 	@Override
 	public void endRecord() {
-		if (nestingLevel < 1) { 
+		if (nestingLevel < 1) {
 			throw new WellformednessException(NOT_IN_RECORD);
 		} else if (nestingLevel > 1) {
-			throw new WellformednessException(IN_ENTITY);			
+			throw new WellformednessException(IN_ENTITY);
 		}
 		nestingLevel -= 1;
 	}
@@ -86,7 +90,7 @@ public void literal(final String name, final String value) {
 			throw new WellformednessException(NOT_IN_RECORD);
 		}
 	}
-	
+
 	@Override
 	public void resetStream() {
 		nestingLevel = 0;

src/test/java/org/culturegraph/mf/stream/converter/MapToStreamTest.java

@@ -19,7 +19,6 @@
 import java.util.Map;
 
 import org.culturegraph.mf.exceptions.FormatException;
-import org.culturegraph.mf.stream.converter.MapToStream;
 import org.culturegraph.mf.stream.sink.EventList;
 import org.culturegraph.mf.stream.sink.StreamValidator;
 import org.culturegraph.mf.util.StreamConstants;
@@ -29,9 +28,9 @@
 
 /**
  * Test {@link MapToStream}
- * 
+ *
  * @author Christoph Böhme
- * 
+ *
  */
 public final class MapToStreamTest {
 
@@ -51,16 +50,16 @@ public void testStringStringMap() {
 		expected.literal(KEYS[1], VALUES[1]);
 		expected.literal(StreamConstants.ID, RECORD_ID);
 		expected.endRecord();
-		
+
 		final Map<String, String> map = new HashMap<String, String>();
 		map.put(KEYS[0], VALUES[0]);
 		map.put(KEYS[1], VALUES[1]);
 		map.put(StreamConstants.ID, RECORD_ID);
 
 		final MapToStream mapToStream = new MapToStream();
-		
+
 		Assert.assertEquals(StreamConstants.ID, mapToStream.getIdKey());
-		
+
 		checkResults(expected, mapToStream, map);
 	}
 
@@ -73,48 +72,48 @@ public void testIntIntMap() {
 		expected.literal(KEYS[1], VALUES[1]);
 		expected.literal(Integer.toString(INT_ID_KEY), RECORD_ID);
 		expected.endRecord();
-		
+
 		final Map<Integer, Integer> map = new HashMap<Integer, Integer>();
 		map.put(INT_KEYS[0], INT_VALUES[0]);
 		map.put(INT_KEYS[1], INT_VALUES[1]);
 		map.put(INT_ID_KEY, INT_RECORD_ID);
-		
+
 		final MapToStream mapToStream = new MapToStream();
 		mapToStream.setIdKey(-1);
-		
+
 		Assert.assertEquals(-1, mapToStream.getIdKey());
-		
+
 		checkResults(expected, mapToStream, map);
 	}
-	
+
 	@Test
 	public void testStringIntMap() {
 		final EventList expected = new EventList();
-		expected.startRecord(null);
+		expected.startRecord("");
 		expected.literal(KEYS[0], VALUES[0]);
 		expected.literal(KEYS[1], VALUES[1]);
 		expected.endRecord();
-		
+
 		final Map<String, Integer> map = new HashMap<String, Integer>();
 		map.put(KEYS[0], INT_VALUES[0]);
 		map.put(KEYS[1], INT_VALUES[1]);
-		
+
 		final MapToStream mapToStream = new MapToStream();
-		
+
 		checkResults(expected, mapToStream, map);
 	}
-	
+
 	private void checkResults(final EventList expected, final MapToStream mapToStream, final Map<?, ?> map) {
-		
+
 		final StreamValidator validator = new StreamValidator(expected.getEvents());
-		
+
 		mapToStream.setReceiver(validator);
-		
+
 		try {
 			mapToStream.process(map);
 			mapToStream.closeStream();
-		} catch (FormatException e){
+		} catch (final FormatException e){
 			Assert.fail(e.toString());
-		}		
+		}
 	}
 }

src/test/java/org/culturegraph/mf/stream/converter/xml/CGXmlReaderTest.java

@@ -16,8 +16,6 @@
 package org.culturegraph.mf.stream.converter.xml;
 
 import org.culturegraph.mf.stream.DataFilePath;
-import org.culturegraph.mf.stream.converter.xml.CGXmlHandler;
-import org.culturegraph.mf.stream.converter.xml.XmlDecoder;
 import org.culturegraph.mf.stream.sink.EventList;
 import org.culturegraph.mf.stream.sink.StreamValidator;
 import org.culturegraph.mf.stream.source.ResourceOpener;
@@ -35,27 +33,27 @@ public final class CGXmlReaderTest {
 	private static final String NUMBER = "Number";
 	private static final String POSTCODE = "Postcode";
 	private static final String CITY = "City";
-	
+
 	@Test
 	public void testReadStringStreamReceiver() {
 		final ResourceOpener opener = new ResourceOpener();
 		final XmlDecoder saxReader = new XmlDecoder();
 		final CGXmlHandler cgHandler = new CGXmlHandler();
 		final EventList writer = new EventList();
-		
-		
+
+
 		opener.setReceiver(saxReader)
 				.setReceiver(cgHandler)
 				.setReceiver(writer);
-		
+
 		opener.process(DataFilePath.CG_XML);
 		opener.closeStream();
-		
+
 		final StreamValidator validator = new StreamValidator(writer.getEvents());
 		validator.setStrictRecordOrder(true);
 		validator.setStrictKeyOrder(true);
 		validator.setStrictValueOrder(true);
-		
+
 		validator.startRecord("1");
 			validator.literal(NAME, "Thomas Mann");
 			validator.startEntity(ADDRESS);
@@ -67,7 +65,7 @@ public void testReadStringStreamReceiver() {
 				validator.literal(POSTCODE, null);
 			validator.endEntity();
 		validator.endRecord();
-		validator.startRecord(null);
+		validator.startRecord("");
 			validator.literal(NAME, "Günter Grass");
 			validator.startEntity(ADDRESS);
 				validator.startEntity(STREET);