Fixed filter implementation to allow filename only matches (also fixed non-deterministic unit tests)

Author: andybrodie

src/com/locima/xml2csv/configuration/filter/FileNameInputFilter.java

@@ -15,19 +15,22 @@
 public class FileNameInputFilter extends FilterContainer {
 
 	private static final Logger LOG = LoggerFactory.getLogger(FileNameInputFilter.class);
+	private boolean matchLocalFileNameOnly;
 	private Pattern pattern;
 
 	/**
 	 * Sets the regex that this filter will use to match on {@link #include(File)}.
 	 *
 	 * @param regex the regular expression that will be used to match. Must not be null.
+	 * @param matchLocalFileNameOnly if true, then only the relative filename (i.e. no directory information) will be passed to the filter.
 	 * @throws PatternSyntaxException if the regular expression passed by <code>regex</code> is invalid.
 	 */
 	// CHECKSTYLE:OFF I don't care if PatternSyntaxException is a runtime exception, it's pertinent!
-	public FileNameInputFilter(String regex) throws PatternSyntaxException {
+	public FileNameInputFilter(String regex, boolean matchLocalFileNameOnly) throws PatternSyntaxException {
 		// CHECKSTYLE:ON
 		LOG.debug("Compiling regex {}", regex);
 		this.pattern = Pattern.compile(regex);
+		this.matchLocalFileNameOnly = matchLocalFileNameOnly;
 	}
 
 	/**
@@ -39,16 +42,18 @@ public FileNameInputFilter(String regex) throws PatternSyntaxException {
 	@Override
 	public boolean include(File xmlInputFile) {
 		boolean match;
-		String absPath = xmlInputFile.getAbsolutePath();
+		String pathToTest = this.matchLocalFileNameOnly ? xmlInputFile.getName() : xmlInputFile.getAbsolutePath();
 		if (this.pattern == null) {
 			LOG.warn("Regex pattern not specified on FileNameInputFilter, returning true");
 			match = false;
 		} else {
-			match = this.pattern.matcher(absPath).find();
-			LOG.debug("Input file {} did {}match file name input filter {}", absPath, match ? "" : "not ", this.pattern);
+			match = this.pattern.matcher(pathToTest).find();
+			LOG.debug("Input file {} did {}match file name input filter {}", pathToTest, match ? "" : "not ", this.pattern);
 			if (match) {
 				match = this.executeNestedFilters(xmlInputFile);
-				LOG.trace("Input file {} excluded by nested filter", absPath);
+				if (!match) {
+					LOG.trace("Nested filter of {} rejected {}", this, xmlInputFile);
+				}
 			}
 		}
 		return match;

src/com/locima/xml2csv/inputparser/xml/ConfigContentHandler.java

@@ -7,6 +7,7 @@
 import java.util.regex.PatternSyntaxException;
 
 import javax.xml.XMLConstants;
+import javax.xml.bind.DatatypeConverter;
 
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -51,6 +52,7 @@ private static enum ElementNames {
 
 	private static final String CUSTOM_NAME_FORMAT_ATTR = "customNameFormat";
 
+	private static final String FILENAME_FILTER_MATCH_LOCAL_ATTR = "matchLocalFileNameOnly";
 	private static final String GROUP_NUMBER_ATTR = "group";
 	private static final String KEY_XPATH_ATTR = "keyXPath";
 	private static final String KVPAIR_ROOT_XPATH_ATTR = "kvPairRoot";
@@ -322,13 +324,38 @@ private void endMappingList() {
 		}
 	}
 
+	/**
+	 * Retrieve an attribute value cast as a Boolean, according the rules specified by XSD spec ({@link DatatypeConverter#parseBoolean(String)}).
+	 *
+	 * @param atts the set of of attributes to retrieve from.
+	 * @param attrName the name of the attribute to retrieve.
+	 * @return true or false, depending on the value of the attribute, as per the XSD spec.
+	 * @throws SAXException if the value specified by <code>attrName</code> is present but not valid according to the XSD spec for XSD boolean value
+	 *             types.
+	 */
+	private boolean getAttributeValueAsBoolean(Attributes atts, String attrName) throws SAXException {
+		String attrValueAsString = atts.getValue(attrName);
+		if (attrValueAsString == null) {
+			return false;
+		} else {
+			try {
+				boolean parsedBoolean = DatatypeConverter.parseBoolean(attrValueAsString);
+				LOG.trace("Parsed {} value {} as {}", attrName, attrValueAsString, parsedBoolean);
+				return parsedBoolean;
+			} catch (IllegalArgumentException iae) {
+				// Thrown by DatatypeConverter.parseBoolean for an invalid argument
+				throw getException(iae, "Invalid value \"%s\" found in attriute \"%s\"", attrValueAsString, attrName);
+			}
+		}
+	}
+
 	/**
 	 * Retrieve an attribute value cast as an Integer, null if one wasn't specified, or throw an exception if malformed.
 	 *
 	 * @param atts the set of of attributes to retrieve from.
 	 * @param attrName the name of the attribute to retrieve.
-	 * @return the value retrieved, or value in <code>defaultValue</code> if a value could not be found.
-	 * @throws SAXException thrown if the value found in the attribute is an integer.
+	 * @return an integer value of the value in <code>attrName</code>, null if no value was specified.
+	 * @throws SAXException thrown if the value found in the attribute is not an integer.
 	 */
 	private Integer getAttributeValueAsInteger(Attributes atts, String attrName) throws SAXException {
 		String attrValueAsString = atts.getValue(attrName);
@@ -343,27 +370,6 @@ private Integer getAttributeValueAsInteger(Attributes atts, String attrName) thr
 		}
 	}
 
-	/**
-	 * Gets a list of the known child names of the parent container.  These are used as variable names, available in the XPath evaluation of subsequent
-	 * mappings with the same parent.
-	 * @return an array, possibly empty, of known siblings of the current mapping (as determined by the top of {@link #mappingListStack}.
-	 */
-	private String[] getPreviousSiblingNames() {
-		List<String> variables = new ArrayList<String>();
-		MappingList currentContainer = this.mappingListStack.peek();
-		for (IMapping m : currentContainer) {
-			if (m instanceof IValueMapping) {
-				IValueMapping vm = (IValueMapping) m;
-				variables.add(vm.getName());
-			}
-		}
-		String[] varArray = variables.toArray(new String[0]);
-		if (LOG.isDebugEnabled()) {
-			LOG.debug("Created variable set for {}: {}", currentContainer.getName(), StringUtil.toStringList(varArray));
-		}
-		return varArray;
-	}
-
 	/**
 	 * Given an element's local name (namespace URI checking is left to the caller), return an enum value that can be used in switch statements to
 	 * branch logic depending on the input.
@@ -393,9 +399,8 @@ private ElementNames getElementNameEnum(String localName) throws SAXException {
 	 */
 	private SAXException getException(Exception inner, String message, Object... parameters) {
 		String temp = String.format(message, parameters);
-		XMLException de =
-						new XMLException(inner, "Error parsing %s(%d:%d) %s", this.documentLocator.getSystemId(),
-										this.documentLocator.getLineNumber(), this.documentLocator.getColumnNumber(), temp);
+		XMLException de = new XMLException(inner, "Error parsing %s(%d:%d) %s", this.documentLocator.getSystemId(),
+						this.documentLocator.getLineNumber(), this.documentLocator.getColumnNumber(), temp);
 		SAXException se = new SAXException(de);
 		return se;
 	}
@@ -409,6 +414,28 @@ public MappingConfiguration getMappings() {
 		return this.mappingConfiguration;
 	}
 
+	/**
+	 * Gets a list of the known child names of the parent container. These are used as variable names, available in the XPath evaluation of subsequent
+	 * mappings with the same parent.
+	 *
+	 * @return an array, possibly empty, of known siblings of the current mapping (as determined by the top of {@link #mappingListStack}.
+	 */
+	private String[] getPreviousSiblingNames() {
+		List<String> variables = new ArrayList<String>();
+		MappingList currentContainer = this.mappingListStack.peek();
+		for (IMapping m : currentContainer) {
+			if (m instanceof IValueMapping) {
+				IValueMapping vm = (IValueMapping) m;
+				variables.add(vm.getName());
+			}
+		}
+		String[] varArray = variables.toArray(new String[0]);
+		if (LOG.isDebugEnabled()) {
+			LOG.debug("Created variable set for {}: {}", currentContainer.getName(), StringUtil.toStringList(varArray));
+		}
+		return varArray;
+	}
+
 	@Override
 	public void setDocumentLocator(Locator locator) {
 		this.documentLocator = locator;
@@ -466,7 +493,7 @@ public void startElement(String uri, String localName, String qName, Attributes
 					startXPathFilter(atts.getValue("xPath"));
 					break;
 				case FileNameInputFilter:
-					startFileNameFilter(atts.getValue("fileNameRegex"));
+					startFileNameFilter(atts.getValue("fileNameRegex"), getAttributeValueAsBoolean(atts, FILENAME_FILTER_MATCH_LOCAL_ATTR));
 					break;
 				default:
 					LOG.warn("Ignoring element ({}):{} as it isn't supported in this version of xml2csv", uri, localName);
@@ -480,11 +507,13 @@ public void startElement(String uri, String localName, String qName, Attributes
 	 * Adds filters to the mapping configuration.
 	 *
 	 * @param fileNameRegex a regular expression to match against the filename. May be null.
+	 * @param matchLocalFileNameOnly if false then the filter must match the absolute path name, if true then only the file name (no directory
+	 *            information) will be used.
 	 * @throws SAXException If any errors occur whilst adding the filters.
 	 */
-	private void startFileNameFilter(String fileNameRegex) throws SAXException {
+	private void startFileNameFilter(String fileNameRegex, boolean matchLocalFileNameOnly) throws SAXException {
 		try {
-			IInputFilter filter = new FileNameInputFilter(fileNameRegex);
+			IInputFilter filter = new FileNameInputFilter(fileNameRegex, matchLocalFileNameOnly);
 			addFilter(filter);
 		} catch (PatternSyntaxException pse) {
 			throw getException(pse, "Invalid Regular Expression {} specified for input filter.", fileNameRegex);

src/com/locima/xml2csv/inputparser/xml/MappingConfiguration.xsd

@@ -203,6 +203,13 @@
 				</documentation>
 			</annotation>
 		</attribute>
+		<attribute name="matchLocalFileNameOnly" type="boolean" use="optional">
+			<annotation>
+				<documentation>
+					Enter true to match on only the file name (exclude all directory information), false or omitted means that the absolute file name is used.
+				</documentation>
+			</annotation>
+		</attribute>
 	</complexType>
 
 	<complexType name="XPathInputFilter">

testsrc/com/locima/xml2csv/configuration/filter/FilterTests.java

@@ -23,7 +23,8 @@ public class FilterTests {
 	private void checkAllInputFileFilterInputs(IInputFilter filter, boolean expectedResult, String... filenames) {
 		for (String filename : filenames) {
 			File f = new File(filename);
-			assertEquals(expectedResult, filter.include(f));
+			boolean result = filter.include(f);
+			assertEquals("Test for " + f  +" failed",expectedResult, result);
 		}
 	}
 
@@ -35,18 +36,18 @@ private void checkAllXmlInputs(IInputFilter filter, boolean expectedResult, XdmN
 
 	@Test
 	public void fileNameTest() {
-		FileNameInputFilter filter = new FileNameInputFilter("\\.xml$");
+		FileNameInputFilter filter = new FileNameInputFilter("\\.xml$",true);
 		checkAllInputFileFilterInputs(filter, true, "wibble.xml", "test.xml", ".xml");
 		checkAllInputFileFilterInputs(filter, false, "wibble.xml ", "wibble.xml.bak", "wibble.bak");
 	}
 
 	@Test
 	public void nestedTest() {
-		FileNameInputFilter xmlFilter = new FileNameInputFilter("\\.xml$");
-		FileNameInputFilter andyFilter = new FileNameInputFilter("a");
-		xmlFilter.addNestedFilter(andyFilter);
-		FileNameInputFilter tomFilter = new FileNameInputFilter("b");
-		andyFilter.addNestedFilter(tomFilter);
+		FileNameInputFilter xmlFilter = new FileNameInputFilter("\\.xml$", true);
+		FileNameInputFilter aFilter = new FileNameInputFilter("a", true);
+		xmlFilter.addNestedFilter(aFilter);
+		FileNameInputFilter bFilter = new FileNameInputFilter("b", true);
+		aFilter.addNestedFilter(bFilter);
 
 		checkAllInputFileFilterInputs(xmlFilter, false, "a.xml", "b.xml", "c.xml", "ab.bak", "ab.xml2");
 		checkAllInputFileFilterInputs(xmlFilter, true, "ba.xml", "ab.xml", "abba.xml", "ab.xml.bak.xml");

testsrc/com/locima/xml2csv/output/OutputManagerTests.java

@@ -1,8 +1,5 @@
 package com.locima.xml2csv.output;
 
-import static org.junit.Assert.assertEquals;
-import static org.junit.Assert.assertNull;
-
 import java.io.File;
 import java.io.IOException;
 import java.util.Arrays;
@@ -25,7 +22,6 @@
 import com.locima.xml2csv.configuration.MappingList;
 import com.locima.xml2csv.configuration.MultiValueBehaviour;
 import com.locima.xml2csv.configuration.NameFormat;
-import com.locima.xml2csv.util.StringUtil;
 import com.locima.xml2csv.util.XmlUtil;
 
 public class OutputManagerTests {