Merge branch 'fix-570'

marytts-runtime/src/main/java/marytts/modules/PraatTextGridGenerator.java

@@ -23,18 +23,17 @@
 // DOM classes
 import marytts.datatypes.MaryData;
 import marytts.datatypes.MaryDataType;
-import marytts.datatypes.MaryXML;
 import marytts.util.data.text.PraatInterval;
 import marytts.util.data.text.PraatIntervalTier;
 import marytts.util.data.text.PraatTextGrid;
-import marytts.util.dom.NameNodeFilter;
+import marytts.util.dom.DomUtils;
 
 import org.w3c.dom.Document;
 import org.w3c.dom.Element;
-import org.w3c.dom.traversal.DocumentTraversal;
-import org.w3c.dom.traversal.NodeFilter;
 import org.w3c.dom.traversal.NodeIterator;
 
+import static marytts.datatypes.MaryXML.*;
+
 /**
  * Transforms a full MaryXML document into a Praat TextGrid containing various interesting information; in particular, the source
  * units and basenames used in unit selection synthesis
@@ -45,7 +44,7 @@
 public class PraatTextGridGenerator extends InternalModule {
 
 	public PraatTextGridGenerator() {
-		super("Praat TextGrid generator", MaryDataType.AUDIO, MaryDataType.PRAAT_TEXTGRID, null);
+		super("Praat TextGrid generator", MaryDataType.REALISED_ACOUSTPARAMS, MaryDataType.PRAAT_TEXTGRID, null);
 	}
 
 	public MaryData process(MaryData d) throws Exception {
@@ -60,13 +59,9 @@ public MaryData process(MaryData d) throws Exception {
 		Document doc = d.getDocument();
 
 		// initialize various variables:
-		Double xmin = 0.0;
-		Double xmax = 0.0;
-		Double duration;
+		Double duration = 0.0;
 		String phone = null;
 
-		Double sentenceStart = 0.0; // this is cumulatively added to counter sentence-initial timing resets
-
 		// initialize some class variables:
 		PraatIntervalTier phoneTier = new PraatIntervalTier("phones");
 		Double basenameDuration = 0.0;
@@ -79,30 +74,27 @@ public MaryData process(MaryData d) throws Exception {
 		PraatIntervalTier sourceTier = new PraatIntervalTier("sources");
 
 		// prepare to iterate only over the PHONE, SENTENCE, and BOUNDARY nodes in the MaryXML:
-		NodeIterator ni = ((DocumentTraversal) doc).createNodeIterator(doc, NodeFilter.SHOW_ELEMENT, new NameNodeFilter(
-				new String[] { MaryXML.SENTENCE, MaryXML.PHONE, MaryXML.BOUNDARY }), false);
+		NodeIterator ni = DomUtils.createNodeIterator(doc, PHONE, BOUNDARY);
 		Element element;
 
 		// now iterate over these nodes:
 		while ((element = (Element) ni.nextNode()) != null) {
-			String tagName = element.getTagName(); // <s>, <ph>, or <boundary> as specified above
-			if (tagName.equals(MaryXML.SENTENCE)) {
-				sentenceStart = xmax;
-				continue; // goto next node, do not collect $200
-			} else if (tagName.equals(MaryXML.PHONE)) {
+			switch (element.getTagName()) { // <s>, <ph>, or <boundary> as specified above
+			case PHONE:
 				phone = element.getAttribute("p");
-				xmin = xmax;
-				xmax = Double.parseDouble(element.getAttribute("end")) + sentenceStart; // TODO: diphone voices have end in ms!
-				duration = xmax - xmin;
-				xmin = xmax;
-			} else { // boundary
+				duration = Integer.parseInt(element.getAttribute("d")) / 1000.0; // duration is always in ms
+				break;
+			case BOUNDARY:
 				phone = "_"; // TODO: perhaps we should access TargetFeatureComputer.getPauseSymbol() instead
-				try {
+				if (element.hasAttribute("duration")) {
 					duration = Double.parseDouble(element.getAttribute("duration")) / 1000.0; // duration is always in ms
-					xmax += duration;
-				} catch (NumberFormatException nfe) {
-					continue; // HMM voices can have duration-less <boundary/> tags, which can't be processed here
+				} else {
+					duration = 0.0; // HMM voices can have duration-less <boundary/> tags
 				}
+				break;
+			default:
+				logger.error("NodeIterator should not find an element of type " + element.getTagName() + " here!");
+				break;
 			}
 
 			PraatInterval phoneInterval = new PraatInterval(duration, phone);
@@ -182,7 +174,7 @@ public MaryData process(MaryData d) throws Exception {
 		}
 
 		// return raw TextGrid as result:
-		MaryData result = new MaryData(outputType(), d.getLocale());
+		MaryData result = new MaryData(getOutputType(), d.getLocale());
 		result.setPlainText(textGrid.toString());
 		return result;
 	}

marytts-signalproc/src/main/java/marytts/util/data/text/PraatInterval.java

@@ -169,8 +169,13 @@ public void setText(String text) {
 	 * 
 	 * @param duration
 	 *            duration
+	 * @throws IllegalArgumentException
+	 *             if duration is negative
 	 */
-	public void setDuration(double duration) {
+	public void setDuration(double duration) throws IllegalArgumentException {
+		if (Double.isNaN(duration) || duration < 0.0) {
+			throw new IllegalArgumentException("duration must be a non-negative value, but was " + duration);
+		}
 		this.duration = duration;
 	}
 

pom.xml

@@ -237,7 +237,7 @@
 			<dependency>
 				<groupId>org.codehaus.groovy</groupId>
 				<artifactId>groovy-all</artifactId>
-				<version>2.4.6</version>
+				<version>2.4.5</version>
 			</dependency>
 
 			<dependency>

src/site/markdown/download/releasenotes-5.2.md

@@ -37,6 +37,7 @@ Development news
 Fixed Issues/Bugs
 -----------------
 
+* [#570](https://github.com/marytts/marytts/issues/570): Praat TextGrid output is invalid with boundaries (times are not monotonic)
 * [#564](https://github.com/marytts/marytts/pull/564): add missing TOKENS examples
 * [#555](https://github.com/marytts/marytts/issues/555): HMMModel generates malformed XML duration attributes
 * [#531](https://github.com/marytts/marytts/issues/531): java.awt.HeadlessException in (Half)PhoneLabelFeatureAligner

voice-cmu-slt-hsmm/src/test/groovy/marytts/modules/PraatTextGridGeneratorIT.groovy

@@ -0,0 +1,58 @@
+package marytts.modules
+
+import org.junit.*
+
+import groovy.util.logging.Log4j
+
+import marytts.LocalMaryInterface
+import marytts.datatypes.MaryDataType
+
+@Log4j
+class PraatTextGridGeneratorIT {
+
+	def mary
+
+	@Before
+	void setUp() {
+		mary = new LocalMaryInterface()
+		mary.outputType = MaryDataType.PRAAT_TEXTGRID.name
+	}
+
+	@Test
+	void checkBoundaryTimes() {
+		def textGrid = mary.generateText('Foo, bar.\n\nBaz!')
+		def xmin = 0
+		def xmax = 0
+		def newInterval = false
+		textGrid.eachLine { line->
+			switch(line) {
+				case ~/.*intervals \[\d\]:.*/:
+					newInterval = true // begin interval block
+					log.info "begin new interval block: $line"
+					break
+				case { line =~ /.*xmin =.*/ && newInterval }:
+					xmin = Float.parseFloat(line.split().last())
+					log.info "xmin = $xmin"
+					assert xmin == xmax: "interval xmin must be equal to previous interval's xmax"
+					break
+				case { line =~ /.*xmax =.*/ && newInterval }:
+					xmax = Float.parseFloat(line.split().last())
+					log.info "xmax = $xmax"
+					assert xmax > xmin: "interval xmax must be bigger than xmin"
+					break
+				case ~/.*text = .*/:
+					newInterval = false // end interval block
+					log.info "end new interval block: $line"
+					break
+				case ~/.*class = "IntervalTier".*/:
+				// starting new tier; reset times
+					xmin = 0
+					xmax = 0
+					break
+				default:
+				// ignore other lines
+					break
+			}
+		}
+	}
+}