/ - Diff - Plateforme TXM - Forge du Centre Blaise Pascal

Révision 3030

     @Field @Option(name="nonPrimarySpeakerRegex", usage="other non primary id of the other turns", widget="String", required=false, def="")
     		String nonPrimarySpeakerRegex
     @Field @Option(name="newSectionMarker", usage="section marker", widget="String", required=false, def="*#")
     		String newSectionMarker
     @Field @Option(name="debug", usage="speaker ID of the primary speaker", widget="Boolean", required=false, def="false")
     		Boolean debug
-...
     gse.runMacro(org.txm.macro.transcription.Vocapia2TranscriberMacro, ["vocapiaDirectory":vocapiaDirectory, "resultDirectory":trsDirectory])
     println "DETECTING MARKED SECTIONS..."
     trsFiles = trsDirectory.listFiles().findAll(){it.getName().toLowerCase().endsWith(".trs")}
     if (trsFiles.size() == 0) {
-...
     	return false
+    }
     File sectionsDirectory = new File(resultDirectory, "sections")
     sectionsDirectory.mkdir()
     gse.runMacro(org.txm.macro.transcription.SegmentTRSInSectionFromMarkerMacro, ["trsDirectory":trsDirectory, "resultDirectory":sectionsDirectory, "newSectionMarker":newSectionMarker])
     println "CREATE THE 'OTHER' TURNS..."
     trsFiles = sectionsDirectory.listFiles().findAll(){it.getName().toLowerCase().endsWith(".trs")}
     if (trsFiles.size() == 0) {
     	println "No XML file found in $sectionsDirectory"
     	return false
+    }
     File otherDirectory = new File(resultDirectory, "otherturns")
     otherDirectory.mkdir()
     ConsoleProgressBar cpb = new ConsoleProgressBar(trsFiles.size())
     cpb = new ConsoleProgressBar(trsFiles.size())
     for (File file : trsFiles) {
     	cpb.tick()
-...
+    }
     cpb.done()
     println "NORMALIZING LOCUTORS..."
     trsFiles = otherDirectory.listFiles().findAll(){it.getName().toLowerCase().endsWith(".trs")}
     if (trsFiles.size() == 0) {
     	println "No XML file found in $otherDirectory"
     	return false
+    }
     // FIX LOCUTORS
     println "NORMALIZING LOCUTORS..."
     cpb = new ConsoleProgressBar(trsFiles.size())
     for (File file : trsFiles) {
     	cpb.tick()

     @Field @Option(name="trsDirectory", usage="Dossier qui contient les fichiers TRS", widget="Folder", required=true, def="")
     File trsDirectory;
     @Field @Option(name="resultDirectory", usage="Dossier résultat TRS", widget="Folder", required=true, def="")
     File resultDirectory;
     @Field @Option(name="newSectionMarker", usage="The marker, spaces included", widget="String", required=true, def=" *#")
     def newSectionMarker
     if (!ParametersDialog.open(this)) return;
     if (!trsDirectory.exists()) {
-...
     	return
+    }
     File outputDirectory = new File(trsDirectory, "out")
     println "Writing result to $outputDirectory..."
     println "Writing result to $resultDirectory..."
     dateTimeFormatter = DateTimeFormatter.ISO_LOCAL_TIME
     def strTotime(def str) {
-...
+    }
     try {
     	def trsFiles = trsDirectory.list(new FilenameFilter() { public boolean accept(File dir, String name) { return name.toLowerCase().endsWith(".trs"); } });
     	def trsFiles = trsDirectory.listFiles().findAll() { it.getName().toLowerCase().endsWith(".trs") }
     	if (trsFiles.size() == 0) {
     		println "No TRS file to process in $trsDirectory"
     		return;
+    	}
     	ConsoleProgressBar cpb = new ConsoleProgressBar(sectionGroupsToInsert.keySet().size())
     	ConsoleProgressBar cpb = new ConsoleProgressBar(trsFiles.size())
     	for (File trsFile : trsFiles) {
     		cpb.tick()
-...
     		def trsEpisode = trsEpisodes[0]
     		def trsSections =  trs.Episode.Section // 1
     		if (trsSections.size() > 1) {
     			println "multiple Section node in $trsFile"
     			println "Error: multiple Section node in $trsFile"
     			continue
+    		}
     		def trsSection = trsSections[0]
-...
     		def turns = trsSection.Turn
     		def newSections = []
     		def iSection = 0;
     		def currentSection = null
     		def currentNode = null
     		def previousSection = trsSection
     		def currentSection = trsSection
     		for (def turn : turns) {
     		for (int iTurn = 0 ; iTurn < turns.size() ; iTurn++) {
     			def turn = turns[iTurn]
     			def start = Float.parseFloat(turn.@startTime)
     			def end = Float.parseFloat(turn.@endTime)
     			def found = null
     			for (int i = iSection ; i < sections.size() ; i++) {
     				if (end < sections[i][0]) { // Turn is before section
     				} else if (sections[i][1] < start) { // Turn is before section
     				} else {
     					found = sections[i]
     					iSection = i
     					break // stop searching and set iSection to accelerate next search
+    				}
+    			}
     			if (found == null) {
     				if (currentSection != null || currentNode == null) {
     					currentNode = new Node(trsEpisode, "Section", ["type":"Sujet non synchronisé", "startTime":turn.@startTime, "endTime":"", "synchronized":"false"] )
     					currentSection = null;
+    				}
     			} else {
     				if (found != currentSection) {
     					if (currentNode != null && currentNode.@synchronized == "false") {
     						def tmp = currentNode.Turn
     						currentNode.@endTime = tmp[-1].@endTime
     			String content = turn.text().trim()
     			println "TURN: "+content
     			if (content.startsWith(newSectionMarker)) {
     				println "NEW SECTION !"
     				previousSection = currentSection
     				currentSection = new Node(trsEpisode, "Section", new LinkedHashMap(["type":newSectionMarker, "startTime":turn.@startTime, "endTime":previousSection.attributes()["endTime"]]))
     				previousSection.attributes()["endTime"] = turn.@startTime
     				turns[iTurn-1].attributes()["endTime"] = turn.@startTime
     					def children = turn.children()
     					for (int i = 0 ; i < children.size() ; i++) {
     						def node = children[i]
     						if (node instanceof String) {
     							println "FIXING: $node "
     							children.remove(i)
     							i--
     						} else if (node instanceof groovy.util.Node && node.name() == "w" && node.text().contains(newSectionMarker)) {
     							println "FIXING: w $node "
     							children.remove(node)
+    						}
+    					}
     					currentSection = found
     					currentNode = new Node(trsEpisode, "Section", currentSection[2])
+    				}
+    			}
     			trsSection.remove(turn)
     			currentNode.append(turn)
     			currentSection.append(turn)
+    		}
     		//remove the initial section
     		trsEpisode.remove(trsSection)
     		outputDirectory.mkdir()
     		File outfile = new File(outputDirectory, trsFile.getName())
     		resultDirectory.mkdir()
     		File outfile = new File(resultDirectory, trsFile.getName())
     		outfile.withWriter("UTF-8") { writer ->
     			writer.write('<?xml version="1.0" encoding="UTF-8"?>\n<!DOCTYPE Trans SYSTEM "trans-14.dtd">\n')
     			def printer = new groovy.util.XmlNodePrinter(new PrintWriter(writer))
-...
+    		}
+    	}
     	cpb.done()
     	reader.close()
     	println "Done."
     } catch(Exception e) {

     		def turninfos = new LinkedHashMap()
     		boolean other = false;
     		String word = ""
     		String duration = "0.0"
     		try {
     			for (int event = parser.next(); event != XMLStreamConstants.END_DOCUMENT; event = parser.next()) {
-...
     								writer.writeEndElement()
     								writer.writeCharacters("\n")
     								break;
     							case "Channel":
     								duration = parser.getAttributeValue(null, "sigdur")
     								break;
     							case "SegmentList":
     								writer.writeStartElement("Episode")
     							//<Section type="report" startTime="0" endTime="3617.593">
     								writer.writeStartElement("Section")
     								writer.writeAttribute("startTime", "0.0")
     								writer.writeAttribute("endTime", duration)
     								for (int i = 0 ; i < parser.getAttributeCount() ; i++) {
     									writer.writeAttribute(parser.getAttributeLocalName(i), parser.getAttributeValue(i))
+    								}

Formats disponibles : Unified diff

Laboratoire ICAR » Plateforme TXM

Révision 3030