Laboratoire ICAR » Plateforme TXM

package org.txm.macro.nlp;

        import org.kohsuke.args4j.*

        import groovy.transform.Field

        import java.nio.charset.Charset

        import org.txm.rcpapplication.swt.widget.parameters.*

        import org.txm.utils.*

        import javax.xml.stream.*

        @Field @Option(name="inputDirectory", usage="TXT directory", widget="Folder", required=true, def="dir")

        File inputDirectory

        @Field @Option(name="encoding", usage="File encoding", widget="String", required=false, def="UTF-8")

        String encoding

        @Field @Option(name="debug", usage="Debug mode", widget="Boolean", required=false, def="false")

        Boolean debug

        if (!ParametersDialog.open(this)) return

        encoding = encoding.trim()

        outputDirectory = new File(inputDirectory, "xml")

        outputDirectory.mkdir()

        println "Processing: "+inputDirectory

        XMLOutputFactory factory = XMLOutputFactory.newInstance()

        def files = inputDirectory.listFiles()

        if (files == null || files.length == 0) {

                println "Error: no file to process in $inputDirectory"

                return false;

        }

        for (File inputfile : files.sort()) {

                if (inputfile.isDirectory() || inputfile.isHidden() || !inputfile.getName().endsWith(".tt")) continue // ignore

                println " file: "+inputfile

                name = inputfile.getName()

                idx = name.lastIndexOf(".")

                if (idx > 0) name = name.substring(0, idx)

                outputfile = new File(outputDirectory, name+".xml")

                Writer output = new OutputStreamWriter(new FileOutputStream(outputfile) , "UTF-8")

                XMLStreamWriter writer = factory.createXMLStreamWriter(output)

                writer.writeStartDocument("UTF-8","1.0")

                writer.writeCharacters("\n")

                writer.writeStartElement("text")

                writer.writeCharacters("\n")

                inSent = 0

                nLine = 0

                inputfile.eachLine(encoding) { line ->

                        nLine++

                        if (debug) {

                                println nLine+": "+line

                        }

                        def sent = false

                def sent_pattern = ~/^([^\t]+)\t(SENT)\t([^\t]+)$/

                def sent_m = (line =~ sent_pattern)

                if (sent_m) sent = true

                        if (sent && inSent) {

                                writer.writeStartElement("w")

                                writer.writeAttribute("frpos", sent_m[0][2])

                                writer.writeAttribute("frlemma", sent_m[0][3])

                                writer.writeCharacters(sent_m[0][3])

                                writer.writeEndElement() // close w                

                                writer.writeComment("\n")

                                writer.writeEndElement() // close s

                                writer.writeEndElement() // close p (sent)

                                writer.writeCharacters("\n")

                                inSent=0

                        } else {

                                if (!inSent) {

                                        writer.writeStartElement("p")

                                        writer.writeStartElement("s")

                                        inSent=1

                                }

                                def w_pattern = ~/^([^\t]+)\t([^\t]+)\t([^\t]+)$/

                        def w_m = (line =~ w_pattern)

                        if (w_m.size() > 0) {

                                writer.writeStartElement("w")

                                        writer.writeAttribute("frpos", w_m[0][2])

                                        writer.writeAttribute("frlemma", w_m[0][3])

                                        writer.writeCharacters(w_m[0][1])

                                        writer.writeEndElement() // close w

                                        writer.writeCharacters("\n")

                                } else {

                                    writer.writeStartElement("w")

                                        writer.writeAttribute("frpos", "NAM")

                                        writer.writeAttribute("frlemma", "<unknown>")

                                        writer.writeCharacters(line)

                                        writer.writeEndElement() // close w

                                        writer.writeCharacters("\n")

                                        println "** TT2XMLInDirectory: unknown [word] line pattern, "+nLine+": <"+line+">."

                                }

                        }

                }

                if (inSent) {

                        writer.writeEndElement() // s

                        writer.writeEndElement() // p

                        writer.writeCharacters("\n")

                }

                writer.writeEndElement()

                writer.writeCharacters("\n")

                writer.close()

                output.close()

        }