Laboratoire ICAR » Plateforme TXM

package org.txm.scripts.importer

import java.io.File;

import javax.xml.stream.XMLStreamConstants;

import javax.xml.stream.XMLOutputFactory;

import javax.xml.stream.XMLStreamWriter;

import java.net.URL;

import javax.xml.stream.*;

import org.txm.importer.filters.*;

import org.txm.importer.PersonalNamespaceContext;

import org.txm.importer.StaxIdentityParser

/**

 * add 2 attributes per element to encode the distance to the previous milestone and next milestone

 * TODO: only the first attribute is implemented

 * 

 * @author mdecorde

 *

 */

class MileStoneProjection extends StaxIdentityParser {

        String wordTag

        String mileStoneTag

        String startTag

        boolean start = false

        int mileStoneDistance = 0

        String mileStoneID = ""

        def milestonesLength = []

        int milestonesCounter = 0

        boolean secondPass = false

        String msIdAttributeName

        String msStartAttributeName

        String msEndAttributeName

        public MileStoneProjection(File inputFile, String startTag, String wordTag, String mileStoneTag) {

                super(inputFile)

                this.wordTag = wordTag

                this.mileStoneTag = mileStoneTag

                mileStoneID = mileStoneTag+"_0"

                this.startTag = startTag

                this.start = false;

                msIdAttributeName = mileStoneTag+"id";

                msStartAttributeName = mileStoneTag+"start";

                msEndAttributeName = mileStoneTag+"end";

                fetchMilestoneLengths();

                start = false // reset

        }

        public void fetchMilestoneLengths() {

                def inputData = inputurl.openStream();

                def factory = XMLInputFactory.newInstance();

                def parser = factory.createXMLStreamReader(inputData);

                for (int event = parser.next(); event != XMLStreamConstants.END_DOCUMENT; event = parser.next()) {

                        switch (event) {

                                case XMLStreamConstants.START_ELEMENT:

                                        localname = parser.getLocalName()

                                        if (start && localname.equals(mileStoneTag)) {

                                                milestonesLength << mileStoneDistance;

                                                mileStoneDistance = 0

                                                milestonesCounter++;

                                        } else if (start && localname.equals(wordTag)) {

                                                mileStoneDistance++

                                        } else if (localname.equals(startTag)) {

                                                start = true

                                        }

                                        break;

                        }

                }

                milestonesLength[milestonesCounter] = mileStoneDistance;

                if (parser != null) parser.close();

                if (inputData != null) inputData.close();

                milestonesCounter = 0;

                //println milestonesLength

        }

        public void processStartElement() {

                super.processStartElement();

                if (start && localname.equals(mileStoneTag)) {

                        mileStoneDistance = 0

                        mileStoneID = parser.getAttributeValue(null, "id")

                        if (mileStoneID == null) mileStoneID = parser.getAttributeValue("xml", "id");

                        if (mileStoneID == null) mileStoneID = "0";

                        milestonesCounter++;

                } else if (start && localname.equals(wordTag)) {

                        // println "end of $milestonesCounter len="+milestonesLength[milestonesCounter]+" dist="+mileStoneDistance

                        writer.writeAttribute(msEndAttributeName, Integer.toString((milestonesLength[milestonesCounter] - mileStoneDistance - 1)))

                        writer.writeAttribute(msStartAttributeName, Integer.toString(mileStoneDistance))

                        writer.writeAttribute(msIdAttributeName, mileStoneID)

                        mileStoneDistance++

                } else if (localname.equals(startTag)) {

                        start = true

                }

        }

        public static void main(String[] args) {

                File inputFile = new File(System.getProperty("user.home"), "TXM-0.8.0-dev/corpora/XTZMILESTONES/tokenized/test.xml")

                File outputFile = new File(System.getProperty("user.home"), "TXM-0.8.0-dev/corpora/XTZMILESTONES/tokenized/result.xml")

                MileStoneProjection msp = new MileStoneProjection(inputFile, "text", "w", "lb");

                println "Sucess: "+msp.process(outputFile)

        }

}