Laboratoire ICAR » Plateforme TXM

package org.txm.scripts.importer

import java.io.Writer;

import java.io.File;

import javax.xml.stream.*;

import java.net.URL;

import org.txm.importer.filters.*;

class SplitBy 

{

        XMLOutputFactory factory = XMLOutputFactory.newInstance();

        FileOutputStream output;

        XMLStreamWriter writer;

        def parser, inputData;

        ArrayList<String> newfiles = [];

        public SplitBy(File xmlfile)

        {

                try {

                        inputData = xmlfile.toURI().toURL().openStream();

                        def inputFactory = XMLInputFactory.newInstance();

                        parser = inputFactory.createXMLStreamReader(inputData);

                } catch (XMLStreamException ex) {

                        System.out.println(ex);

                }catch (IOException ex) {

                        System.out.println("IOException while parsing ");

                }

        }

        boolean inby = false;

        public boolean process(File outdir, String by, String idAttribute)

        {

                outdir.mkdirs();

                try

                {

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

                                switch(event)

                                {

                                        case XMLStreamConstants.START_ELEMENT:

                                                if(parser.getLocalName() == by)

                                                {

                                                        inby = true;

                                                        if(writer != null)

                                                        {

                                                                writer.close();

                                                                output.close();

                                                        }

                                                        String filename = "";

                                                        for(int i = 0 ; i < parser.getAttributeCount(); i++)

                                                                if(parser.getAttributeLocalName(i) == idAttribute)

                                                                {

                                                                        filename = parser.getAttributeValue(i);

                                                                        break;

                                                                }

                                                        File newfile = new File(outdir,filename+".xml")

                                                        if(files.contains(newfile))

                                                        {

                                                                println "Item declared twice "+filename+" at location "+parser.getLocation();

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

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

                                                                return false;

                                                        }

                                                        files.add(newfile);

                                                        output = new FileOutputStream(newfile)

                                                        writer = factory.createXMLStreamWriter(output, "UTF-8");//create a new file

                                                }

                                                if(writer != null)

                                                {

                                                        writeEvent(event);

                                                        if(parser.getLocalName() == "TEI")

                                                        { // write namespaces

                                                                writer.writeNamespace("tei", "http://www.tei-c.org/ns/1.0")

                                                                writer.writeNamespace("txm", "http://textometrie.org/1.0")

                                                        }

                                                }

                                                break;

                                        case XMLStreamConstants.END_ELEMENT:

                                                if(inby)

                                                {

                                                        if(writer != null)

                                                        {

                                                                writer.writeEndElement();

                                                        }

                                                }

                                                if(parser.getLocalName() == by)

                                                {

                                                        inby = false; // write the end element $by before

                                                }

                                                break;

                                        case XMLStreamConstants.CHARACTERS:

                                                if(writer != null)

                                                {

                                                        writer.writeCharacters(parser.getText());

                                                }

                                                break;

                                }

                        }

                        if(writer != null)

                                writer.close();

                        if(output != null)

                                output.close();

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

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

                        return true;

                }

                catch(Exception e){

                        println "XML Error at location: "+parser.getLocation()

                        e.printStackTrace()

                        if (writer != null)

                                writer.close()

                                output.close()

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

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

                                return false

                        }

        }

        /**

         * write the current event.

         *

         * @param event the stax event

         */

        private void writeEvent(int event)

        {

                def prefix = parser.getPrefix();

                if (event == XMLStreamConstants.START_ELEMENT ) 

                {

                        def localname = parser.getLocalName(); // write element

                        if (prefix != null && prefix.length() > 0)

                                writer.writeStartElement(prefix+":"+localname);

                        else

                                writer.writeStartElement(localname);

                        for(int i = 0 ; i < parser.getNamespaceCount() ; i++) //write namespaces

                        {

                                writer.writeNamespace(parser.getNamespacePrefix(i), parser.getNamespaceURI(i));

                        }

                        String attrprefix; // write attributes

                        for (int i = 0 ; i < parser.getAttributeCount() ; i++)

                        {

                                attrprefix = parser.getAttributePrefix(i);

                                if (attrprefix != null & attrprefix.length() > 0)

                                        writer.writeAttribute(attrprefix+":"+parser.getAttributeLocalName(i), parser.getAttributeValue(i))

                                else

                                        writer.writeAttribute(parser.getAttributeLocalName(i), parser.getAttributeValue(i))

                        }

                        //writer.writeCharacters("\n");

                }

                else if (event == XMLStreamConstants.END_ELEMENT)

                {

                        writer.writeEndElement();

                        //writer.writeCharacters("\n");

                }

                else if (event == XMLStreamConstants.CHARACTERS)

                {

                        writer.writeCharacters(parser.getText());

                }

                else if (event == XMLStreamConstants.COMMENT)

                {

                        writer.writeComment(parser.getText());

                }

        }

        public ArrayList<File> getFiles()

        {

                return newfiles;

        }

        public static void main(String[] args)

        {

                File xmlfile = new File("C:\\Documents and Settings\\mdecorde\\Bureau\\MISAT2011\\corpus\\tiempo\\ELTIEMPO.2002-2010.MIN.xml");

                File okfile = new File("C:\\Documents and Settings\\mdecorde\\Bureau\\MISAT2011\\corpus\\tiempo\\");

                File outdir = new File("C:\\Documents and Settings\\mdecorde\\Bureau\\MISAT2011\\corpus\\tiempo\\tiempo");

                outdir.deleteDir();

                outdir.mkdir();

                //                Reader reader = new InputStreamReader(new FileInputStream(xmlfile) , "ISO-8859-1");

                //                Writer writer = new OutputStreamWriter(new FileOutputStream(okfile) , "UTF-8");

                //                String line = reader.readLine();

                //                while(line != null)

                //                {

                //                        line = line.replaceAll(' ([a-zA-Z0-9]+)="(.*)""', ' $1="$2"')

                //                        writer.write(line+"\n");

                //                        line = reader.readLine();

                //                }

                //                reader.close();

                //                writer.close();

                //                

                //                

                ////                if(!ValidateXml.test(okfile))

                ////                {

                ////                        println "XML not valid";

                ////                        return;

                ////                }

                def splitter = new SplitBy(xmlfile);

                if(splitter.process(outdir, "texte", "id"))

                        println "success";

                else

                        println "failed !"

        }

}