Laboratoire ICAR » Plateforme TXM

	 * Warning: at this point we don't know if all parameters are set -> you should avoid NPE

	 * 

	// FIXME: SJ: this method must become abstract in TXM 0.8.1 and implemented by subclasses

package org.txm.importer;

import java.io.File;

import java.io.IOException;

import java.io.InputStream;

import java.net.MalformedURLException;

import java.util.ArrayList;

import java.util.HashMap;

import java.util.HashSet;

import javax.xml.stream.*;

import org.txm.utils.io.IOUtils;

/**

 * Read an XML file and find out : XML elements, their attributes and recursivity level

 * names are lowercases

 * 

 * @author mdecorde

 *

 */

public class SAttributesListener {

	public HashMap<String,HashSet<String>> structs = new HashMap<String, HashSet<String>>();

	public HashSet<String> anatypes = new HashSet<String>();

	public HashMap<String, Integer> structsCountProf = new HashMap<String, Integer>();

	public HashMap<String, Integer> structsMaxProf = new HashMap<String, Integer>();

	private String structPath = "/";

	private XMLStreamReader parser;

	SAttributesListener() {

		structs = new HashMap<String, HashSet<String>>();

		structsCountProf = new HashMap<String, Integer>();

		structsMaxProf = new HashMap<String, Integer>();

		structPath = "/";

		anatypes = new HashSet<String>(); // store scanned word attributes

	}

	SAttributesListener(XMLStreamReader parser) {

		this();

		this.parser = parser;

	}

	public void appendResultsTo(SAttributesListener another) {

		structs = another.structs;

		structsCountProf = another.structsCountProf;

		structsMaxProf = another.structsMaxProf;

		anatypes = another.anatypes; // store scanned word attributes

	}

	public void start(XMLStreamReader parser) {

		this.parser = parser;

	}

	String W = "w";

	String ANA = "ana";

	String FORM = "form";

	/**

	 * Call this method for each START_ELEMENT stax event

	 * @param localname the element localname

	 */

	public void startElement(String localname) {

		localname = localname.toLowerCase();

		//String localname = parser.getLocalName();

		if(localname.equals(W)) return;

		if(localname.equals(ANA)) return;

		if(localname.equals(FORM)) return;

		structPath += localname+"/";

		//println "add: "+structPath

		HashSet<String> attrs = structs.get(localname);

		if (!structs.containsKey(localname)) {

			attrs = new HashSet<String>();

			structs.put(localname, attrs);

			structsCountProf.put(localname, 0);

			structsMaxProf.put(localname, 0);

		} //else {

		// get structure recursion

		int prof = structsCountProf.get(localname)+1;

		structsCountProf.put(localname, prof);

		if (structsMaxProf.get(localname) < prof) {

			structsMaxProf.put(localname, prof);

		}

		// get the structure attributes

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

			attrs.add(parser.getAttributeLocalName(i).toLowerCase());

		}

	}

	/**

	 * Call this method for each END_ELEMENT stax event

	 * @param localname the element localname

	 */

	public void endElement(String localname) {

		localname = localname.toLowerCase();

		//String localname = parser.getLocalName();

		if(localname.equals(W)) return;

		if(localname.equals(ANA)) return;

		if(localname.equals(FORM)) return;

		if (structPath.length() > 1) {

			int idx = structPath.lastIndexOf("/");

			if (idx > 0) {

				structPath = structPath.substring(0, idx);

				//println "end of $localname "+(structsCountProf.get(localname))

				//if (structsCountProf.get(localname) != null)

				structsCountProf.put(localname, structsCountProf.get(localname)-1);

			}

			//println "pop: "+structPath

		}

	}

//	boolean firstGetStructs = true;

	public HashMap<String,HashSet<String>> getStructs() {

		if (structsCountProf.containsKey("div")) {

			if (structsCountProf.get("div") > 0)

				structs.remove("div1");

			if (structsCountProf.get("div") > 1)

				structs.remove("div2");

			if (structsCountProf.get("div") > 2)

				structs.remove("div3");

			if (structsCountProf.get("div") > 3)

				structs.remove("div4");

			if (structsCountProf.get("div") > 4)

				structs.remove("div5");

			if (structsCountProf.get("div") > 5)

				structs.remove("div6");

		}

//		if (firstGetStructs) {

//			firstGetStructs = false;

		// fix min&maj names for CQP

			ArrayList<String> keys = new ArrayList<String>();

			keys.addAll(structs.keySet());

			for (String key : keys) {

				HashSet<String> value = structs.get(key);

				structs.remove(key);

				structs.put(key.toLowerCase(), value);

			}

//		}

		return structs;

	}

	boolean firstGetstructsCountProf = true;

	public HashMap<String, Integer> getProfs() {

		//		if (firstGetstructsCountProf) {

		//			firstGetstructsCountProf = false;

		//			def keys = []

		//			keys.addAll(structsCountProf.keySet());

		//			for( String key : keys) {

		//				def value = structsCountProf.get(key);

		//				structsCountProf.remove(key)

		//				structsCountProf.put(key.toLowerCase(), value);

		//			}

		//		}

		HashMap<String, Integer> clone = new HashMap<String, Integer>();

		for (String key : structsMaxProf.keySet()) {

			if (structsMaxProf.get(key) > 0)

				clone.put(key, structsMaxProf.get(key)-1);

			else

				clone.put(key, 0);

		}

		return clone;

	}

	public void initialize(ArrayList<String> pattributes, HashMap<String, HashSet<String>> sAttributesMap, HashMap<String, Integer> sAttributesProfs) {

		this.anatypes.addAll(pattributes);

		for (String s : sAttributesMap.keySet()) {

			this.structsMaxProf.put(s, sAttributesProfs.get(s));

			this.structsCountProf.put(s, 0);

			this.structs.put(s, sAttributesMap.get(s));

		}

	}

	public HashSet<String> getAnatypes() {

		return anatypes;

	}

//	public SAttributesListener scanFile(File xmlFile) throws MalformedURLException, IOException, XMLStreamException {

//		return scanFile(xmlFile, this);

//	}

	public void setParser(XMLStreamReader parser) {

		this.parser = parser;

	}

	/**

	 * Merge results in the parentListener

	 * 

	 * @param xmlFile

	 * @param parentListener results are appended to the parentListener if any

	 * @return

	 * @throws IOException 

	 * @throws MalformedURLException 

	 * @throws XMLStreamException 

	 */

	public SAttributesListener scanFile(File xmlFile) throws MalformedURLException, IOException, XMLStreamException {

		boolean startText = false;

		boolean startWord = false;

		InputStream inputData = xmlFile.toURI().toURL().openStream();

		XMLInputFactory factory = XMLInputFactory.newInstance();

		XMLStreamReader parser = factory.createXMLStreamReader(inputData);

//		SAttributesListener listener;

//		if (parentListener != null) {

//			listener = parentListener;

//			listener.setParser(parser);

//		} else {

//			listener = new SAttributesListener(parser);

//		}

		String TEXT = "text";

		String ANA = "ana";

		String TYPE = "type";

		//HashSet<String> types = new HashSet<String>();

		this.setParser(parser);

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

			if (event == XMLStreamConstants.START_ELEMENT) { // start elem

				if (TEXT.equals(parser.getLocalName())) startText = true;

				if (startText) this.startElement(parser.getLocalName());

				if (this.W.equals(parser.getLocalName())) {

					startWord = true;

				} else if (startWord && ANA.equals(parser.getLocalName())) { // ana elem

					for (int i = 0 ; i < parser.getAttributeCount(); i++) { // find @type

						if (TYPE.equals(parser.getAttributeLocalName(i))) { // @type

							this.anatypes.add(parser.getAttributeValue(i).substring(1)); //remove the #

							break;

						}

					}

				}

			} else if (event == XMLStreamConstants.END_ELEMENT) { // end elem

				if (startText) this.endElement(parser.getLocalName());

				if (TEXT.equals(parser.getLocalName())) startText = false;

				if (this.W.equals(parser.getLocalName())) {

					startWord = false;

				}

			}

		}

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

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

		return this;

	}

	/**

	 * scan the XML files of a directory to list the structures with their properties and levels. Also list the word properties

	 * @param xmlDirectory

	 * @param wordTag

	 * @return

	 * @throws XMLStreamException 

	 * @throws IOException 

	 * @throws MalformedURLException 

	 */

	public static SAttributesListener scanFiles(File xmlDirectory, String wordTag) throws MalformedURLException, IOException, XMLStreamException {

		SAttributesListener listener = new SAttributesListener();

		listener.W = wordTag;

		for (File xmlFile : xmlDirectory.listFiles(IOUtils.HIDDENFILE_FILTER)) {

			if (xmlFile.isFile() && !xmlFile.isHidden() && xmlFile.getName().toLowerCase().endsWith(".xml")) {

				listener.scanFile(xmlFile); // results saved in 'listener' data

				//				println "LISTENER RESULT with ${xmlFile.getName()}: "+listener

				//				println " prof: "+listener.getStructs()

				//				println " prof: "+listener.getProfs()

				//				println " path: "+listener.structPath

			}

		}

		return listener;

	}

}

		registryDirectory = new File(module.getBinaryDirectory(), "registry");

		} 

		cqpDirectory.mkdir();