Révision 2279

tmp/org.txm.groovy.core/src/java/org/txm/groovy/core/GroovyScriptedImportEngine.java (revision 2279)
119 119 
			}
120 120 

  		





  121
  121
  
    
			ExecTimer.start();


  		





  122
  
  
    
			Log.info(NLS.bind(GroovyMessages.startingTheP0GroovyImportScript, mainScript.getName())); //$NON-NLS-1$


  		





  
  122
  
    
			//Log.info(NLS.bind(GroovyMessages.startingTheP0GroovyImportScript, mainScript.getName())); //$NON-NLS-1$


  		





  
  123
  
    
			Log.info(NLS.bind(GroovyMessages.startingTheP0importModule, project.getImportModuleName()));


  		





  123
  124
  
    
			gse.run(mainScript.toURI().toURL().toString(), binding); // run the groovy import script


  		





  124
  
  
    
			Log.fine("\nEnd of Groovy import script: " + ExecTimer.stop()); //$NON-NLS-1$


  		





  125
  125
  
    
			monitor.worked(90);


  		





  126
  126
  
    

  		





  127
  127
  
    
			TokenizerClasses.reset(); // restore default tokenizer classes


  		





  ......




  134
  134
  
    
					


  		





  135
  135
  
    
					project.saveParameters(true);


  		





  136
  136
  
    
					


  		





  
  137
  
    
					if (monitor != null) monitor.setTaskName(TXMCoreMessages.common_done);


  		





  
  138
  
    
					


  		





  137
  139
  
    
//					Log.info(GroovyMessages.restartingToolboxSearchengines);


  		





  138
  140
  
    
//					Toolbox.getEngineManager(EngineType.SEARCH).restartEngines();


  		





  139
  
  
    
//					Log.info(TXMCoreMessages.common_done);


  		





  
  141
  
    
					if (project.getDoUpdate()) {


  		





  
  142
  
    
						Log.info(NLS.bind(TXMCoreMessages.corpusImportDoneInP0, ExecTimer.stop()));


  		





  
  143
  
    
					} else {


  		





  
  144
  
    
						Log.info(NLS.bind(TXMCoreMessages.corpusUpdateDoneInP0, ExecTimer.stop()));


  		





  
  145
  
    
					}


  		





  140
  146
  
    
					


  		





  141
  147
  
    
				} else {


  		





  142
  148
  
    
					Log.fine(GroovyMessages.errorImportNotCorrectlyEndedSeeConsoleMessages);


  		












  

    
      tmp/org.txm.groovy.core/src/java/org/txm/groovy/core/messages/GroovyMessages.java	(revision 2279)
    
  






  11
  11
  
    
	public static String startingTheP0GroovyImportScript;


  		





  12
  12
  
    
	public static String errorImportNotCorrectlyEndedSeeConsoleMessages;


  		





  13
  13
  
    
	public static String errorTreeTaggerAnnotationEngineIsNotReady;


  		





  
  14
  
    

  		





  
  15
  
    
	public static String startingTheP0importModule;


  		





  14
  16
  
    
	


  		





  15
  17
  
    
	static {


  		





  16
  18
  
    
		// initialize resource bundle


  		












  

    
      tmp/org.txm.groovy.core/src/java/org/txm/groovy/core/messages/messages.properties	(revision 2279)
    
  






  1
  1
  
    
restartingToolboxSearchengines=Restarting search engine...


  		





  2
  
  
    
startingTheP0GroovyImportScript=Starting the {0} Groovy import script.


  		





  
  2
  
    
startingTheP0GroovyImportScript=Starting the {0} Groovy import script...


  		





  
  3
  
    
startingTheP0importModule=Starting the  "{0}" import module...


  		





  3
  4
  
    
errorImportNotCorrectlyEndedSeeConsoleMessages=Error: import not correctly ended. See previous messages.


  		





  4
  5
  
    
errorTreeTaggerAnnotationEngineIsNotReady=** Error: TreeTagger annotation engine is not ready please check TXM > Advance > NLP > TreeTagger preferences. Aborting.


  		












  

    
      tmp/org.txm.groovy.core/src/java/org/txm/groovy/core/messages/messages_fr.properties	(revision 2279)
    
  






  1
  1
  
    
restartingToolboxSearchengines=Redémarrage des moteurs de recherche...


  		





  2
  
  
    
startingTheP0GroovyImportScript=Démarrage du script d'import Groovy {0}.


  		





  
  2
  
    
startingTheP0GroovyImportScript=Démarrage du script d'import Groovy {0}...


  		





  
  3
  
    
startingTheP0importModule=Démarrage du module d'import "{0}"...


  		





  3
  4
  
    
errorImportNotCorrectlyEndedSeeConsoleMessages=** Erreur: l'import ne s'est pas terminé correctement. Voir les messages précédent.


  		





  4
  5
  
    
errorTreeTaggerAnnotationEngineIsNotReady=** Erreur : l'annotation treetagger n'est pas possible. Veuillez vérifier les préférences de la page TXM > Avancé > TAL > TreeTagger. Abandon.


  		












  

    
      tmp/org.txm.groovy.core/src/groovy/org/txm/scripts/importer/xtz/XTZDefaultPagerStep.groovy	(revision 2279)
    
  






  202
  202
  
    
	private boolean createNextOutput() {


  		





  203
  203
  
    
		wordcount = 0;


  		





  204
  204
  
    
		try {


  		





  205
  
  
    
			def tags = closeMultiWriter();


  		





  
  205
  
    
			def tags = closeMultiWriter()


  		





  206
  206
  
    
			


  		





  207
  207
  
    
			outfile = new File(outdir, txtname+"_"+(++pagecount)+".html")


  		





  208
  
  
    
			pages.add(outfile);


  		





  
  208
  
    
			pages.add(outfile)


  		





  209
  209
  
    
			firstWord = true; // waiting for next word


  		





  210
  210
  
    

  		





  211
  
  
    
			pagedWriter = new StaxStackWriter(outfile, "UTF-8");


  		





  
  211
  
    
			pagedWriter = new StaxStackWriter(outfile, "UTF-8")


  		





  212
  212
  
    

  		





  213
  213
  
    
			//pagedWriter.writeStartDocument()


  		





  214
  214
  
    
			pagedWriter.writeDTD("<!DOCTYPE html>")


  		





  215
  215
  
    
			pagedWriter.writeCharacters("\n")


  		





  216
  
  
    
			pagedWriter.writeStartElement("html");


  		





  
  216
  
    
			pagedWriter.writeStartElement("html")


  		





  217
  217
  
    
			pagedWriter.writeCharacters("\n")


  		





  218
  
  
    
			pagedWriter.writeEmptyElement("meta", ["http-equiv":"Content-Type", "content":"text/html","charset":"UTF-8"]);


  		





  
  218
  
    
			pagedWriter.writeEmptyElement("meta", ["http-equiv":"Content-Type", "content":"text/html","charset":"UTF-8"])


  		





  219
  219
  
    
			for (String css : cssList) {


  		





  220
  
  
    
				pagedWriter.writeEmptyElement("link", ["rel":"stylesheet", "type":"text/css","href":"$css"]);


  		





  
  220
  
    
				pagedWriter.writeEmptyElement("link", ["rel":"stylesheet", "type":"text/css","href":"$css"])


  		





  221
  221
  
    
			}


  		





  222
  
  
    
			pagedWriter.writeStartElement("head");


  		





  
  222
  
    
			pagedWriter.writeStartElement("head")


  		





  223
  223
  
    
			pagedWriter.writeStartElement("title")


  		





  224
  224
  
    
			pagedWriter.writeCharacters(basename.toUpperCase()+" Edition - Page "+pagecount)


  		





  225
  225
  
    
			pagedWriter.writeEndElement(); // </title>


  		





  ......




  228
  228
  
    
			pagedWriter.writeStartElement("body") //<body>


  		





  229
  229
  
    
			pagedWriter.writeStartElement("div", ["class": "txmeditionpage"]) //<div>


  		





  230
  230
  
    
//			println "OPENING: $tags"


  		





  231
  
  
    
			pagedWriter.writeStartElements(tags);


  		





  
  231
  
    
			pagedWriter.writeStartElements(tags)


  		





  232
  232
  
    
			return true;


  		





  233
  233
  
    
		} catch (Exception e) {


  		





  234
  
  
    
			System.out.println(e.getLocalizedMessage());


  		





  
  234
  
    
			System.out.println(e.getLocalizedMessage())


  		





  235
  235
  
    
			e.printStackTrace()


  		





  236
  236
  
    
			return false;


  		





  237
  237
  
    
		}


  		





  ......




  245
  245
  
    
	 */


  		





  246
  246
  
    
	private boolean createOutput() {


  		





  247
  247
  
    
		try {


  		





  248
  
  
    
			return createNextOutput();


  		





  
  248
  
    
			return createNextOutput()


  		





  249
  249
  
    
		} catch (Exception e) {


  		





  250
  
  
    
			System.out.println(e.getLocalizedMessage());


  		





  
  250
  
    
			System.out.println(e.getLocalizedMessage())


  		





  251
  251
  
    
			return false;


  		





  252
  252
  
    
		}


  		





  253
  253
  
    
	}


  		





  ......




  304
  304
  
    
				rend = "";


  		





  305
  305
  
    
				switch (event) {


  		





  306
  306
  
    
					case XMLStreamConstants.START_ELEMENT:


  		





  307
  
  
    
						localname = parser.getLocalName();


  		





  
  307
  
    
						localname = parser.getLocalName()


  		





  308
  308
  
    
						if (outOfTextElements.contains(localname)) {


  		





  309
  309
  
    
							currentOutOfTextElements << localname


  		





  310
  310
  
    
							writeOutOfTextToEditText = true;


  		





  ......




  313
  313
  
    
						}


  		





  314
  314
  
    

  		





  315
  315
  
    
						if (localname == paginationElement) {


  		





  316
  
  
    
							createNextOutput();


  		





  
  316
  
    
							createNextOutput()


  		





  317
  317
  
    
							wordcount=0;


  		





  318
  
  
    
							pagedWriter.write("\n");


  		





  
  318
  
    
							pagedWriter.write("\n")


  		





  319
  319
  
    
							if (getAttributeValue(parser, null,"n") != null) {


  		





  320
  320
  
    
								pagedWriter.writeElement("p", ["class":"txmeditionpb", "align":"center"], getAttributeValue(parser, null,"n"))


  		





  321
  321
  
    
							}


  		





  ......




  331
  331
  
    
									attributes[parser.getAttributeLocalName(i)] = parser.getAttributeValue(i).toString()


  		





  332
  332
  
    
								}


  		





  333
  333
  
    

  		





  334
  
  
    
								pagedWriter.write("\n");


  		





  
  334
  
    
								pagedWriter.write("\n")


  		





  335
  335
  
    
								pagedWriter.writeStartElement("p")


  		





  336
  
  
    
								pagedWriter.writeAttribute("class", rend);


  		





  
  336
  
    
								pagedWriter.writeAttribute("class", rend)


  		





  337
  337
  
    
								if (attributes.containsKey("id")) {


  		





  338
  338
  
    
									pagedWriter.writeElement("h3", attributes["id"])


  		





  339
  339
  
    
								}


  		





  340
  340
  
    

  		





  341
  
  
    
								pagedWriter.writeStartElement("table");


  		





  
  341
  
    
								pagedWriter.writeStartElement("table")


  		





  342
  342
  
    
								for (String k : attributes.keySet()) {


  		





  343
  343
  
    
									if (k == "id") continue;


  		





  344
  344
  
    
									if (k == "rend") continue;


  		





  345
  345
  
    

  		





  346
  
  
    
									pagedWriter.writeStartElement("tr");


  		





  347
  
  
    
									pagedWriter.writeElement("td", k);


  		





  348
  
  
    
									pagedWriter.writeElement("td", attributes[k]);


  		





  349
  
  
    
									pagedWriter.writeEndElement(); //tr


  		





  
  346
  
    
									pagedWriter.writeStartElement("tr")


  		





  
  347
  
    
									pagedWriter.writeElement("td", k)


  		





  
  348
  
    
									pagedWriter.writeElement("td", attributes[k])


  		





  
  349
  
    
									pagedWriter.writeEndElement() //tr


  		





  350
  350
  
    
								}


  		





  351
  351
  
    
								pagedWriter.writeEndElement() // table


  		





  352
  352
  
    
								pagedWriter.writeEndElement() // p


  		





  353
  
  
    
								pagedWriter.writeCharacters("\n");


  		





  
  353
  
    
								pagedWriter.writeCharacters("\n")


  		





  354
  354
  
    
								break;


  		





  355
  355
  
    
							case "ref":


  		





  356
  356
  
    
								pagedWriter.writeStartElement("a")


  		





  357
  
  
    
								pagedWriter.writeAttribute("href", getAttributeValue(parser, null, "target"));


  		





  358
  
  
    
								pagedWriter.writeAttribute("target", "_blank");


  		





  359
  
  
    
								pagedWriter.writeAttribute("class", rend);


  		





  
  357
  
    
								pagedWriter.writeAttribute("href", getAttributeValue(parser, null, "target"))


  		





  
  358
  
    
								pagedWriter.writeAttribute("target", "_blank")


  		





  
  359
  
    
								pagedWriter.writeAttribute("class", rend)


  		





  360
  360
  
    
								break;


  		





  361
  361
  
    
							case "head":


  		





  362
  
  
    
								pagedWriter.write("\n");


  		





  
  362
  
    
								pagedWriter.write("\n")


  		





  363
  363
  
    
								pagedWriter.writeStartElement("h2", ["class":rend])


  		





  364
  364
  
    
								break;


  		





  365
  365
  
    
							case "graphic":


  		





  ......




  383
  383
  
    
								pagedWriter.writeStartElement("td", ["class":rend])


  		





  384
  384
  
    
								break;


  		





  385
  385
  
    
							case "list":


  		





  386
  
  
    
								String type = getAttributeValue(parser, null,"type");


  		





  
  386
  
    
								String type = getAttributeValue(parser, null,"type")


  		





  387
  387
  
    
								if ("unordered" == type) {


  		





  388
  388
  
    
									pagedWriter.writeStartElement("ul", ["class":rend])


  		





  389
  389
  
    
								} else {


  		





  ......




  409
  409
  
    
								break;


  		





  410
  410
  
    
							case "p":


  		





  411
  411
  
    
							//case "lg":


  		





  412
  
  
    
								pagedWriter.write("\n");


  		





  
  412
  
    
								pagedWriter.write("\n")


  		





  413
  413
  
    
								pagedWriter.writeStartElement("p", ["class":rend])


  		





  414
  414
  
    
								break;


  		





  
  415
  
    
							case "sp":


  		





  
  416
  
    
								pagedWriter.writeStartElement("p", ["class":"turn"])


  		





  
  417
  
    
								pagedWriter.writeStartElement("span")


  		





  
  418
  
    
								pagedWriter.writeAttribute("class", "spk")


  		





  
  419
  
    
								pagedWriter.writeCharacters(parser.getAttributeValue(null,"speaker")+": ")


  		





  
  420
  
    
								pagedWriter.writeEndElement() // span@class=spk


  		





  
  421
  
    
								break;


  		





  
  422
  
    
							case "u":


  		





  
  423
  
    
								//pagedWriter.writeStartElement("p", ["class":"u"])


  		





  
  424
  
    
								pagedWriter.writeStartElement("span")


  		





  
  425
  
    
								pagedWriter.writeAttribute("class", "sync")


  		





  
  426
  
    
								pagedWriter.writeCharacters(parser.getAttributeValue(null,"time"))


  		





  
  427
  
    
								//pagedWriter.writeEndElement(); // span@class=spk


  		





  
  428
  
    
								break;


  		





  415
  429
  
    
							case "div":


  		





  416
  430
  
    
							case "div1":


  		





  417
  431
  
    
							case "div2":


  		





  ......




  443
  457
  
    
								interpvalue+=" "+getAttributeValue(parser, null, "type").substring(1)+":"


  		





  444
  458
  
    
								break;


  		





  445
  459
  
    
							case "form":


  		





  446
  
  
    
								wordvalue="";


  		





  447
  
  
    
								interpvalue ="";


  		





  448
  
  
    
								flagform=true;


  		





  
  460
  
    
								wordvalue=""


  		





  
  461
  
    
								interpvalue =""


  		





  
  462
  
    
								flagform=true


  		





  449
  463
  
    
								break;


  		





  450
  464
  
    
							default:


  		





  451
  465
  
    
								if (noteElements.contains(localname)) {


  		





  ......




  466
  480
  
    
						switch (localname) {


  		





  467
  481
  
    
							case "text":


  		





  468
  482
  
    
								break;


  		





  
  483
  
    
							case "u":


  		





  
  484
  
    
								pagedWriter.writeEmptyElement("br")


  		





  
  485
  
    
								break;


  		





  469
  486
  
    
							case "p":


  		





  
  487
  
    
							case "sp":


  		





  470
  488
  
    
							//case "lg":


  		





  471
  489
  
    
								pagedWriter.writeEndElement() // </p>


  		





  472
  
  
    
								pagedWriter.write("\n");


  		





  
  490
  
    
								pagedWriter.write("\n")


  		





  473
  491
  
    
								break;


  		





  474
  492
  
    
							case "div":


  		





  475
  493
  
    
							case "div1":


  		





  ......




  482
  500
  
    
								break;


  		





  483
  501
  
    
							case "head":


  		





  484
  502
  
    
								pagedWriter.writeEndElement() // </h2>


  		





  485
  
  
    
								pagedWriter.write("\n");


  		





  
  503
  
    
								pagedWriter.write("\n")


  		





  486
  504
  
    
								break;


  		





  487
  505
  
    
							case "list":


  		





  488
  
  
    
								pagedWriter.writeEndElement(); // ul or ol


  		





  489
  
  
    
								pagedWriter.write("\n");


  		





  
  506
  
    
								pagedWriter.writeEndElement() // ul or ol


  		





  
  507
  
    
								pagedWriter.write("\n")


  		





  490
  508
  
    
								break


  		





  491
  509
  
    
							case "item":


  		





  492
  
  
    
								pagedWriter.writeEndElement(); // li


  		





  493
  
  
    
								pagedWriter.write("\n");


  		





  
  510
  
    
								pagedWriter.writeEndElement() // li


  		





  
  511
  
    
								pagedWriter.write("\n")


  		





  494
  512
  
    
								break;


  		





  495
  513
  
    
							case "hi":


  		





  496
  
  
    
								pagedWriter.writeEndElement(); // b


  		





  
  514
  
    
								pagedWriter.writeEndElement() // b


  		





  497
  515
  
    
								break;


  		





  498
  516
  
    
							case "emph":


  		





  499
  
  
    
								pagedWriter.writeEndElement(); // i


  		





  
  517
  
    
								pagedWriter.writeEndElement() // i


  		





  500
  518
  
    
								break;


  		





  501
  519
  
    
							case "table":


  		





  502
  
  
    
								pagedWriter.writeEndElement(); // table


  		





  
  520
  
    
								pagedWriter.writeEndElement() // table


  		





  503
  521
  
    
								pagedWriter.write("\n");


  		





  504
  522
  
    
								break;


  		





  505
  523
  
    
							case "row":


  		





  506
  
  
    
								pagedWriter.writeEndElement(); // tr


  		





  
  524
  
    
								pagedWriter.writeEndElement() // tr


  		





  507
  525
  
    
								break;


  		





  508
  526
  
    
							case "cell":


  		





  509
  
  
    
								pagedWriter.writeEndElement(); // td


  		





  
  527
  
    
								pagedWriter.writeEndElement() // td


  		





  510
  528
  
    
								break;


  		





  511
  529
  
    
							case "ref":


  		





  512
  530
  
    
								pagedWriter.writeEndElement() // </a>


  		





  ......




  519
  537
  
    
								break;


  		





  520
  538
  
    
							case wordTag:


  		





  521
  539
  
    
								int l = lastword.length();


  		





  522
  
  
    
								String endOfLastWord = "";


  		





  523
  
  
    
								if (l > 0)


  		





  524
  
  
    
									endOfLastWord = lastword.subSequence(l-1, l);


  		





  525
  
  
    

  		





  526
  
  
    
								if (interpvalue != null)


  		





  527
  
  
    
									interpvalue = interpvalue;


  		





  528
  
  
    

  		





  
  540
  
    
								String endOfLastWord = ""


  		





  
  541
  
    
								if (l > 0) {


  		





  
  542
  
    
									endOfLastWord = lastword.subSequence(l-1, l)


  		





  
  543
  
    
								}


  		





  
  544
  
    
								


  		





  
  545
  
    
								if (interpvalue != null) {


  		





  
  546
  
    
									interpvalue = interpvalue


  		





  
  547
  
    
								}


  		





  529
  548
  
    
								if (NoSpaceBefore.contains(wordvalue) ||


  		





  530
  549
  
    
								NoSpaceAfter.contains(lastword) ||


  		





  531
  550
  
    
								wordvalue.startsWith("-") ||


  		





  532
  551
  
    
								NoSpaceAfter.contains(endOfLastWord)) {


  		





  533
  
  
    
									pagedWriter.writeStartElement("span", ["title":interpvalue, "id":wordid]);


  		





  
  552
  
    
									pagedWriter.writeStartElement("span", ["title":interpvalue, "id":wordid])


  		





  534
  553
  
    
								} else {


  		





  535
  
  
    
									pagedWriter.writeCharacters("\n");


  		





  536
  
  
    
									pagedWriter.writeStartElement("span", ["title":interpvalue, "id":wordid]);


  		





  
  554
  
    
									pagedWriter.writeCharacters("\n")


  		





  
  555
  
    
									pagedWriter.writeStartElement("span", ["title":interpvalue, "id":wordid])


  		





  537
  556
  
    
								}


  		





  538
  557
  
    

  		





  539
  
  
    
								pagedWriter.writeCharacters(wordvalue);


  		





  540
  
  
    
								pagedWriter.writeEndElement();


  		





  
  558
  
    
								pagedWriter.writeCharacters(wordvalue)


  		





  
  559
  
    
								pagedWriter.writeEndElement()


  		





  541
  560
  
    
								//pagedWriter.writeComment("\n")


  		





  542
  561
  
    
								lastword=wordvalue;


  		





  543
  562
  
    
								break;


  		





  ......




  561
  580
  
    
						break;


  		





  562
  581
  
    
					case XMLStreamConstants.CHARACTERS:


  		





  563
  582
  
    
						if (flagform && parser.getText().length() > 0) {


  		





  564
  
  
    
							wordvalue+=(parser.getText());


  		





  565
  
  
    
							if (flagNote == parser.getText().length() > 0)


  		





  566
  
  
    
								noteContent += parser.getText().replace("\n", " ");


  		





  
  583
  
    
							wordvalue+=(parser.getText())


  		





  
  584
  
    
							if (flagNote == parser.getText().length() > 0) {


  		





  
  585
  
    
								noteContent += parser.getText().replace("\n", " ")


  		





  
  586
  
    
							}


  		





  567
  587
  
    
						} else	if (flaginterp && parser.getText().length() > 0) {


  		





  568
  
  
    
							interpvalue+=(parser.getText());


  		





  
  588
  
    
							interpvalue+=(parser.getText())


  		





  569
  589
  
    
						} else if (flagNote == parser.getText().length() > 0) {


  		





  570
  
  
    
							noteContent += parser.getText().replace("\n", " ");


  		





  
  590
  
    
							noteContent += parser.getText().replace("\n", " ")


  		





  571
  591
  
    
						} else if (writeOutOfTextToEditText) {


  		





  572
  592
  
    
							pagedWriter.writeCharacters(parser.getText())


  		





  573
  593
  
    
						}


  		












Formats disponibles :
  Unified diff