Laboratoire ICAR » Plateforme TXM

		isSuccessFul = filesToProcess.size() > 0

			new File(module.getBinaryDirectory(), "tokenized").deleteDir()

			new File(module.getBinaryDirectory(), "src").deleteDir()

			new File(module.getBinaryDirectory(), "split").deleteDir()

		if (!doCWBEncodeStep()) return;

		if (!doCWBMakeAllStep()) return;

			new File(module.getBinaryDirectory(), "cqp").deleteDir()

		}

		if (cleanDirectories) {

			new File(module.getBinaryDirectory(), "treetagger").deleteDir()

			new File(module.getBinaryDirectory(), "annotations").deleteDir()

		}

			ex.printStackTrace();

			return false;

		}

		return true;

	}

		} catch (Exception ex) { System.err.println(ex); return false;}

		return true;

	}

	 * @param basename the basename

	 * @return true, if successful

	 */

	{

		String rootDir = srcDir.getAbsolutePath()+"/"

		// scanning directory brut/*.cnr

			} else print "."

		}

		println ""

		return true;

	}

if (MONITOR != null && MONITOR.isCanceled()) { return MONITOR.done(); }

println "-- IMPORTER - Reading source files"

def imp = new importer();

	println "import process stopped";

	return;

}

	def engine = Toolbox.getEngineManager(EngineType.ANNOTATION).getEngine("TreeTagger")

	if (engine.processDirectory(txmDir, binDir, ["lang":model])) {

		annotate_status = true;

	}

}

imp.doTokenize(doTokenizeStep) // change this, to not tokenize xml

imp.setStopIfMalformed(stopIfMalformed)

	println "import process stopped"

	return

}

	def engine = Toolbox.getEngineManager(EngineType.ANNOTATION).getEngine("TreeTagger")

	if (engine.processDirectory(txmDir, binDir, ["lang":model])) {

		annotationSuccess = true

	}

}

			cwbMa.run(corpusname, outDir + "/registry");

		} catch (Exception ex) {System.out.println(ex); return false;}

		return true;

	}

	 * @param basename the basename

	 * @return true, if successful

	 */

	{

		new File(binDir,"tokenized").deleteDir()

		new File(binDir,"tokenized").mkdir()

		cpb.done()

		okfiles = txmDir.listFiles()

		return okfiles != null && okfiles.size() > 0;

	}

 */

class importer {

	{

		new File(binDir, "ptokenized").deleteDir();

		new File(binDir, "ptokenized").mkdir();

			}

		}

		println("")

		return true;

	}

			}

			cwbMa.run(corpusname, outDir + "/registry");

		} catch (Exception ex) {System.out.println(ex);return false;}

		return true;

	}

if (MONITOR != null && MONITOR.isCanceled()) { return MONITOR.done(); }

println "-- IMPORTER - Reading source files"

def imp = new importer()

	println "import process stopped";

	return;

}

	def engine = Toolbox.getEngineManager(EngineType.ANNOTATION).getEngine("TreeTagger")

	if (engine.processDirectory(txmDir, binDir, ["lang":model])) {

		annotate_status = true;

	}

}

def imp = new importer();

imp.doValidation(true) // change this to not validate xml

imp.doTokenize(true) // change this, to not tokenize xml

	println "import process stopped";

	return;

}

	def engine = Toolbox.getEngineManager(EngineType.ANNOTATION).getEngine("TreeTagger")

	if (engine.processDirectory(txmDir, binDir, ["lang":model])) {

		annotationSuccess = true;

	}

}

if (MONITOR != null && MONITOR.isCanceled()) { return MONITOR.done(); }

if (MONITOR != null) MONITOR.worked(20, "IMPORTER - Reading source files");

def imp = new importer()

def corpusIDS = imp.getCorpusIDS()

if (MONITOR != null) MONITOR.worked(20, "ANNOTATE - Running NLP tools");

	def engine = Toolbox.getEngineManager(EngineType.ANNOTATION).getEngine("TreeTagger")

	if (engine.processDirectory(txmDir, binDir, ["lang":model])) {

		annotationSuccess = true;

	}

}

println "langs : "+textLangs

	 * @param rootDirFile the root dir file

	 * @param base the base

	 */

	{

		String filename = binDir.getName();

			}

		}

		println ""

	}

	public def getCorpusIDS() {

				}

			}

		}

		return true;

	}

imp.doValidation(true) // change this to not validate xml

imp.doTokenize(true) // change this, to not tokenize xml

imp.setStopIfMalformed(stopIfMalformed);

	println "import process stopped";

	return;

}

	def engine = Toolbox.getEngineManager(EngineType.ANNOTATION).getEngine("TreeTagger")

	if (engine.processDirectory(txmDir, binDir, ["lang":model])) {

		annotationSuccess = true;

	}

}

	String lang; // language used by the tokenizer

	/**

	 * Instantiates a new importer.

	 * @param outdir the outdir

	 * @param metadatas the metadatas

	 */

		this.trsfiles = trsfiles;

		this.txmDir = txmDir;

		this.binDir = binDir;

		this.metadatas = metadatas;

		this.lang = lang;

	}

	/**

			}

		}

		}

		cpb.done()

String userDir = System.getProperty("user.home");

def MONITOR;

{	}

String basename = corpusname

String model = lang

File srcDir = new File(rootDir);

binDir.mkdirs();

if (!binDir.exists()) {

	println "Could not create binDir "+binDir

		if (MONITOR != null) MONITOR.worked(1, "IMPORTER")

		if (MONITOR != null && MONITOR.isCanceled()) { return MONITOR.done(); }

		println "-- IMPORTER"

		if (!imp.run()) {

			println "Failed to prepare files - Aborting";

			return;

			if (engine.processDirectory(txmDir, binDir, ["lang":model])) {

				annotationSuccess = true;

				}

			}

		}

	if(debug) comp.setDebug();

	comp.removeInterviewers(removeInterviewer);

	comp.setIgnoreTranscriberMetadata(ignoreTranscriberMetadata);

		println "Failed to compile files";

		return;

	}

			List<String> NoSpaceBefore = LangFormater.getNoSpaceBefore(lang);

			List<String> NoSpaceAfter = LangFormater.getNoSpaceAfter(lang);

			if (t == null) {

				t.setName(txtname);

			}

			t.setSourceFile(txmFile)

			String txtname = txmFile.getName();

			int i = txtname.lastIndexOf(".");

			if(i > 0) txtname = txtname.substring(0, i);

			if (mediaFile.exists()) {

				File copy = new File(binDir, "media/"+mediaFile.getName())

if (MONITOR != null && MONITOR.isCanceled()) { return MONITOR.done(); }

if (MONITOR != null) MONITOR.worked(20, "FINALIZING")

	 * @param outdir the outdir

	 * @return true, if successful

	 */

		//println "run compiler with $xmlfiles, $basename and $outdir"

		this.outdir = binDir;

		sectionAttrs = new HashSet<String>() // reset section attributs set

		if (corpus != null) {

				corpus.clean(); // remove old files

			} else {

				corpus.delete(); // remove old files and TXMResult children

			}

		} else {

		}

		corpus.setDescription("Built with the XML-TRS import module");

		} catch (Exception ex) {System.out.println(ex); return false;}

		}

		return true;

				writer.writeAttribute("time", formatedTime)

				writer.writeAttribute("start", time)

				for (int i = 0; i < parser.getAttributeCount(); i++) { // write other attributes if any

					String v = parser.getAttributeLocalName(i);

					if (!("who".equals(v)) && !("overlap".equals(v)) && !("time".equals(v)) && !("speaker".equals(v)) && !("endTime".equals(v)) && !("startTime".equals(v))) {

println "-- IMPORTER - Reading source files with extension "+suffixes

if (MONITOR != null && MONITOR.isCanceled()) { return MONITOR.done(); }

if (MONITOR != null) MONITOR.worked(20, "IMPORTER - Reading source files with extension "+suffixes)

	println "Import process stopped";

	return;

}

	def engine = Toolbox.getEngineManager(EngineType.ANNOTATION).getEngine("TreeTagger")

	if (engine.processDirectory(txmDir, binDir, ["lang":model])) {

		annotationSuccess = true;

	}

}

			cwbMa.run(corpusname, outDir + "/registry");

		} catch (Exception ex) {System.out.println("CWB error: "+ex); return false;}

		return true;

	}

	 * @param basename the basename

	 * @return true, if successful

	 */

		File stokenizedDir = new File(binDir,"stokenized");

		stokenizedDir.deleteDir();

		files = txmDir.listFiles()

		if (files == null || files.size() == 0) return false

		return true;

	}

}

	 * @param basename the basename

	 * @return true, if successful

	 */

	{

		if (rootDirFile.listFiles() == null || rootDirFile.listFiles().size() == 0) {

			println "Error: no file to process in "+rootDirFile;

			}

		}

		cpb.done()

		return true;

	}

}

if (MONITOR != null && MONITOR.isCanceled()) { return MONITOR.done(); }

println "-- IMPORTER - Reading source files"

	println "import process stopped";

	return;

}

	if (engine.processDirectory(txmDir, binDir, ["lang":model])) {

		annotationSuccess = true;

	}

}

if (MONITOR != null && MONITOR.isCanceled()) { return MONITOR.done(); }

		} catch (Exception ex) {System.out.println(ex); return false;}

		return true;

	}

	def engine = Toolbox.getEngineManager(EngineType.ANNOTATION).getEngine("TreeTagger")

	if (engine.processDirectory(txmDir, binDir, ["lang":model])) {

		annotationSuccess = true;

	}

}

		} catch (Exception ex) {System.out.println(ex); return false;}

		return true;

	}

			cwbMa.run(corpusname, outDir + "/registry");

			return true;

		} catch (Exception ex) {System.out.println(ex); return false;}

		return true;

	}