root / tmp / org.txm.treetagger.core / src / org / txm / importer / xmltxm / Annotate.groovy @ 1681
History | View | Annotate | Download (18.3 kB)
| 1 |
// Copyright © 2010-2013 ENS de Lyon.
|
|---|---|
| 2 |
// Copyright © 2007-2010 ENS de Lyon, CNRS, INRP, University of
|
| 3 |
// Lyon 2, University of Franche-Comté, University of Nice
|
| 4 |
// Sophia Antipolis, University of Paris 3.
|
| 5 |
//
|
| 6 |
// The TXM platform is free software: you can redistribute it
|
| 7 |
// and/or modify it under the terms of the GNU General Public
|
| 8 |
// License as published by the Free Software Foundation,
|
| 9 |
// either version 2 of the License, or (at your option) any
|
| 10 |
// later version.
|
| 11 |
//
|
| 12 |
// The TXM platform is distributed in the hope that it will be
|
| 13 |
// useful, but WITHOUT ANY WARRANTY; without even the implied
|
| 14 |
// warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
|
| 15 |
// PURPOSE. See the GNU General Public License for more
|
| 16 |
// details.
|
| 17 |
//
|
| 18 |
// You should have received a copy of the GNU General
|
| 19 |
// Public License along with the TXM platform. If not, see
|
| 20 |
// http://www.gnu.org/licenses.
|
| 21 |
//
|
| 22 |
//
|
| 23 |
//
|
| 24 |
// $LastChangedDate: 2016-10-03 15:30:36 +0200 (lun. 03 oct. 2016) $
|
| 25 |
// $LastChangedRevision: 3313 $
|
| 26 |
// $LastChangedBy: mdecorde $
|
| 27 |
//
|
| 28 |
package org.txm.importer.xmltxm
|
| 29 |
|
| 30 |
import java.io.File |
| 31 |
import java.text.DateFormat |
| 32 |
import java.util.Date |
| 33 |
import java.util.concurrent.* |
| 34 |
|
| 35 |
import org.txm.Toolbox |
| 36 |
import org.txm.importer.* |
| 37 |
import org.txm.scripts.importer.* |
| 38 |
import org.txm.importer.scripts.xmltxm.AnnotationInjection |
| 39 |
import org.txm.importer.cwb.* |
| 40 |
import org.txm.objects.* |
| 41 |
import org.txm.treetagger.core.preferences.TreeTaggerPreferences |
| 42 |
import org.txm.utils.ConsoleProgressBar |
| 43 |
import org.txm.utils.LangDetector; |
| 44 |
import org.txm.utils.io.IOUtils |
| 45 |
import org.txm.utils.logger.Log; |
| 46 |
import org.txm.utils.treetagger.TreeTagger |
| 47 |
|
| 48 |
/**
|
| 49 |
* Annotate and replace the TEI-TXM files of the folder $rootDirFile/txm with TreeTagger.
|
| 50 |
* creates $rootDirFile/interp and $rootDirFile/treetagger
|
| 51 |
*
|
| 52 |
*/
|
| 53 |
class Annotate { |
| 54 |
boolean cancelNow = false; |
| 55 |
|
| 56 |
/** The report file. */
|
| 57 |
File reportFile;//contains the txm:application tag content |
| 58 |
|
| 59 |
/** The resp person. */
|
| 60 |
String respPerson;
|
| 61 |
|
| 62 |
/** The resp id. */
|
| 63 |
String respId;
|
| 64 |
|
| 65 |
/** The resp desc. */
|
| 66 |
String respDesc;
|
| 67 |
|
| 68 |
/** The resp date. */
|
| 69 |
String respDate;
|
| 70 |
|
| 71 |
/** The resp when. */
|
| 72 |
String respWhen;
|
| 73 |
|
| 74 |
/** The app ident. */
|
| 75 |
String appIdent;
|
| 76 |
|
| 77 |
/** The app version. */
|
| 78 |
String appVersion;
|
| 79 |
|
| 80 |
/** The distributor. */
|
| 81 |
String distributor;
|
| 82 |
|
| 83 |
/** The publi stmt. */
|
| 84 |
String publiStmt;
|
| 85 |
|
| 86 |
/** The source stmt. */
|
| 87 |
String sourceStmt;
|
| 88 |
|
| 89 |
/** The types. */
|
| 90 |
def types;
|
| 91 |
|
| 92 |
/** The types title. */
|
| 93 |
def typesTITLE;
|
| 94 |
|
| 95 |
/** The types desc. */
|
| 96 |
def typesDesc;
|
| 97 |
|
| 98 |
/** The types tagset. */
|
| 99 |
def typesTAGSET;
|
| 100 |
|
| 101 |
/** The types web. */
|
| 102 |
def typesWEB;
|
| 103 |
|
| 104 |
/** The idform. */
|
| 105 |
String idform;
|
| 106 |
|
| 107 |
/** The debug. */
|
| 108 |
boolean debug = false; |
| 109 |
|
| 110 |
File modelsDirectory;
|
| 111 |
|
| 112 |
public Annotate() {
|
| 113 |
modelsDirectory = new File(TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.MODELS_PATH)); // default models directory is set in the Toolbox |
| 114 |
} |
| 115 |
|
| 116 |
/**
|
| 117 |
* Sets the debug.
|
| 118 |
*/
|
| 119 |
public void setDebug() { |
| 120 |
debug = true;
|
| 121 |
} |
| 122 |
|
| 123 |
String id;
|
| 124 |
/**
|
| 125 |
* Inits the tt outfile infos.
|
| 126 |
*
|
| 127 |
* @param rootDirFile the root dir file
|
| 128 |
* @param modelfile the modelfile
|
| 129 |
*/
|
| 130 |
public void initTTOutfileInfos(File rootDirFile, File modelfile, String modelfilename) |
| 131 |
{
|
| 132 |
initTTOutfileInfos(rootDirFile, modelfile, modelfilename, null);
|
| 133 |
} |
| 134 |
|
| 135 |
/**
|
| 136 |
* Inits the tt outfile infos.
|
| 137 |
*
|
| 138 |
* @param rootDirFile the root dir file
|
| 139 |
* @param modelfile the modelfile
|
| 140 |
* @param properties : 2 element array that contains the word properties to create. It can be null (the modelfilename will be used)
|
| 141 |
*/
|
| 142 |
public void initTTOutfileInfos(File rootDirFile, File modelfile, String modelfilename, String[] properties) |
| 143 |
{
|
| 144 |
id = modelfilename; |
| 145 |
String[] split = id.split("\\."); |
| 146 |
if (split.length > 0) id = split[0]; |
| 147 |
if (id.equals("??")) id = "xx" |
| 148 |
|
| 149 |
reportFile = new File(rootDirFile,"NLPToolsParameters.xml"); |
| 150 |
|
| 151 |
respPerson = System.getProperty("user.name"); |
| 152 |
respId = "txm";
|
| 153 |
respDesc = "NLP annotation tool";
|
| 154 |
respDate = DateFormat.getDateInstance(DateFormat.SHORT, Locale.UK).format(new Date()); |
| 155 |
respWhen = DateFormat.getDateInstance(DateFormat.FULL, Locale.UK).format(new Date()); |
| 156 |
|
| 157 |
appIdent = "TreeTagger";
|
| 158 |
appVersion = "3.2";
|
| 159 |
|
| 160 |
distributor = "";
|
| 161 |
publiStmt = """""";
|
| 162 |
sourceStmt = """""";
|
| 163 |
|
| 164 |
if (properties != null && properties.length == 2) { |
| 165 |
types = [properties[0],properties[1]]; |
| 166 |
typesTITLE = [properties[0],properties[1]]; |
| 167 |
} else {
|
| 168 |
types = [id+"pos",id+"lemma"]; |
| 169 |
typesTITLE = [id+"pos",id+"lemma"]; |
| 170 |
} |
| 171 |
|
| 172 |
//TODO: the tagset, website and description should be referenced in the model catalog
|
| 173 |
if(modelfile.getName() == "rgaqcj.par") { |
| 174 |
typesDesc = ["CATTEX pos tagset built with BFM texts","fr lemma of the model "+modelfile+" - "] |
| 175 |
typesTAGSET = ["http://bfm.ens-lyon.fr/IMG/pdf/Cattex2009_Manuel.pdf",""] |
| 176 |
typesWEB = ["http://bfm.ens-lyon.fr/",""] |
| 177 |
} else {
|
| 178 |
typesDesc = ["pos tagset built from model "+modelfile,id+" lemma of the model "+modelfile+" - "] |
| 179 |
typesTAGSET = ["",""] |
| 180 |
typesWEB = ["",""] |
| 181 |
} |
| 182 |
|
| 183 |
idform ="w";
|
| 184 |
} |
| 185 |
|
| 186 |
/**
|
| 187 |
* Apply tt.
|
| 188 |
*
|
| 189 |
* @param ttsrcfile the ttsrcfile
|
| 190 |
* @param ttoutfile the ttoutfile
|
| 191 |
* @param modelfile the modelfile
|
| 192 |
* @return true, if successful
|
| 193 |
*/
|
| 194 |
public boolean applyTT(File ttsrcfile, File ttoutfile, File modelfile) { |
| 195 |
applyTT(ttsrcfile, ttoutfile, modelfile, null)
|
| 196 |
} |
| 197 |
|
| 198 |
/**
|
| 199 |
* Apply tt.
|
| 200 |
*
|
| 201 |
* @param ttsrcfile the ttsrcfile
|
| 202 |
* @param ttoutfile the ttoutfile
|
| 203 |
* @param modelfile the modelfile
|
| 204 |
* @param options, if null use value set in Toolbox preferences
|
| 205 |
* @return true, if successful
|
| 206 |
*/
|
| 207 |
public boolean applyTT(File ttsrcfile, File ttoutfile, File modelfile, String[] options) |
| 208 |
{
|
| 209 |
try {
|
| 210 |
File infile = ttsrcfile;
|
| 211 |
File outfile = ttoutfile;
|
| 212 |
|
| 213 |
def tt = new TreeTagger(TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.INSTALL_PATH)+"/bin/", options); |
| 214 |
tt.settoken(); |
| 215 |
tt.setlemma(); |
| 216 |
tt.setsgml(); |
| 217 |
if (TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.OPTIONS_UNKNOWN)) {
|
| 218 |
tt.setnounknown(); |
| 219 |
} |
| 220 |
tt.seteostag("<s>");
|
| 221 |
if (TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.OPTIONS_DEBUG)) {
|
| 222 |
tt.debug(true);
|
| 223 |
} else {
|
| 224 |
tt.setquiet(); |
| 225 |
} |
| 226 |
if (TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.OPTIONS_CAPHEURISTIC)) {
|
| 227 |
tt.setcapheuristics(); |
| 228 |
} |
| 229 |
if (TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.OPTIONS_HYPHENHEURISTIC)) {
|
| 230 |
tt.sethyphenheuristics(); |
| 231 |
} |
| 232 |
if (TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.OPTIONS_PROB)) {
|
| 233 |
tt.setprob(); |
| 234 |
} |
| 235 |
|
| 236 |
String lex = TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.OPTIONS_LEX);
|
| 237 |
if (lex !=null && lex.length() > 0) { |
| 238 |
tt.setlex(lex); |
| 239 |
} |
| 240 |
String wc = TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.OPTIONS_WC);
|
| 241 |
if (wc !=null && wc.length() > 0) { |
| 242 |
tt.setwc(wc); |
| 243 |
} |
| 244 |
tt.treetagger(modelfile.getAbsolutePath(), infile.getAbsolutePath(), outfile.getAbsolutePath()) |
| 245 |
infile.delete(); |
| 246 |
} catch(Exception e) { |
| 247 |
Log.printStackTrace(e); |
| 248 |
System.out.println("Failed to apply TreeTagger on $ttsrcfile input file with the $modelfile model file."); |
| 249 |
return false; |
| 250 |
} |
| 251 |
return true; |
| 252 |
} |
| 253 |
|
| 254 |
/**
|
| 255 |
* Write standoff file.
|
| 256 |
*
|
| 257 |
* @param ttoutfile the ttoutfile
|
| 258 |
* @param posfile the posfile
|
| 259 |
* @return true, if successful
|
| 260 |
*/
|
| 261 |
public boolean writeStandoffFile(File ttoutfile, File posfile) |
| 262 |
{
|
| 263 |
def encoding ="UTF-8"; |
| 264 |
def transfo = new CSV2W_ANA(); |
| 265 |
//println("build w-interp "+ttfile.getName()+ ">>"+posfile.getName())
|
| 266 |
transfo.setAnnotationTypes( types, typesDesc, typesTAGSET, typesWEB, idform); |
| 267 |
transfo.setResp(respId, respDesc, respDate, respPerson, respWhen); |
| 268 |
transfo.setApp(appIdent, appVersion); |
| 269 |
transfo.setTarget(ttoutfile.getAbsolutePath(), reportFile); |
| 270 |
transfo.setInfos(distributor, publiStmt, sourceStmt); |
| 271 |
return transfo.process( ttoutfile, posfile, encoding );
|
| 272 |
} |
| 273 |
|
| 274 |
/**
|
| 275 |
* Run step by step : build TT src files, run TT, build xml-standoff files, inject standoff annotations
|
| 276 |
*
|
| 277 |
* @param rootDirFile the root dir file
|
| 278 |
* @param modelfilename the modelfilename
|
| 279 |
* @return true, if successful
|
| 280 |
*/
|
| 281 |
public boolean run(File binDir, File txmDir, String modelfilename) |
| 282 |
{
|
| 283 |
//test if modelfile exists
|
| 284 |
if (debug) {
|
| 285 |
println "rootDirFile "+binDir
|
| 286 |
println "txmDir "+txmDir
|
| 287 |
println "TREETAGGER INSTALL PATH : "+TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.INSTALL_PATH);
|
| 288 |
println "TREETAGGER MODELS PATH : "+TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.MODELS_PATH)
|
| 289 |
} |
| 290 |
|
| 291 |
//test if the Toolbox know TreeTagger
|
| 292 |
if (!new File(TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.INSTALL_PATH)+"/bin/").exists()) { |
| 293 |
println("Could not find TreeTagger binaries in "+TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.INSTALL_PATH)+"/bin/") |
| 294 |
return false; |
| 295 |
} |
| 296 |
String langAll = null |
| 297 |
String lang;
|
| 298 |
if (modelfilename.startsWith("??")) { |
| 299 |
langAll = new LangDetector(binDir).getLang();
|
| 300 |
println "General lang $langAll"
|
| 301 |
} |
| 302 |
|
| 303 |
//cleaning
|
| 304 |
new File(binDir, "annotations").deleteDir(); |
| 305 |
new File(binDir, "annotations").mkdir(); |
| 306 |
new File(binDir, "treetagger").deleteDir(); |
| 307 |
new File(binDir, "treetagger").mkdir(); |
| 308 |
|
| 309 |
ArrayList<String> milestones = []; |
| 310 |
|
| 311 |
//BUILD TT FILE READY TO BE TAGGED
|
| 312 |
List<File> files = txmDir.listFiles(IOUtils.HIDDENFILE_FILTER) |
| 313 |
|
| 314 |
println("Building TT source files ("+files.size()+") from directory "+txmDir) |
| 315 |
ConsoleProgressBar cpb = new ConsoleProgressBar(files.size())
|
| 316 |
for (File f : files) { |
| 317 |
cpb.tick() |
| 318 |
File srcfile = f;
|
| 319 |
File resultfile = new File(binDir, "treetagger/"+f.getName()+".tt"); |
| 320 |
if(debug)
|
| 321 |
println "build tt src : "+srcfile+" >> "+resultfile |
| 322 |
def ttsrcbuilder = new BuildTTSrc(srcfile.toURI().toURL()) |
| 323 |
if (!ttsrcbuilder.process(resultfile, null)) |
| 324 |
System.out.println("Failed to build tt src file of "+srcfile); |
| 325 |
} |
| 326 |
|
| 327 |
if (cancelNow) return; |
| 328 |
|
| 329 |
File modelDirectory = new File(TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.MODELS_PATH)); |
| 330 |
if (!modelDirectory.exists()) {
|
| 331 |
println "Skipping ANNOTATE: TreeTagger language model file directory not found: "+modelDirectory.getAbsolutePath();
|
| 332 |
return false; |
| 333 |
} else if (!modelDirectory.canRead()) { |
| 334 |
println "Skipping ANNOTATE: impossible to access the TreeTagger language model file directory: "+modelDirectory.getAbsolutePath();
|
| 335 |
return false; |
| 336 |
} |
| 337 |
println("")
|
| 338 |
//Convert encoding if needed
|
| 339 |
|
| 340 |
//APPLY TREETAGGER
|
| 341 |
files = new File(binDir, "treetagger").listFiles(IOUtils.HIDDENFILE_FILTER) |
| 342 |
println("Applying $modelfilename TreeTagger model on dir: "+new File(binDir, "treetagger")+ " ("+files.size()+" files)") |
| 343 |
if (files == null || files.size() == 0) |
| 344 |
return false; |
| 345 |
File modelfile;
|
| 346 |
cpb = new ConsoleProgressBar(files.size())
|
| 347 |
for (File f : files) { |
| 348 |
String tmpModelFileName = modelfilename
|
| 349 |
if (modelfilename.startsWith("??")) { |
| 350 |
lang = langAll; |
| 351 |
if (f.length() > LangDetector.MINIMALSIZE) {
|
| 352 |
lang = new LangDetector(f).getLang();
|
| 353 |
//println "guessing lang $f : $lang"
|
| 354 |
} |
| 355 |
tmpModelFileName = lang+".par"
|
| 356 |
} |
| 357 |
modelfile = new File(modelsDirectory, tmpModelFileName); |
| 358 |
if (debug)
|
| 359 |
println "model file : "+modelfile;
|
| 360 |
|
| 361 |
File
|
| 362 |
|
| 363 |
if (!modelfile.exists()) {
|
| 364 |
println "Skipping ANNOTATE: '$modelfile' TreeTagger language model file not found."
|
| 365 |
if(System.getProperty("os.name").startsWith("Windows") || System.getProperty("os.name").startsWith("Mac")) |
| 366 |
println "Windows&Mac users: the operating system might be hiding file extensions. Use your file explorer to check the file name."
|
| 367 |
return false; |
| 368 |
} else if (!modelfile.canRead()) { |
| 369 |
println "Skipping ANNOTATE: impossible to access the '$modelfile' TreeTagger language model file."
|
| 370 |
return false; |
| 371 |
} |
| 372 |
|
| 373 |
// if (modelfile.getName().equals("sp.par")) {//UTF >> Latin1
|
| 374 |
// if(debug)
|
| 375 |
// println "fix encoding for model "+modelfile
|
| 376 |
// new EncodingConverter(f, "UTF-8", "ISO-8859-1")
|
| 377 |
// }
|
| 378 |
|
| 379 |
cpb.tick() |
| 380 |
File infile = f;
|
| 381 |
File outfile = new File(f.getParent(),f.getName()+"-out.tt"); |
| 382 |
if (!applyTT(infile, outfile, modelfile)) {
|
| 383 |
return false; |
| 384 |
} |
| 385 |
|
| 386 |
// //Reconvert encoding if needed
|
| 387 |
// if (modelfile.getName().equals("sp.par")) {
|
| 388 |
// if(debug)
|
| 389 |
// println "convert "+f+" latin1 >> UTF-8"
|
| 390 |
// new EncodingConverter(f, "ISO-8859-1", "UTF-8")
|
| 391 |
// }
|
| 392 |
} |
| 393 |
println("")
|
| 394 |
|
| 395 |
if (cancelNow) return; |
| 396 |
|
| 397 |
//BUILD STAND-OFF FILES
|
| 398 |
if (modelfile == null) { |
| 399 |
println "no model applied"
|
| 400 |
return false; |
| 401 |
} |
| 402 |
|
| 403 |
initTTOutfileInfos(binDir, modelfile, modelfilename); |
| 404 |
files = new File(binDir, "treetagger").listFiles(IOUtils.HIDDENFILE_FILTER)// now contains the result files of TT |
| 405 |
println "Building stdoff files ("+files.size()+") from dir:"+new File(binDir, "treetagger")+" to "+new File(binDir, "annotations"); |
| 406 |
if (files == null || files.size() == 0) |
| 407 |
return false; |
| 408 |
cpb = new ConsoleProgressBar(files.size())
|
| 409 |
for (File ttrezfile : files) { |
| 410 |
cpb.tick() |
| 411 |
File annotfile = new File(binDir, "annotations/"+ttrezfile.getName()+"-STOFF.xml"); |
| 412 |
if (!writeStandoffFile(ttrezfile, annotfile))
|
| 413 |
println("Failed to build standoff file of "+ttrezfile);
|
| 414 |
} |
| 415 |
println("")
|
| 416 |
|
| 417 |
if (cancelNow) return; |
| 418 |
|
| 419 |
//INJECT ANNOTATIONS
|
| 420 |
List<File> interpfiles = new File(binDir, "annotations").listFiles(IOUtils.HIDDENFILE_FILTER); |
| 421 |
List<File> txmfiles = txmDir.listFiles(IOUtils.HIDDENFILE_FILTER); |
| 422 |
if (txmfiles == null) { |
| 423 |
println "No file to annotate in "+txmDir.getAbsolutePath()
|
| 424 |
return false; |
| 425 |
} |
| 426 |
interpfiles.sort(); // same order
|
| 427 |
txmfiles.sort(); //same order
|
| 428 |
println "Injecting stdoff files ("+interpfiles.size()+") data from "+new File(binDir, "annotations")+ " to xml-txm files of "+txmDir; |
| 429 |
if (interpfiles == null || interpfiles.size() == 0) |
| 430 |
return false; |
| 431 |
cpb = new ConsoleProgressBar(interpfiles.size())
|
| 432 |
for (int i = 0 ; i < interpfiles.size() ; i++) { |
| 433 |
cpb.tick() |
| 434 |
File srcfile = txmfiles.get(i);
|
| 435 |
File pos1file = interpfiles.get(i);
|
| 436 |
File temp = File.createTempFile("Annotate", "temp", srcfile.getParentFile()); |
| 437 |
def builder = new AnnotationInjection(srcfile.toURI().toURL(), pos1file.toURI().toURL()); |
| 438 |
if (!builder.process(temp)) {
|
| 439 |
return false; |
| 440 |
} |
| 441 |
builder = null;
|
| 442 |
|
| 443 |
//println "renaming files..."
|
| 444 |
if (!(srcfile.delete() && temp.renameTo(srcfile)))
|
| 445 |
println "Warning can't rename file "+temp+" to "+srcfile |
| 446 |
} |
| 447 |
|
| 448 |
println("")
|
| 449 |
return true; |
| 450 |
} |
| 451 |
|
| 452 |
public void setModelsDirectory(File modelsDirectory) { |
| 453 |
this.modelsDirectory = modelsDirectory;
|
| 454 |
} |
| 455 |
|
| 456 |
/**
|
| 457 |
* Run file by file. Allow to have one different lang per file. Default behavior add new word properties
|
| 458 |
*
|
| 459 |
* @param binDir
|
| 460 |
* @param txmDir
|
| 461 |
* @param lang associate a file name with a model filename
|
| 462 |
* @return true, if successful
|
| 463 |
*/
|
| 464 |
public boolean run(File binDir, File txmDir, HashMap<String, String> langs) |
| 465 |
{
|
| 466 |
return run(binDir, txmDir, langs, false, new String[0], new String[0]); |
| 467 |
} |
| 468 |
|
| 469 |
/**
|
| 470 |
* Run file by file. Allow to have one different lang per file
|
| 471 |
*
|
| 472 |
* @param binDir
|
| 473 |
* @param txmDir
|
| 474 |
* @param lang associate a file name with a model filename
|
| 475 |
* @param replace, replace or create a word property
|
| 476 |
* @return true, if successful
|
| 477 |
*/
|
| 478 |
public boolean run(File binDir, File txmDir, HashMap<String, String> langs, boolean replace, String[] properties, String[] options) |
| 479 |
{
|
| 480 |
if (!new File(TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.INSTALL_PATH)+"/bin/").exists()) { |
| 481 |
println("Path to TreeTagger is wrong "+TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.INSTALL_PATH)+"/bin/") |
| 482 |
return true; |
| 483 |
} |
| 484 |
|
| 485 |
List<File> listfiles = txmDir.listFiles(IOUtils.HIDDENFILE_FILTER); |
| 486 |
|
| 487 |
//cleaning
|
| 488 |
File annotDir = new File(binDir,"annotations"); |
| 489 |
annotDir.deleteDir(); |
| 490 |
annotDir.mkdir(); |
| 491 |
File ptreetaggerDir = new File(binDir,"ptreetagger"); |
| 492 |
ptreetaggerDir.deleteDir(); |
| 493 |
ptreetaggerDir.mkdir(); |
| 494 |
File treetaggerDir = new File(binDir,"treetagger"); |
| 495 |
treetaggerDir.deleteDir(); |
| 496 |
treetaggerDir.mkdir(); |
| 497 |
|
| 498 |
int cores = Runtime.getRuntime().availableProcessors() |
| 499 |
int coresToUse = Math.max(1.0, cores * 0.7) |
| 500 |
ExecutorService pool = Executors.newFixedThreadPool(coresToUse) |
| 501 |
|
| 502 |
def files = txmDir.listFiles(IOUtils.HIDDENFILE_FILTER)
|
| 503 |
ConsoleProgressBar cpb = new ConsoleProgressBar(files.size())
|
| 504 |
for (File teiFile : files) { |
| 505 |
int counter = 1; |
| 506 |
ThreadFile t = new ThreadFile("TT_"+counter++, teiFile) { |
| 507 |
|
| 508 |
public void run() { |
| 509 |
|
| 510 |
if (cancelNow) return; |
| 511 |
if (langs.get(f.getName()) == null) { |
| 512 |
println "Error: no lang defined for file $f"
|
| 513 |
return;
|
| 514 |
} |
| 515 |
|
| 516 |
String lang = langs.get(f.getName());
|
| 517 |
run(f, lang, binDir, txmDir, replace, properties, options, annotDir, ptreetaggerDir, treetaggerDir) |
| 518 |
|
| 519 |
cpb.tick(); |
| 520 |
} |
| 521 |
}; |
| 522 |
|
| 523 |
pool.execute(t) |
| 524 |
} |
| 525 |
|
| 526 |
pool.shutdown() |
| 527 |
pool.awaitTermination(10, TimeUnit.HOURS) |
| 528 |
println ""
|
| 529 |
return true; |
| 530 |
} |
| 531 |
public boolean run(File f, String lang, File binDir, File txmDir) { |
| 532 |
|
| 533 |
File annotDir = new File(binDir,"annotations"); |
| 534 |
annotDir.mkdir(); |
| 535 |
File ptreetaggerDir = new File(binDir,"ptreetagger"); |
| 536 |
ptreetaggerDir.mkdir(); |
| 537 |
File treetaggerDir = new File(binDir,"treetagger"); |
| 538 |
treetaggerDir.mkdir(); |
| 539 |
|
| 540 |
return run(f, lang, binDir, txmDir, false, new String[0], new String[0], annotDir, ptreetaggerDir, treetaggerDir) |
| 541 |
} |
| 542 |
|
| 543 |
|
| 544 |
public boolean run(File f, String lang, File binDir, File txmDir, boolean replace, String[] properties, String[] options, File annotDir, File ptreetaggerDir, File treetaggerDir) { |
| 545 |
|
| 546 |
File modelfile = new File(modelsDirectory, lang+".par"); |
| 547 |
if (!modelfile.exists()) {
|
| 548 |
println "Error: No Modelfile available for lang "+modelfile+". Continue import process "; |
| 549 |
return false; |
| 550 |
} |
| 551 |
File annotfile = new File(annotDir, f.getName()+"-STDOFF.xml"); |
| 552 |
File ttsrcfile = new File(ptreetaggerDir, f.getName()+"-src.tt"); |
| 553 |
File ttrezfile = new File(treetaggerDir, f.getName()+"-out.tt"); |
| 554 |
//println ("TT with $model "+f+"+"+annotfile+" > "+ttsrcfile+" > "+ttrezfile);
|
| 555 |
|
| 556 |
//BUILD TT FILE READY TO BE TAGGED
|
| 557 |
def builder = new BuildTTSrc(f.toURL()); |
| 558 |
builder.process(ttsrcfile, null);
|
| 559 |
|
| 560 |
//Apply TT
|
| 561 |
applyTT(ttsrcfile, ttrezfile, modelfile, options); |
| 562 |
|
| 563 |
//CREATE STANDOFF FILES
|
| 564 |
initTTOutfileInfos(binDir, modelfile, lang, properties); |
| 565 |
writeStandoffFile(ttrezfile, annotfile) |
| 566 |
|
| 567 |
//INJECT ANNOTATIONS
|
| 568 |
File tmpFile = new File(txmDir, "temp_"+f.getName()) |
| 569 |
builder = new AnnotationInjection(f.toURL(), annotfile.toURL(), replace);
|
| 570 |
builder.process(tmpFile); |
| 571 |
if (!(f.delete() && tmpFile.renameTo(f))) println "Warning can't rename file "+tmpFile+" to "+f |
| 572 |
|
| 573 |
return f.exists();
|
| 574 |
} |
| 575 |
|
| 576 |
public void setCancelNow() { |
| 577 |
cancelNow = true;
|
| 578 |
} |
| 579 |
|
| 580 |
public class ThreadFile extends Thread { |
| 581 |
File f;
|
| 582 |
public ThreadFile(String name, File f) { |
| 583 |
this.setName(name)
|
| 584 |
this.f = f;
|
| 585 |
} |
| 586 |
} |
| 587 |
} |