Révision 3768
| TXM/trunk/bundles/org.txm.groovy.core/src/groovy/org/txm/scripts/importer/xtz/TTAnnotater.groovy (revision 3768) | ||
|---|---|---|
| 2 | 2 |
|
| 3 | 3 |
import java.io.File; |
| 4 | 4 |
|
| 5 |
import org.txm.core.preferences.TBXPreferences |
|
| 5 | 6 |
import org.txm.importer.xmltxm.Annotate |
| 6 | 7 |
import org.txm.importer.xtz.* |
| 7 | 8 |
import org.txm.Toolbox |
| 8 | 9 |
import org.txm.core.engines.* |
| 10 |
import org.txm.importer.ApplyXsl2 |
|
| 9 | 11 |
|
| 10 | 12 |
/** |
| 11 | 13 |
* Wraps current Annotate class into the import workflow |
| ... | ... | |
| 25 | 27 |
//String model = module.getParameters().get(ImportKeys.TTMODEL); |
| 26 | 28 |
boolean cleanDirectories = module.getProject().getCleanAfterBuild(); |
| 27 | 29 |
|
| 30 |
boolean expertMode = TBXPreferences.getInstance().getBoolean(TBXPreferences.EXPERT_USER); |
|
| 31 |
|
|
| 28 | 32 |
String corpusname = module.getProject().getName(); |
| 29 | 33 |
String lang = module.getProject().getLang(); |
| 30 | 34 |
|
| ... | ... | |
| 44 | 48 |
} |
| 45 | 49 |
} |
| 46 | 50 |
|
| 51 |
if (isSuccessFul && expertMode) doPostNLPXSLStep(); |
|
| 52 |
|
|
| 47 | 53 |
if (cleanDirectories) {
|
| 48 | 54 |
new File(module.getBinaryDirectory(), "treetagger").deleteDir() |
| 49 | 55 |
new File(module.getBinaryDirectory(), "ptreetagger").deleteDir() |
| ... | ... | |
| 52 | 58 |
|
| 53 | 59 |
if (!isSuccessFul) reason = "TreeTagger annotation failed." |
| 54 | 60 |
} |
| 61 |
|
|
| 62 |
/** |
|
| 63 |
* read from $inputDirectory and write the result in $bindir/txm |
|
| 64 |
* |
|
| 65 |
*/ |
|
| 66 |
public boolean doPostNLPXSLStep() {
|
|
| 67 |
|
|
| 68 |
//filesToProcess = inputDirectory.listFiles(); |
|
| 69 |
|
|
| 70 |
File postnlpXSLdirectory = new File(module.getSourceDirectory(), "xsl/4-postnlp") |
|
| 71 |
|
|
| 72 |
def xslFiles = postnlpXSLdirectory.listFiles(new FileFilter() {
|
|
| 73 |
public boolean accept(File pathname) { return pathname.getName().endsWith(".xsl");}
|
|
| 74 |
}); |
|
| 75 |
def xslParams = project.getXsltParameters() |
|
| 76 |
xslParams["output-directory"] = outputDirectory.getAbsoluteFile().toURI().toString(); |
|
| 77 |
|
|
| 78 |
if (postnlpXSLdirectory.exists() && xslFiles != null && xslFiles.size() > 0) {
|
|
| 79 |
|
|
| 80 |
def files = inputDirectory.listFiles() |
|
| 81 |
|
|
| 82 |
println "-- PostNLP XSL Step with the $postnlpXSLdirectory directory." |
|
| 83 |
xslFiles.sort() |
|
| 84 |
for (File xslFile : xslFiles) {
|
|
| 85 |
if (xslFile.isDirectory() || xslFile.isHidden() || !xslFile.getName().endsWith(".xsl")) continue;
|
|
| 86 |
//if (!xslFile.getName().matches("[1-9]-.+")) continue;
|
|
| 87 |
|
|
| 88 |
if (ApplyXsl2.processImportSources(xslFile, files, xslParams)) {
|
|
| 89 |
inputDirectory = outputDirectory; // the files to process are now in the "txm" directory |
|
| 90 |
println "" |
|
| 91 |
} else {
|
|
| 92 |
reason = "Fail to apply post NLP XSL: $xslFile" |
|
| 93 |
return false; |
|
| 94 |
} |
|
| 95 |
} |
|
| 96 |
} else {
|
|
| 97 |
//println "Nothing to do." |
|
| 98 |
} |
|
| 99 |
return true; |
|
| 100 |
} |
|
| 55 | 101 |
|
| 56 | 102 |
@Override |
| 57 | 103 |
public void cancel() {
|
| TXM/trunk/bundles/org.txm.groovy.core/src/groovy/org/txm/scripts/importer/xtz/XTZImporter.groovy (revision 3768) | ||
|---|---|---|
| 126 | 126 |
|
| 127 | 127 |
// main workflow of XTZ importer step |
| 128 | 128 |
|
| 129 |
boolean expertMode = TBXPreferences.getInstance().getBoolean(TBXPreferences.EXPERT_USER); |
|
| 130 |
|
|
| 129 | 131 |
if (!doFixSurrogates()) return; |
| 130 | 132 |
if (!doSplitMergeXSLStep()) return; |
| 131 | 133 |
if (!doFrontXSLStep()) return; |
| ... | ... | |
| 135 | 137 |
if (!doEncodeMileStonesStep()) return; |
| 136 | 138 |
if (!doToXMLTXMStep()) return; |
| 137 | 139 |
if (!doInjectMetadataStep()) return; |
| 140 |
if (expertMode && !doBackXSLStep()) return; |
|
| 138 | 141 |
|
| 139 | 142 |
isSuccessFul = filesToProcess.size() > 0 |
| 140 | 143 |
|
| ... | ... | |
| 289 | 292 |
inputDirectory = outputDirectory; // the files to process are now in the "txm" directory |
| 290 | 293 |
println "" |
| 291 | 294 |
} else {
|
| 292 |
reason = "Fail to apply front XSL: $xslPath"
|
|
| 295 |
reason = "Fail to apply front XSL: $xslFile"
|
|
| 293 | 296 |
return false; |
| 294 | 297 |
} |
| 295 | 298 |
} |
| ... | ... | |
| 299 | 302 |
return true; |
| 300 | 303 |
} |
| 301 | 304 |
|
| 305 |
/** |
|
| 306 |
* read from $inputDirectory and write the result in $bindir/txm |
|
| 307 |
* |
|
| 308 |
*/ |
|
| 309 |
public boolean doBackXSLStep() {
|
|
| 310 |
|
|
| 311 |
//filesToProcess = inputDirectory.listFiles(); |
|
| 312 |
|
|
| 313 |
File backXSLdirectory = new File(module.getSourceDirectory(), "xsl/4-back") |
|
| 314 |
|
|
| 315 |
def xslFiles = backXSLdirectory.listFiles(new FileFilter() {
|
|
| 316 |
public boolean accept(File pathname) { return pathname.getName().endsWith(".xsl");}
|
|
| 317 |
}); |
|
| 318 |
|
|
| 319 |
xslParams["output-directory"] = outputDirectory.getAbsoluteFile().toURI().toString(); |
|
| 320 |
|
|
| 321 |
if (backXSLdirectory.exists() && xslFiles != null && xslFiles.size() > 0) {
|
|
| 322 |
|
|
| 323 |
println "-- Back XSL Step with the $backXSLdirectory directory." |
|
| 324 |
xslFiles.sort() |
|
| 325 |
for (File xslFile : xslFiles) {
|
|
| 326 |
if (xslFile.isDirectory() || xslFile.isHidden() || !xslFile.getName().endsWith(".xsl")) continue;
|
|
| 327 |
//if (!xslFile.getName().matches("[1-9]-.+")) continue;
|
|
| 328 |
|
|
| 329 |
if (ApplyXsl2.processImportSources(xslFile, filesToProcess, xslParams)) {
|
|
| 330 |
println "" |
|
| 331 |
} else {
|
|
| 332 |
reason = "Fail to apply back XSL: $xslFile" |
|
| 333 |
return false; |
|
| 334 |
} |
|
| 335 |
} |
|
| 336 |
} else {
|
|
| 337 |
//println "Nothing to do." |
|
| 338 |
} |
|
| 339 |
return true; |
|
| 340 |
} |
|
| 341 |
|
|
| 302 | 342 |
public boolean doCheckXMLFileStep() {
|
| 303 | 343 |
filesToProcess = [] |
| 304 | 344 |
println "-- Checking XML-TEI files for well-formedness." |
| TXM/trunk/bundles/org.txm.core/src/java/org/txm/importer/ApplyXsl2.java (revision 3768) | ||
|---|---|---|
| 436 | 436 |
cpb.tick(); |
| 437 | 437 |
File outfile = new File(outdir, name); |
| 438 | 438 |
|
| 439 |
if (!f.getAbsolutePath().equals(outfile.getAbsolutePath()) && f.lastModified() <= outfile.lastModified()) {
|
|
| 440 |
if (debug) System.out.println("skipped: " + f);
|
|
| 441 |
continue; // input file is older than output file |
|
| 442 |
} |
|
| 439 |
// if (!f.getAbsolutePath().equals(outfile.getAbsolutePath()) && f.lastModified() <= outfile.lastModified()) {
|
|
| 440 |
// if (debug) System.out.println("Skipping: " + f);
|
|
| 441 |
// continue; // input file is older than output file
|
|
| 442 |
// }
|
|
| 443 | 443 |
|
| 444 | 444 |
if (deleteOutput) outfile = null; // if outfile is null builder.process won't create the output file adn will replace f |
| 445 | 445 |
|
Formats disponibles : Unified diff