Révision 1371
| tmp/org.txm.tigersearch.rcp/groovy/org/txm/scripts/importer/srcmf/srcmfLoader.groovy (revision 1371) | ||
|---|---|---|
| 54 | 54 |
debug = true |
| 55 | 55 |
params = new BaseParameters(new File(userDir, "xml/slethgier/import.xml")) |
| 56 | 56 |
params.load() |
| 57 |
if (!org.txm.Toolbox.isInitialized()) {
|
|
| 58 |
Toolbox.setParam(Toolbox.INSTALL_DIR,new File("/usr/lib/TXM"));
|
|
| 59 |
Toolbox.setParam(Toolbox.TREETAGGER_INSTALL_PATH,new File(userDir,"treetagger")); |
|
| 60 |
Toolbox.setParam(Toolbox.TREETAGGER_MODELS_PATH,new File(userDir,"treetagger/models")); |
|
| 61 |
Toolbox.setParam(Toolbox.METADATA_ENCODING, "UTF-8"); |
|
| 62 |
Toolbox.setParam(Toolbox.METADATA_COLSEPARATOR, ","); |
|
| 63 |
Toolbox.setParam(Toolbox.METADATA_TXTSEPARATOR, "\""); |
|
| 64 |
Toolbox.setParam(Toolbox.USER_TXM_HOME, new File(System.getProperty("user.home"), "TXM"));
|
|
| 65 |
} |
|
| 66 | 57 |
} |
| 67 | 58 |
if (params == null) { println "no parameters. Aborting"; return; }
|
| 68 | 59 |
|
| tmp/org.txm.tigersearch.rcp/groovy/org/txm/scripts/importer/tigersearch/tigersearchLoader.groovy (revision 1371) | ||
|---|---|---|
| 52 | 52 |
debug = true |
| 53 | 53 |
params = new BaseParameters(new File(userDir, "xml/roland/import.xml")) |
| 54 | 54 |
params.load() |
| 55 |
if (!org.txm.Toolbox.isInitialized()) {
|
|
| 56 |
|
|
| 57 |
TokenizerClasses.loadFromNode(params.getTokenizerElement(params.getCorpusElement())); |
|
| 58 |
Toolbox.setParam(Toolbox.INSTALL_DIR,new File("/usr/lib/TXM"));
|
|
| 59 |
//Toolbox.setParam(Toolbox.INSTALL_DIR,new File("C:\\Program Files\\TXM"));//For Windows
|
|
| 60 |
Toolbox.setParam(Toolbox.TREETAGGER_INSTALL_PATH,new File(userDir,"treetagger")); |
|
| 61 |
//Toolbox.setParam(Toolbox.TREETAGGER_INSTALL_PATH,new File("C:\\Program Files\\treetagger"));//for Windows
|
|
| 62 |
Toolbox.setParam(Toolbox.TREETAGGER_MODELS_PATH,new File(userDir,"treetagger/models")); |
|
| 63 |
Toolbox.setParam(Toolbox.METADATA_ENCODING, "UTF-8"); |
|
| 64 |
Toolbox.setParam(Toolbox.METADATA_COLSEPARATOR, ","); |
|
| 65 |
Toolbox.setParam(Toolbox.METADATA_TXTSEPARATOR, "\""); |
|
| 66 |
//Toolbox.setParam(Toolbox.TREETAGGER_MODELS_PATH,new File("C:\\Program Files\\treetagger\\models"));//for Windows
|
|
| 67 |
Toolbox.setParam(Toolbox.USER_TXM_HOME, new File(System.getProperty("user.home"), "TXM"));
|
|
| 68 |
} |
|
| 69 | 55 |
} |
| 70 | 56 |
|
| 71 | 57 |
if (params == null) { println "no parameters. Aborting"; return; }
|
| tmp/org.txm.groovy.core/src/groovy/org/txm/scripts/importer/xtz/XTZImport.groovy (revision 1371) | ||
|---|---|---|
| 116 | 116 |
} |
| 117 | 117 |
} |
| 118 | 118 |
} |
| 119 |
|
|
| 120 |
public static void main(String[] args) {
|
|
| 121 |
|
|
| 122 |
// long start = System.currentTimeMillis() |
|
| 123 |
// |
|
| 124 |
// String userDir = System.getProperty("user.home")
|
|
| 125 |
// Toolbox.setParam(Toolbox.INSTALL_DIR,new File("/usr/lib/TXM"));
|
|
| 126 |
// //Toolbox.setParam(Toolbox.INSTALL_DIR,new File("C:\\Program Files\\TXM"));//For Windows
|
|
| 127 |
// Toolbox.setParam(Toolbox.METADATA_ENCODING, "UTF-8"); |
|
| 128 |
// Toolbox.setParam(Toolbox.METADATA_COLSEPARATOR, ","); |
|
| 129 |
// Toolbox.setParam(Toolbox.METADATA_TXTSEPARATOR, "\""); |
|
| 130 |
// //Toolbox.setParam(Toolbox.TREETAGGER_MODELS_PATH,new File("C:\\Program Files\\treetagger\\models"));//for Windows
|
|
| 131 |
// Toolbox.setParam(Toolbox.USER_TXM_HOME, new File(userDir, "TXM")); |
|
| 132 |
// |
|
| 133 |
// //File importParametersFile = new File("/home/mdecorde/xml/brown-for-xtz/import.xml");
|
|
| 134 |
// // File importParametersFile = new File("/home/mdecorde/xml/qgraalc/qgraal_cw/import.xml"); // new import
|
|
| 135 |
// //File importParametersFile = new File("/home/mdecorde/TXM/corpora/QGRAALXTZ/import.xml"); // corpus update
|
|
| 136 |
// // File importParametersFile = new File("/home/mdecorde/xml/annotation/import.xml"); // new import
|
|
| 137 |
// //File importParametersFile = new File("/home/mdecorde/xml/baiptest/import.xml"); // new import
|
|
| 138 |
// //File importParametersFile = new File("/home/mdecorde/TXM/corpora/BAIP/import.xml"); // corpus update
|
|
| 139 |
// File importParametersFile = new File("/home/mdecorde/xml/xtzsmall/import.xml");
|
|
| 140 |
// BaseParameters b = new BaseParameters(importParametersFile); |
|
| 141 |
// b.getKeyValueParameters().put(ImportKeys.CLEAN, "false") |
|
| 142 |
// b.getKeyValueParameters().put(ImportKeys.MULTITHREAD, "false") |
|
| 143 |
// b.getKeyValueParameters().put(ImportKeys.DEBUG, "true") |
|
| 144 |
// b.getKeyValueParameters().put(ImportKeys.UPDATECORPUS, "false") |
|
| 145 |
// |
|
| 146 |
// XTZImport i = new XTZImport(b); |
|
| 147 |
// i.start(); |
|
| 148 |
// |
|
| 149 |
// println "Done: "+i.isSuccessful |
|
| 150 |
// |
|
| 151 |
// long end = System.currentTimeMillis() |
|
| 152 |
// println "TIME: "+(end-start) |
|
| 153 |
} |
|
| 154 | 119 |
} |
| tmp/org.txm.groovy.core/src/groovy/org/txm/scripts/teitxm/AnnotateWTC.groovy (revision 1371) | ||
|---|---|---|
| 39 | 39 |
import org.txm.utils.DeleteDir; |
| 40 | 40 |
import org.txm.utils.LangDetector; |
| 41 | 41 |
import org.txm.utils.treetagger.TreeTagger |
| 42 |
import org.txm.treetagger.core.preferences.TreeTaggerPreferences |
|
| 42 | 43 |
|
| 43 |
// TODO: Auto-generated Javadoc |
|
| 44 | 44 |
/** |
| 45 | 45 |
* Annotate and replace the TEI-TXM files of the folder $rootDirFile/txm with TreeTagger. |
| 46 | 46 |
* creates $rootDirFile/interp and $rootDirFile/treetagger |
| ... | ... | |
| 107 | 107 |
if(debug) {
|
| 108 | 108 |
println "rootDirFile "+binDir |
| 109 | 109 |
println "txmDir "+txmDir |
| 110 |
println "TREETAGGER INSTALL PATH : "+Toolbox.getPreference(Toolbox.TREETAGGER_INSTALL_PATH)
|
|
| 111 |
println "TREETAGGER MODELS PATH : "+Toolbox.getPreference(Toolbox.TREETAGGER_MODELS_PATH)
|
|
| 110 |
println "TREETAGGER INSTALL PATH : "+TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.INSTALL_PATH))
|
|
| 111 |
println "TREETAGGER MODELS PATH : "+TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.MODELS_PATH))
|
|
| 112 | 112 |
} |
| 113 | 113 |
|
| 114 | 114 |
//test if the Toolbox know TreeTagger |
| 115 |
if (!new File(Toolbox.getPreference(Toolbox.TREETAGGER_INSTALL_PATH)+"/bin/").exists())
|
|
| 115 |
if (!new File(TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.INSTALL_PATH), "bin/").exists())
|
|
| 116 | 116 |
{
|
| 117 |
println("Could not find TreeTagger binaries in "+Toolbox.getPreference(Toolbox.TREETAGGER_INSTALL_PATH)+"/bin/")
|
|
| 117 |
println("Could not find TreeTagger binaries in "+TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.INSTALL_PATH),"bin/")
|
|
| 118 | 118 |
return false; |
| 119 | 119 |
} |
| 120 | 120 |
|
| ... | ... | |
| 127 | 127 |
List<File> files = txmDir.listFiles() |
| 128 | 128 |
|
| 129 | 129 |
// get model file and check it |
| 130 |
File modelfile = new File(Toolbox.getPreference(Toolbox.TREETAGGER_MODELS_PATH), modelfilename);
|
|
| 130 |
File modelfile = new File(TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.MODELS_PATH), modelfilename);
|
|
| 131 | 131 |
if (debug) |
| 132 | 132 |
println "model file : "+modelfile; |
| 133 | 133 |
|
| tmp/org.txm.groovy.core/src/groovy/org/txm/scripts/teitxm/Annotate.groovy (revision 1371) | ||
|---|---|---|
| 108 | 108 |
File modelsDirectory; |
| 109 | 109 |
|
| 110 | 110 |
public Annotate() {
|
| 111 |
modelsDirectory = new File(Toolbox.getPreference(Toolbox.TREETAGGER_MODELS_PATH)); // default models directory is set in the Toolbox
|
|
| 111 |
modelsDirectory = new File(TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.INSTALL_PATH)); // default models directory is set in the Toolbox
|
|
| 112 | 112 |
} |
| 113 | 113 |
|
| 114 | 114 |
/** |
| ... | ... | |
| 267 | 267 |
if (debug) {
|
| 268 | 268 |
println "rootDirFile "+binDir |
| 269 | 269 |
println "txmDir "+txmDir |
| 270 |
println "TREETAGGER INSTALL PATH : "+Toolbox.getPreference(Toolbox.TREETAGGER_INSTALL_PATH)
|
|
| 271 |
println "TREETAGGER MODELS PATH : "+Toolbox.getPreference(Toolbox.TREETAGGER_MODELS_PATH)
|
|
| 270 |
println "TREETAGGER INSTALL PATH : "+TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.INSTALL_PATH)
|
|
| 271 |
println "TREETAGGER MODELS PATH : "+TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.MODELS_PATH)
|
|
| 272 | 272 |
} |
| 273 | 273 |
|
| 274 | 274 |
//test if the Toolbox know TreeTagger |
| ... | ... | |
| 310 | 310 |
|
| 311 | 311 |
if (cancelNow) return; |
| 312 | 312 |
|
| 313 |
File modelDirectory = new File(Toolbox.getPreference(Toolbox.TREETAGGER_MODELS_PATH));
|
|
| 313 |
File modelDirectory = new File(TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.MODELS_PATH));
|
|
| 314 | 314 |
if (!modelDirectory.exists()) {
|
| 315 | 315 |
println "Skipping ANNOTATE: TreeTagger language model file directory not found: "+modelDirectory.getAbsolutePath(); |
| 316 | 316 |
return false; |
| tmp/org.txm.treetagger.core/src/org/txm/importer/xmltxm/AnnotateWTC.groovy (revision 1371) | ||
|---|---|---|
| 36 | 36 |
import org.txm.importer.cwb.* |
| 37 | 37 |
import org.txm.objects.* |
| 38 | 38 |
import org.txm.stat.utils.ConsoleProgressBar |
| 39 |
import org.txm.treetagger.core.preferences.TreeTaggerPreferences |
|
| 39 | 40 |
import org.txm.utils.DeleteDir; |
| 40 | 41 |
import org.txm.utils.LangDetector; |
| 41 | 42 |
import org.txm.utils.io.IOUtils |
| ... | ... | |
| 108 | 109 |
if(debug) {
|
| 109 | 110 |
println "rootDirFile "+binDir |
| 110 | 111 |
println "txmDir "+txmDir |
| 111 |
println "TREETAGGER INSTALL PATH : "+Toolbox.getPreference(Toolbox.TREETAGGER_INSTALL_PATH)
|
|
| 112 |
println "TREETAGGER MODELS PATH : "+Toolbox.getPreference(Toolbox.TREETAGGER_MODELS_PATH)
|
|
| 112 |
println "TREETAGGER INSTALL PATH : "+TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.INSTALL_PATH)
|
|
| 113 |
println "TREETAGGER MODELS PATH : "+TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.MODELS_PATH)
|
|
| 113 | 114 |
} |
| 114 | 115 |
|
| 115 | 116 |
//test if the Toolbox know TreeTagger |
| ... | ... | |
| 127 | 128 |
List<File> files = txmDir.listFiles(IOUtils.FILTER_HIDDEN) |
| 128 | 129 |
|
| 129 | 130 |
// get model file and check it |
| 130 |
File modelfile = new File(Toolbox.getPreference(Toolbox.TREETAGGER_MODELS_PATH), modelfilename);
|
|
| 131 |
File modelfile = new File(TreeTaggerPreferences.getInstance().getBoolean(TreeTaggerPreferences.MODELS_PATH), modelfilename);
|
|
| 131 | 132 |
if (debug) |
| 132 | 133 |
println "model file : "+modelfile; |
| 133 | 134 |
|
Formats disponibles : Unified diff