Révision 1065
| tmp/org.txm.treetagger.core/src/org/txm/importer/xmltxm/Annotate.groovy (revision 1065) | ||
|---|---|---|
| 209 | 209 |
File infile = ttsrcfile; |
| 210 | 210 |
File outfile = ttoutfile; |
| 211 | 211 |
|
| 212 |
// TODO: why org.txm.utils.treetagger.TreeTagger tt = ..., throw IllegalAccessError ??? |
|
| 213 |
String opt = TreeTaggerPreferences.getString(TreeTaggerPreferences.OPTIONS, TreeTaggerPreferences.PREFERENCES_NODE); |
|
| 214 |
if (options == null) {
|
|
| 215 |
if (opt == null) opt = ""; |
|
| 216 |
options = opt.split(" ");
|
|
| 217 |
} |
|
| 218 |
|
|
| 219 | 212 |
def tt = new TreeTagger(TreeTaggerPreferences.getString(TreeTaggerPreferences.INSTALL_PATH, TreeTaggerPreferences.PREFERENCES_NODE)+"/bin/", options); |
| 220 | 213 |
tt.settoken(); |
| 221 | 214 |
tt.setlemma(); |
| 222 |
tt.setquiet(); |
|
| 223 | 215 |
tt.setsgml(); |
| 224 |
tt.setnounknown(); |
|
| 216 |
if (TreeTaggerPreferences.getBoolean(TreeTaggerPreferences.OPTIONS_UNKNOWN, TreeTaggerPreferences.PREFERENCES_NODE)) {
|
|
| 217 |
tt.setnounknown(); |
|
| 218 |
} |
|
| 225 | 219 |
tt.seteostag("<s>");
|
| 226 |
tt.debug(debug); |
|
| 220 |
if (TreeTaggerPreferences.getBoolean(TreeTaggerPreferences.OPTIONS_DEBUG, TreeTaggerPreferences.PREFERENCES_NODE)) {
|
|
| 221 |
tt.debug(true); |
|
| 222 |
} else {
|
|
| 223 |
tt.setquiet(); |
|
| 224 |
} |
|
| 225 |
if (TreeTaggerPreferences.getBoolean(TreeTaggerPreferences.OPTIONS_CAPHEURISTIC, TreeTaggerPreferences.PREFERENCES_NODE)) {
|
|
| 226 |
tt.setcapheuristics(); |
|
| 227 |
} |
|
| 228 |
if (TreeTaggerPreferences.getBoolean(TreeTaggerPreferences.OPTIONS_HYPHENHEURISTIC, TreeTaggerPreferences.PREFERENCES_NODE)) {
|
|
| 229 |
tt.sethyphenheuristics(); |
|
| 230 |
} |
|
| 231 |
if (TreeTaggerPreferences.getBoolean(TreeTaggerPreferences.OPTIONS_PROB, TreeTaggerPreferences.PREFERENCES_NODE)) {
|
|
| 232 |
tt.setprob(); |
|
| 233 |
} |
|
| 234 |
|
|
| 235 |
String lex = TreeTaggerPreferences.getString(TreeTaggerPreferences.OPTIONS_LEX, TreeTaggerPreferences.PREFERENCES_NODE); |
|
| 236 |
if (lex !=null && lex.length() > 0) {
|
|
| 237 |
tt.setlex(lex); |
|
| 238 |
} |
|
| 239 |
String wc = TreeTaggerPreferences.getString(TreeTaggerPreferences.OPTIONS_WC, TreeTaggerPreferences.PREFERENCES_NODE); |
|
| 240 |
if (wc !=null && wc.length() > 0) {
|
|
| 241 |
tt.setwc(wc); |
|
| 242 |
} |
|
| 227 | 243 |
tt.treetagger(modelfile.getAbsolutePath(), infile.getAbsolutePath(), outfile.getAbsolutePath()) |
| 228 | 244 |
infile.delete(); |
| 229 | 245 |
} catch(Exception e) {
|
| 230 | 246 |
Log.printStackTrace(e); |
| 231 |
System.out.println("Failed to apply TreeTagger on $f input file with the $modelfile model file.");
|
|
| 247 |
System.out.println("Failed to apply TreeTagger on $ttsrcfile input file with the $modelfile model file.");
|
|
| 232 | 248 |
return false; |
| 233 | 249 |
} |
| 234 | 250 |
return true; |
| tmp/org.txm.treetagger.core/src/org/txm/treetagger/core/TreeTaggerEngine.java (revision 1065) | ||
|---|---|---|
| 32 | 32 |
return false; |
| 33 | 33 |
} |
| 34 | 34 |
|
| 35 |
options = TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.OPTIONS).split(" ");
|
|
| 35 |
//options = TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.OPTIONS).split(" ");
|
|
| 36 | 36 |
|
| 37 | 37 |
return ttBinaryDirectory != null && ttBinaryDirectory.exists() && ttModelsDirectory != null && ttModelsDirectory.exists(); |
| 38 | 38 |
} |
| tmp/org.txm.treetagger.core/src/org/txm/treetagger/core/preferences/TreeTaggerPreferences.java (revision 1065) | ||
|---|---|---|
| 42 | 42 |
*/ |
| 43 | 43 |
public static final String MODELS_PATH = "models_path"; //$NON-NLS-1$ |
| 44 | 44 |
|
| 45 |
public static final String OPTIONS = "options"; //$NON-NLS-1$ |
|
| 45 |
//public static final String OPTIONS = "options"; //$NON-NLS-1$
|
|
| 46 | 46 |
|
| 47 |
public static final String OPTIONS_LEX = "lex"; //$NON-NLS-1$ |
|
| 48 |
public static final String OPTIONS_WC = "wc"; //$NON-NLS-1$ |
|
| 49 |
public static final String OPTIONS_UNKNOWN = "unknown"; //$NON-NLS-1$ |
|
| 50 |
public static final String OPTIONS_DEBUG = "debug"; //$NON-NLS-1$ |
|
| 51 |
public static final String OPTIONS_CAPHEURISTIC = "cap_heuristic"; //$NON-NLS-1$ |
|
| 52 |
public static final String OPTIONS_HYPHENHEURISTIC = "hyphen_heuristic"; //$NON-NLS-1$ |
|
| 53 |
public static final String OPTIONS_PROB = "prob"; //$NON-NLS-1$ |
|
| 54 |
|
|
| 47 | 55 |
public static final String FIX_APOSTROPHES = "fix_apostrophes"; //$NON-NLS-1$ |
| 48 | 56 |
|
| 49 | 57 |
|
| ... | ... | |
| 75 | 83 |
preferences.put(INSTALL_PATH, installPath); |
| 76 | 84 |
preferences.put(MODELS_PATH, installPath + "/models"); //$NON-NLS-1$ |
| 77 | 85 |
preferences.putBoolean(FIX_APOSTROPHES, false); |
| 78 |
|
|
| 86 |
|
|
| 87 |
preferences.putBoolean(OPTIONS_DEBUG, false); |
|
| 88 |
preferences.putBoolean(OPTIONS_CAPHEURISTIC, false); |
|
| 89 |
preferences.putBoolean(OPTIONS_HYPHENHEURISTIC, false); |
|
| 90 |
preferences.putBoolean(OPTIONS_UNKNOWN, true); |
|
| 91 |
preferences.putBoolean(OPTIONS_PROB, false); |
|
| 92 |
|
|
| 79 | 93 |
// FIXME: need to validate this code + need to check if it's still useful |
| 80 | 94 |
String bversion = TreeTaggerPreferences.getInstance().getString(INSTALLED_BINARIES_VERSION); |
| 81 | 95 |
String mversion = TreeTaggerPreferences.getInstance().getString(INSTALLED_MODELS_VERSION); |
| ... | ... | |
| 97 | 111 |
BufferedReader reader = IOUtils.getReader(previousPreferenceFile, "ISO-8859-1"); //$NON-NLS-1$ |
| 98 | 112 |
previousProperties.load(reader); |
| 99 | 113 |
|
| 100 |
String [] keys= {INSTALL_PATH, MODELS_PATH, OPTIONS};
|
|
| 114 |
String [] keys= {INSTALL_PATH, MODELS_PATH};
|
|
| 101 | 115 |
for (String k : keys) {
|
| 102 | 116 |
if (previousProperties.getProperty(previousProperties.getProperty(k)) != null) {
|
| 103 | 117 |
preferences.put(k, installPath); |
| tmp/org.txm.rcp/OSGI-INF/l10n/bundle.properties (revision 1065) | ||
|---|---|---|
| 3 | 3 |
#Thu Jul 25 14:46:50 CEST 2013 |
| 4 | 4 |
|
| 5 | 5 |
Bundle-Name = org.txm.rcp |
| 6 |
|
|
| 6 | 7 |
Bundle-Vendor = Textometrie.org |
| 7 | 8 |
|
| 8 | 9 |
category.description = Textometry Functions |
| ... | ... | |
| 442 | 443 |
|
| 443 | 444 |
page.name = User |
| 444 | 445 |
page.name.0 = Advanced |
| 446 |
page.name.1 = Debug |
|
| 445 | 447 |
page.name.10 = Correspondence Analysis |
| 446 | 448 |
page.name.12 = Export |
| 447 | 449 |
page.name.16 = Referencer |
| tmp/org.txm.rcp/OSGI-INF/l10n/bundle_ru.properties (revision 1065) | ||
|---|---|---|
| 6 | 6 |
category.name = TXM |
| 7 | 7 |
category.name.0 = TXM |
| 8 | 8 |
|
| 9 |
|
|
| 10 | 9 |
command.description = \u00D0\u009F\u00D0\u00B5\u00D1\u0080\u00D0\u00B5\u00D0\u00B7\u00D0\u00B0\u00D0\u00B3\u00D1\u0080\u00D1\u0083\u00D0\u00B7\u00D0\u00B8\u00D1\u0082\u00D1\u008C TXM |
| 11 | 10 |
command.description.0 = \u00D0\u009A\u00D0\u00BE\u00D0\u00BD\u00D0\u00B2\u00D0\u00B5\u00D1\u0080\u00D1\u0082\u00D0\u00B8\u00D1\u0080\u00D0\u00BE\u00D0\u00B2\u00D0\u00B0\u00D1\u0082\u00D1\u008C \u00D1\u0081\u00D1\u0082\u00D0\u00B0\u00D1\u0080\u00D1\u008B\u00D0\u00B9 \u00D0\u00B1\u00D0\u00B8\u00D0\u00BD\u00D0\u00B0\u00D1\u0080\u00D0\u00BD\u00D1\u008B\u00D0\u00B9 \u00D1\u0084\u00D0\u00B0\u00D0\u00B9\u00D0\u00BB |
| 12 | 11 |
command.description.1 = Reset corpora |
| ... | ... | |
| 331 | 330 |
|
| 332 | 331 |
page.name = \u00D0\u009F\u00D0\u00BE\u00D0\u00BB\u00D1\u008C\u00D0\u00B7\u00D0\u00BE\u00D0\u00B2\u00D0\u00B0\u00D1\u0082\u00D0\u00B5\u00D0\u00BB\u00D1\u008C |
| 333 | 332 |
page.name.0 = \u00D0\u0094\u00D0\u00BE\u00D0\u00BF\u00D0\u00BE\u00D0\u00BB\u00D0\u00BD\u00D0\u00B8\u00D1\u0082\u00D0\u00B5\u00D0\u00BB\u00D1\u008C\u00D0\u00BD\u00D0\u00BE |
| 333 |
page.name.1 = Debug |
|
| 334 | 334 |
page.name.10 = \u00D0\u00B0\u00D0\u00BD\u00D0\u00B0\u00D0\u00BB\u00D0\u00B8\u00D0\u00B7 \u00D1\u0081\u00D0\u00BE\u00D0\u00BE\u00D1\u0082\u00D0\u00B2\u00D0\u00B5\u00D1\u0082\u00D1\u0081\u00D1\u0082\u00D0\u00B2\u00D0\u00B8\u00D0\u00B9 |
| 335 | 335 |
page.name.12 = \u00D0\u00AD\u00D0\u00BA\u00D1\u0081\u00D0\u00BF\u00D0\u00BE\u00D1\u0080\u00D1\u0082 |
| 336 | 336 |
page.name.16 = \u00D0\u00A1\u00D1\u0081\u00D1\u008B\u00D0\u00BB\u00D0\u00BA\u00D0\u00B8 |
| tmp/org.txm.rcp/OSGI-INF/l10n/bundle_fr.properties (revision 1065) | ||
|---|---|---|
| 3 | 3 |
#Thu Jul 25 14:48:07 CEST 2013 |
| 4 | 4 |
|
| 5 | 5 |
Bundle-Name = org.txm.rcp |
| 6 |
|
|
| 6 | 7 |
Bundle-Vendor = Textometrie.org |
| 7 | 8 |
|
| 8 | 9 |
category.description = Commandes de Textometrie |
| ... | ... | |
| 462 | 463 |
|
| 463 | 464 |
page.name = Utilisateur |
| 464 | 465 |
page.name.0 = Avanc\u00E9 |
| 466 |
page.name.1 = Debug |
|
| 465 | 467 |
page.name.10 = Analyse des correspondances |
| 466 | 468 |
page.name.11 = Cooccurrences |
| 467 | 469 |
page.name.12 = Export |
| tmp/org.txm.treetagger.rcp/src/org/txm/treetagger/rcp/handlers/Train.java (revision 1065) | ||
|---|---|---|
| 342 | 342 |
|
| 343 | 343 |
// Call treetagger-train |
| 344 | 344 |
if (ttSrcFile.exists() && lexique2.exists() && openclassfile.exists()) {
|
| 345 |
System.out.println("Running ");
|
|
| 345 |
System.out.println("Running treetagger-train...");
|
|
| 346 | 346 |
String treetaggerBinDirectory = new File(TreeTaggerPreferences.getInstance().getString(TreeTaggerPreferences.INSTALL_PATH), "bin").getAbsolutePath(); |
| 347 | 347 |
if (!treetaggerBinDirectory.endsWith("/")) treetaggerBinDirectory += "/";
|
| 348 | 348 |
|
| tmp/org.txm.treetagger.rcp/src/org/txm/treetagger/rcp/preferences/TreeTaggerPreferencePage.java (revision 1065) | ||
|---|---|---|
| 29 | 29 |
|
| 30 | 30 |
import org.eclipse.jface.preference.BooleanFieldEditor; |
| 31 | 31 |
import org.eclipse.jface.preference.DirectoryFieldEditor; |
| 32 |
import org.eclipse.jface.preference.StringFieldEditor; |
|
| 32 |
import org.eclipse.jface.preference.FileFieldEditor; |
|
| 33 |
import org.eclipse.swt.SWT; |
|
| 34 |
import org.eclipse.swt.layout.GridData; |
|
| 35 |
import org.eclipse.swt.layout.GridLayout; |
|
| 36 |
import org.eclipse.swt.widgets.Group; |
|
| 33 | 37 |
import org.eclipse.ui.IWorkbench; |
| 34 | 38 |
import org.txm.rcp.preferences.RCPPreferencesPage; |
| 35 | 39 |
import org.txm.rcp.preferences.TXMPreferenceStore; |
| ... | ... | |
| 44 | 48 |
* |
| 45 | 49 |
*/ |
| 46 | 50 |
public class TreeTaggerPreferencePage extends RCPPreferencesPage {
|
| 47 |
|
|
| 48 | 51 |
|
| 49 | 52 |
@Override |
| 50 | 53 |
public void createFieldEditors() {
|
| 54 |
|
|
| 51 | 55 |
this.addField(new DirectoryFieldEditor(TreeTaggerPreferences.INSTALL_PATH, TreeTaggerUIMessages.TreeTaggerPreferencePage_1, this.getFieldEditorParent())); |
| 52 | 56 |
this.addField(new DirectoryFieldEditor(TreeTaggerPreferences.MODELS_PATH, TreeTaggerUIMessages.TreeTaggerPreferencePage_2, this.getFieldEditorParent())); |
| 53 |
this.addField(new StringFieldEditor(TreeTaggerPreferences.OPTIONS, TreeTaggerUIMessages.TreeTaggerPreferencePage_3, this.getFieldEditorParent())); |
|
| 57 |
|
|
| 58 |
//this.addField(new StringFieldEditor(TreeTaggerPreferences.OPTIONS, TreeTaggerUIMessages.TreeTaggerPreferencePage_3, this.getFieldEditorParent())); |
|
| 59 |
Group runOptions = new Group(this.getFieldEditorParent(), SWT.BORDER); |
|
| 60 |
runOptions.setText("TreeTagger process run options");
|
|
| 61 |
GridData gdata = new GridData(GridData.FILL, GridData.FILL, true, false, 3, 1); |
|
| 62 |
runOptions.setLayoutData(gdata); |
|
| 63 |
runOptions.setLayout(new GridLayout(3, false)); |
|
| 64 |
|
|
| 65 |
this.addField(new BooleanFieldEditor(TreeTaggerPreferences.OPTIONS_UNKNOWN, "Print the token rather than <unknown> for unknown lemmas", runOptions)); |
|
| 66 |
this.addField(new BooleanFieldEditor(TreeTaggerPreferences.OPTIONS_CAPHEURISTIC, "Look up unknown capitalized words in the list of lower-case words", runOptions)); |
|
| 67 |
this.addField(new BooleanFieldEditor(TreeTaggerPreferences.OPTIONS_HYPHENHEURISTIC, "Turn on the heuristics fur guessing the parts of speech of unknown hyphenated words", runOptions)); |
|
| 68 |
this.addField(new BooleanFieldEditor(TreeTaggerPreferences.OPTIONS_PROB, "Print tag probabilities", runOptions)); |
|
| 69 |
this.addField(new FileFieldEditor(TreeTaggerPreferences.OPTIONS_LEX, "Read auxiliary lexicon entries from a file", runOptions)); |
|
| 70 |
this.addField(new FileFieldEditor(TreeTaggerPreferences.OPTIONS_WC, "Read a word-class automaton from a file", runOptions)); |
|
| 71 |
|
|
| 72 |
this.addField(new BooleanFieldEditor(TreeTaggerPreferences.OPTIONS_DEBUG, "Activate debug messages", runOptions)); |
|
| 73 |
runOptions.setLayout(new GridLayout(3, false)); |
|
| 74 |
|
|
| 54 | 75 |
this.addField(new BooleanFieldEditor(TreeTaggerPreferences.FIX_APOSTROPHES, TreeTaggerUIMessages.TreeTaggerPreferencePage_0, this.getFieldEditorParent())); |
| 55 | 76 |
} |
| 56 | 77 |
|
| 57 |
|
|
| 58 | 78 |
@Override |
| 59 | 79 |
public void init(IWorkbench workbench) {
|
| 60 | 80 |
this.setPreferenceStore(new TXMPreferenceStore(TreeTaggerPreferences.getInstance().getPreferencesNodeQualifier())); |
Formats disponibles : Unified diff