Révision 3561
| TXM/trunk/org.txm.conllu.rcp/src/org/txm/conllu/rcp/commands/ImportCoNLLUAnnotationsFromFile.java (revision 3561) | ||
|---|---|---|
| 64 | 64 |
@Option(name = "propertiesPrefix", usage = "optional prefix for the properties to create", widget = "String", required = true, def = "ud-") |
| 65 | 65 |
String propertiesPrefix; |
| 66 | 66 |
|
| 67 |
@Option(name = "udPropertiesToImport", usage = "to create the ud properties", widget = "StringArrayMultiple", metaVar="form lemma upos xpos feats head deprel deps misc", required = true, def = "form,lemma,upos,xpos,feats,head,deprel,deps,misc") |
|
| 67 |
@Option(name = "udPropertiesToImport", usage = "to create the ud properties", widget = "StringArrayMultiple", metaVar="id form lemma upos xpos feats head deprel deps misc", required = true, def = "form,lemma,upos,xpos,feats,head,deprel,deps,misc")
|
|
| 68 | 68 |
String udPropertiesToImport; |
| 69 | 69 |
|
| 70 | 70 |
@Option(name = "overwrite_cqp_properties", usage = "if set the CQP properties are replaced", widget = "Boolean", required = true, def = "false") |
| TXM/trunk/org.txm.conllu.rcp/src/org/txm/conllu/rcp/commands/ImportCoNLLUAnnotationsFromDirectory.java (revision 3561) | ||
|---|---|---|
| 72 | 72 |
@Option(name = "propertiesPrefix", usage = "optional prefix for the properties to create", widget = "String", required = true, def = "ud-") |
| 73 | 73 |
String propertiesPrefix; |
| 74 | 74 |
|
| 75 |
@Option(name = "udPropertiesToImport", usage = "to create the ud properties", widget = "StringArrayMultiple", metaVar="form lemma upos xpos feats head deprel deps misc", required = true, def = "form,lemma,upos,xpos,feats,head,deprel,deps,misc") |
|
| 75 |
@Option(name = "udPropertiesToImport", usage = "to create the ud properties", widget = "StringArrayMultiple", metaVar="id form lemma upos xpos feats head deprel deps misc", required = true, def = "form,lemma,upos,xpos,feats,head,deprel,deps,misc")
|
|
| 76 | 76 |
String udPropertiesToImport; |
| 77 | 77 |
|
| 78 | 78 |
@Option(name = "overwrite_cqp_properties", usage = "if set the CQP properties are replaced", widget = "Boolean", required = true, def = "false") |
| ... | ... | |
| 166 | 166 |
int nTextProcessed = 0; |
| 167 | 167 |
int nWordsInserted = 0; |
| 168 | 168 |
ConsoleProgressBar cpb = new ConsoleProgressBar(files.length); |
| 169 |
|
|
| 170 |
if (files.length == 0) {
|
|
| 171 |
Log.warning(NLS.bind("** No *.conllu files found in {0}. Aborting.", conlluDirectory));
|
|
| 172 |
return 0; |
|
| 173 |
} |
|
| 174 |
|
|
| 169 | 175 |
for (File coonluFile : files) {
|
| 170 | 176 |
cpb.tick(); |
| 171 | 177 |
nWordsInserted += ImportCoNLLUAnnotations._importAnnotations(coonluFile, mainCorpus, propertiesPrefix, null, normalizeWordIds, headPropertiesToProject, depsPropertiesToProject, udPropertiesToImport); |
| TXM/trunk/org.txm.rcp/rcpapplication.product (revision 3561) | ||
|---|---|---|
| 80 | 80 |
</launcher> |
| 81 | 81 |
|
| 82 | 82 |
<vm> |
| 83 |
<linux include="false">org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-1.7</linux>
|
|
| 84 |
<macos include="false">org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-1.7</macos>
|
|
| 83 |
<linux include="false">org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-1.8</linux>
|
|
| 84 |
<macos include="false">org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-1.8</macos>
|
|
| 85 | 85 |
<solaris include="false">org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-1.7</solaris> |
| 86 | 86 |
<windows include="false">org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-1.8</windows> |
| 87 | 87 |
</vm> |
| TXM/trunk/org.txm.connlu.core/src/org/txm/conllu/core/function/ImportCoNLLUAnnotations.java (revision 3561) | ||
|---|---|---|
| 30 | 30 |
textId = coonluFile.getName().substring(0, coonluFile.getName().length() - 7); |
| 31 | 31 |
} |
| 32 | 32 |
|
| 33 |
for (String p : headPropertiesToProject) {
|
|
| 34 |
udPropertiesToImport.add("head-"+p);
|
|
| 35 |
} |
|
| 36 |
|
|
| 37 |
for (String p : depsPropertiesToProject) {
|
|
| 38 |
udPropertiesToImport.add("dep-"+p);
|
|
| 39 |
} |
|
| 40 |
|
|
| 33 | 41 |
Log.info("** processing text: " + textId);
|
| 34 | 42 |
Text text = mainCorpus.getProject().getText(textId); |
| 35 | 43 |
if (text == null) {
|
| ... | ... | |
| 181 | 189 |
|
| 182 | 190 |
return processor.getNInsertions(); |
| 183 | 191 |
} |
| 184 |
|
|
| 192 |
|
|
| 185 | 193 |
public static void buildPropertiesProjections(LinkedHashMap<String, HashMap<String, String>> sentenceProperties, Set<String> headPropertiesToProject, Set<String> depsPropertiesToProject) throws IOException, XMLStreamException {
|
| 186 | 194 |
|
| 187 | 195 |
//System.out.println("Building "+headPropertiesToProject+" and "+depsPropertiesToProject+" in "+sentenceProperties);
|
| ... | ... | |
| 194 | 202 |
for (String wheadid : sentenceProperties.keySet()) { // find the word head properties
|
| 195 | 203 |
|
| 196 | 204 |
HashMap<String, String> headProperties = sentenceProperties.get(wheadid); |
| 197 |
|
|
| 205 |
|
|
| 198 | 206 |
if (headProperties.get("id").equals(properties.get("head"))) { // a word which head is the current word id
|
| 199 | 207 |
// { "form", "lemma", "upos", "xpos", "feats", "deprel", "deps", "misc" };
|
| 200 | 208 |
if (headPropertiesToProject.contains("form")) properties.put("head-form", headProperties.get("form"));
|
| ... | ... | |
| 259 | 267 |
HashMap<String, String> properties = sentenceProperties.get(id); |
| 260 | 268 |
ArrayList<String> propertyNames = new ArrayList<String>(properties.keySet()); |
| 261 | 269 |
|
| 262 |
for (String p : propertyNames) {
|
|
| 270 |
for (String p : propertyNames) { // add # to properties to inject, and remove properties not to inject
|
|
| 263 | 271 |
|
| 264 |
if (udPropertiesToImport != null && udPropertiesToImport.size() > 0 && !udPropertiesToImport.contains(p)) continue; // keep only the sentence properties to import |
|
| 265 |
|
|
| 266 |
if (p.startsWith("#")) continue; //property already prefixed with #, do not add the prefix
|
|
| 267 |
|
|
| 272 |
if (p.startsWith("#")) {
|
|
| 273 |
continue; //property already prefixed with #, do not add the prefix |
|
| 274 |
} |
|
| 275 |
else if (udPropertiesToImport != null && udPropertiesToImport.size() > 0 && !udPropertiesToImport.contains(p)) {; // keep only the sentence properties to import
|
|
| 276 |
properties.remove(p); |
|
| 277 |
continue; //property already prefixed with #, do not add the prefix |
|
| 278 |
} |
|
| 268 | 279 |
String value = properties.get(p); |
| 269 | 280 |
properties.remove(p); |
| 270 | 281 |
properties.put("#"+propertiesPrefix+p, value);
|
| 271 | 282 |
} |
| 283 |
|
|
| 272 | 284 |
processor.addProperty(id, properties); |
| 273 | 285 |
} |
| 274 | 286 |
|
Formats disponibles : Unified diff