Révision 38
| SRC/src/fr/triangle/hyperalign/kernel/dico/Dictionary.java (revision 38) | ||
|---|---|---|
| 4 | 4 |
import java.net.URL; |
| 5 | 5 |
import java.util.ArrayList; |
| 6 | 6 |
import java.util.Collections; |
| 7 |
import java.util.Enumeration; |
|
| 7 | 8 |
import java.util.HashMap; |
| 8 | 9 |
import java.util.List; |
| 9 | 10 |
import java.util.Vector; |
| ... | ... | |
| 56 | 57 |
/** |
| 57 | 58 |
* This static value corresponds to the level in the dictionary : "dico" |
| 58 | 59 |
*/ |
| 59 |
public final static String DICO_DICO = "dictionary_dico";
|
|
| 60 |
public final static String DICO_DICO = "dico"; |
|
| 60 | 61 |
/** |
| 61 | 62 |
* This static value corresponds to the level in the dictionary : "form" |
| 62 | 63 |
*/ |
| 63 |
public final static String DICO_FORM = "dictionary_form";
|
|
| 64 |
public final static String DICO_FORM = "form"; |
|
| 64 | 65 |
/** |
| 65 | 66 |
* This static value corresponds to the level in the dictionary : "lemme" |
| 66 | 67 |
*/ |
| 67 |
public final static String DICO_LEMME = "dictionary_lemme";
|
|
| 68 |
public final static String DICO_LEMME = "lemme"; |
|
| 68 | 69 |
/** |
| 69 | 70 |
* This static value corresponds to the level in the dictionary : "occ" |
| 70 | 71 |
*/ |
| 71 |
public final static String DICO_OCC = "dictionary_occ";
|
|
| 72 |
public final static String DICO_OCC = "occ"; |
|
| 72 | 73 |
|
| 73 | 74 |
|
| 74 | 75 |
public final static String OCC = "occ";//Occ |
| ... | ... | |
| 129 | 130 |
public Vector<DictionaryElement> getLemmes(){
|
| 130 | 131 |
Vector<DictionaryElement> children = null; |
| 131 | 132 |
if(root!=null){
|
| 132 |
children = root.children(); |
|
| 133 |
children = root.childrenVect();
|
|
| 133 | 134 |
} |
| 134 | 135 |
return children; |
| 135 | 136 |
} |
| ... | ... | |
| 147 | 148 |
Vector<DictionaryElement> children = getLemmes(); |
| 148 | 149 |
//System.out.println("Terminology.getConcept() with id "+id);
|
| 149 | 150 |
//Concept LEVEL |
| 150 |
if(children.size()>0){
|
|
| 151 |
for(int i = 0 ; i < children.size(); ++i){
|
|
| 152 |
DictionaryElement child = (DictionaryElement) children.get(i); |
|
| 153 |
if(child.getId().equals(id)){
|
|
| 154 |
lemme = child; |
|
| 155 |
//System.out.println("------------- concept => "+child);
|
|
| 156 |
} |
|
| 151 |
for (DictionaryElement child : children) {
|
|
| 152 |
if(child.getId().equals(id)){
|
|
| 153 |
lemme = child; |
|
| 154 |
//System.out.println("------------- concept => "+child);
|
|
| 157 | 155 |
} |
| 158 | 156 |
} |
| 159 | 157 |
return lemme; |
| ... | ... | |
| 186 | 184 |
public Vector<DictionaryElement> getNearLemme(String word, DictionaryElement root){
|
| 187 | 185 |
Vector<DictionaryElement> nearLemmes = new Vector<DictionaryElement>(); |
| 188 | 186 |
boolean isNearLemme = false; |
| 189 |
Vector<DictionaryElement> children = root.children();
|
|
| 187 |
Enumeration<DictionaryElement> children = root.children();
|
|
| 190 | 188 |
if(children!=null){
|
| 191 |
|
|
| 192 | 189 |
//Concept LEVEL |
| 193 |
if(children.size()>0){
|
|
| 194 |
for(int i = 0 ; i < children.size(); ++i){
|
|
| 195 |
DictionaryElement child = (DictionaryElement) children.get(i); |
|
| 196 |
StringDistance stDist = new StringDistance(); |
|
| 197 |
int dist = stDist.computeLevenshteinDistance(child.getName(), word); |
|
| 190 |
while(children.hasMoreElements()){
|
|
| 191 |
DictionaryElement child = (DictionaryElement) children.nextElement(); |
|
| 192 |
StringDistance stDist = new StringDistance(); |
|
| 193 |
int dist = stDist.computeLevenshteinDistance(child.getName(), word); |
|
| 198 | 194 |
|
| 199 |
if(dist <= 2){
|
|
| 200 |
//System.out.println("Distance entre ["+child.getName()+"] et ["+word+"] = "+dist);
|
|
| 195 |
if(dist <= 2){
|
|
| 196 |
//System.out.println("Distance entre ["+child.getName()+"] et ["+word+"] = "+dist);
|
|
| 197 |
nearLemmes.add(child); |
|
| 198 |
} |
|
| 199 |
|
|
| 200 |
else {
|
|
| 201 |
isNearLemme = getForm(word, child, stDist); |
|
| 202 |
if(isNearLemme){
|
|
| 201 | 203 |
nearLemmes.add(child); |
| 202 | 204 |
} |
| 203 |
|
|
| 204 |
else {
|
|
| 205 |
isNearLemme = getForm(word, child, stDist); |
|
| 206 |
if(isNearLemme){
|
|
| 207 |
nearLemmes.add(child); |
|
| 208 |
} |
|
| 209 |
} |
|
| 210 | 205 |
} |
| 211 | 206 |
} |
| 212 | 207 |
} |
| ... | ... | |
| 216 | 211 |
public DictionaryElement getLemme(String word, DictionaryElement root){
|
| 217 | 212 |
DictionaryElement lemme = null; |
| 218 | 213 |
boolean isNearLemme = false; |
| 219 |
Vector<DictionaryElement> children = root.children();
|
|
| 214 |
Enumeration<DictionaryElement> children = root.children();
|
|
| 220 | 215 |
if(children!=null){
|
| 221 | 216 |
|
| 222 | 217 |
//Concept LEVEL |
| 223 |
if(children.size()>0){
|
|
| 224 |
for(int i = 0 ; i < children.size(); ++i){
|
|
| 225 |
DictionaryElement child = (DictionaryElement) children.get(i); |
|
| 226 |
|
|
| 227 |
if(child.getName().equals(word)){
|
|
| 228 |
return child; |
|
| 229 |
} |
|
| 218 |
while(children.hasMoreElements()){
|
|
| 219 |
DictionaryElement child = (DictionaryElement) children.nextElement(); |
|
| 220 |
if(child.getName().equals(word)){
|
|
| 221 |
return child; |
|
| 230 | 222 |
} |
| 231 | 223 |
} |
| 232 | 224 |
} |
| ... | ... | |
| 242 | 234 |
*/ |
| 243 | 235 |
public boolean getForm(String word, DictionaryElement root, StringDistance stDist){
|
| 244 | 236 |
//System.out.println("------------------TerminologyManager.getTerm");
|
| 245 |
Vector<DictionaryElement> children = root.children();
|
|
| 237 |
Enumeration<DictionaryElement> children = root.children();
|
|
| 246 | 238 |
//Concept LEVEL |
| 247 |
if(children.size()>0){
|
|
| 248 |
for(int i = 0 ; i < children.size(); ++i){
|
|
| 249 |
DictionaryElement child = (DictionaryElement) children.get(i); |
|
| 239 |
while(children.hasMoreElements()){
|
|
| 240 |
DictionaryElement child = (DictionaryElement) children.nextElement(); |
|
| 250 | 241 |
|
| 251 |
int dist = stDist.computeLevenshteinDistance(child.getName(), word); |
|
| 252 |
//System.out.println("Distance entre ["+child.getElement()+"] et ["+word+"] = "+dist);
|
|
| 253 |
|
|
| 254 |
//if(child.getElement().equals(word)){
|
|
| 255 |
if(dist < 2){
|
|
| 256 |
return true; |
|
| 257 |
} |
|
| 242 |
int dist = stDist.computeLevenshteinDistance(child.getName(), word); |
|
| 243 |
//System.out.println("Distance entre ["+child.getElement()+"] et ["+word+"] = "+dist);
|
|
| 244 |
//if(child.getElement().equals(word)){
|
|
| 245 |
if(dist < 2){
|
|
| 246 |
return true; |
|
| 258 | 247 |
} |
| 259 | 248 |
} |
| 260 |
|
|
| 261 | 249 |
return false; |
| 262 | 250 |
} |
| 263 | 251 |
|
| ... | ... | |
| 268 | 256 |
public Vector<DictionaryElement> getForms(){
|
| 269 | 257 |
Vector<DictionaryElement> forms = new Vector<DictionaryElement>(); |
| 270 | 258 |
Vector<DictionaryElement> lemmes = getLemmes(); |
| 271 |
for(int i = 0 ; i < lemmes.size() ; ++i){
|
|
| 272 |
DictionaryElement lemme = lemmes.get(i); |
|
| 273 |
Vector<DictionaryElement> formsOfLemme = lemme.children(); |
|
| 274 |
for(int j = 0 ; j < formsOfLemme.size() ; ++j){
|
|
| 275 |
forms.add(formsOfLemme.get(j)); |
|
| 259 |
for (DictionaryElement lemme : lemmes) {
|
|
| 260 |
Enumeration<DictionaryElement> formsOfLemme = lemme.children(); |
|
| 261 |
while(formsOfLemme.hasMoreElements()){
|
|
| 262 |
forms.add(formsOfLemme.nextElement()); |
|
| 276 | 263 |
} |
| 277 | 264 |
} |
| 278 | 265 |
return forms; |
| ... | ... | |
| 289 | 276 |
for(int i = 0 ; i < terms.size() ; ++i){
|
| 290 | 277 |
DictionaryElement termEl = terms.get(i); |
| 291 | 278 |
if(termEl.getName().equals(term)){
|
| 292 |
Vector<DictionaryElement> occs = termEl.children();
|
|
| 293 |
for(int j = 0 ; j < occs.size() ; ++j){
|
|
| 294 |
DictionaryElement occEl = occs.get(j);
|
|
| 279 |
Enumeration<DictionaryElement> occs = termEl.children();
|
|
| 280 |
while(occs.hasMoreElements()){
|
|
| 281 |
DictionaryElement occEl = occs.nextElement();
|
|
| 295 | 282 |
if(occEl.getId().equals(idOcc)){
|
| 296 | 283 |
return occEl; |
| 297 | 284 |
} |
| ... | ... | |
| 345 | 332 |
DictionaryElement term = null; |
| 346 | 333 |
if(element.getReference().equals(Dictionary.DICO_FORM)){
|
| 347 | 334 |
term = element; |
| 348 |
Vector<DictionaryElement> children = term.children();
|
|
| 349 |
System.out.println("Dictionary.getOccurrence() NB OCCS "+children.size()+" for term ["+term+"]");
|
|
| 335 |
Enumeration<DictionaryElement> children = term.children();
|
|
| 336 |
System.out.println("Dictionary.getOccurrence() for term ["+term+"]");
|
|
| 350 | 337 |
|
| 351 |
for(int i = 0 ; i < children.size();++i){
|
|
| 352 |
occEl = children.get(i);
|
|
| 338 |
while(children.hasMoreElements()){
|
|
| 339 |
occEl = children.nextElement();
|
|
| 353 | 340 |
//System.out.println("Terminology.getOccurrence() >>>> "+occEl);
|
| 354 | 341 |
if(occEl.getOccurrence()==occSet){
|
| 355 | 342 |
System.out.println("Dictionary.getOccurrence() >>>> "+occEl);
|
| ... | ... | |
| 391 | 378 |
public Vector<DictionaryElement> getOccurrencesForLemme(DictionaryElement lemme){
|
| 392 | 379 |
Vector<DictionaryElement> occs = new Vector<DictionaryElement>(); |
| 393 | 380 |
if(lemme.children()!=null){
|
| 394 |
for(int i = 0 ; i < lemme.children().size(); ++i){
|
|
| 395 |
DictionaryElement form = (DictionaryElement) lemme.children().get(i);
|
|
| 381 |
while(lemme.children().hasMoreElements()){
|
|
| 382 |
DictionaryElement form = (DictionaryElement) lemme.children().nextElement();
|
|
| 396 | 383 |
if(form.children()!=null){
|
| 397 |
for(int j = 0 ; j < form.children().size(); ++j){
|
|
| 398 |
DictionaryElement occ = (DictionaryElement) form.children().get(j);
|
|
| 384 |
while(form.children().hasMoreElements()){
|
|
| 385 |
DictionaryElement occ = (DictionaryElement) form.children().nextElement();
|
|
| 399 | 386 |
occs.add(occ); |
| 400 | 387 |
} |
| 401 | 388 |
} |
| ... | ... | |
| 430 | 417 |
*/ |
| 431 | 418 |
public DictionaryElement getFormWithLemme(DictionaryElement lemme, String formStr){
|
| 432 | 419 |
DictionaryElement form = null; |
| 433 |
Vector<DictionaryElement> forms = lemme.children();
|
|
| 434 |
for(int i = 0 ; i < forms.size() ; ++i){
|
|
| 435 |
form = forms.get(i);
|
|
| 420 |
Enumeration<DictionaryElement> forms = lemme.children();
|
|
| 421 |
while(forms.hasMoreElements()){
|
|
| 422 |
form = forms.nextElement();
|
|
| 436 | 423 |
if(formStr.equals(form.getName())){
|
| 437 | 424 |
return form; |
| 438 | 425 |
} |
| ... | ... | |
| 483 | 470 |
public DictionaryElement checkElementAsChild(String elementName, DictionaryElement parent){
|
| 484 | 471 |
//System.out.println("Terminology.checkElementAsChild() >> "+elementName+" FROM "+parent.getName());
|
| 485 | 472 |
if(parent!=null){
|
| 486 |
Vector<DictionaryElement> children = parent.children();
|
|
| 473 |
Enumeration<DictionaryElement> children = parent.children();
|
|
| 487 | 474 |
if(children!=null){
|
| 488 |
for(int i = 0 ; i< children.size(); ++i){
|
|
| 489 |
DictionaryElement child = children.get(i);
|
|
| 475 |
while(children.hasMoreElements()){
|
|
| 476 |
DictionaryElement child = children.nextElement();
|
|
| 490 | 477 |
if(child.getName().equals(elementName)){
|
| 491 | 478 |
return child; |
| 492 | 479 |
} |
| ... | ... | |
| 499 | 486 |
public DictionaryElement checkLemme(String idLemme, DictionaryElement root){
|
| 500 | 487 |
//System.out.println("Terminology.checkElementAsChild() >> "+elementName+" FROM "+parent.getName());
|
| 501 | 488 |
if(root!=null){
|
| 502 |
Vector<DictionaryElement> children = root.children();
|
|
| 489 |
Enumeration<DictionaryElement> children = root.children();
|
|
| 503 | 490 |
if(children!=null){
|
| 504 |
for(int i = 0 ; i< children.size(); ++i){
|
|
| 505 |
DictionaryElement child = children.get(i);
|
|
| 491 |
while(children.hasMoreElements()){
|
|
| 492 |
DictionaryElement child = children.nextElement();
|
|
| 506 | 493 |
if(child.getId().equals(idLemme)){
|
| 507 | 494 |
return child; |
| 508 | 495 |
} |
| ... | ... | |
| 520 | 507 |
*/ |
| 521 | 508 |
public DictionaryElement checkAndAddForm(String newForm, DictionaryElement lemme){
|
| 522 | 509 |
if(lemme!=null){
|
| 523 |
Vector<DictionaryElement> forms = lemme.children();
|
|
| 510 |
Enumeration<DictionaryElement> forms = lemme.children();
|
|
| 524 | 511 |
if(forms!=null){
|
| 525 |
for(int i = 0 ; i< forms.size(); ++i){
|
|
| 526 |
DictionaryElement formOfLemme = forms.get(i);
|
|
| 512 |
while(forms.hasMoreElements()){
|
|
| 513 |
DictionaryElement formOfLemme = forms.nextElement();
|
|
| 527 | 514 |
if(formOfLemme.getName().equals(newForm)){
|
| 528 | 515 |
//System.out.println(newTerm+" is already in the terminology !"); |
| 529 | 516 |
return formOfLemme; |
Formats disponibles : Unified diff