Révision 2408
| tmp/org.txm.searchengine.cqp.core/src/org/txm/searchengine/cqp/corpus/Subcorpus.java (revision 2408) | ||
|---|---|---|
| 63 | 63 |
* @author jmague |
| 64 | 64 |
*/ |
| 65 | 65 |
public class Subcorpus extends CQPCorpus {
|
| 66 |
|
|
| 66 |
|
|
| 67 | 67 |
/** |
| 68 | 68 |
* Query used to build the corpus. |
| 69 | 69 |
*/ |
| 70 |
@Parameter(key=TXMPreferences.QUERY)
|
|
| 70 |
@Parameter(key = TXMPreferences.QUERY)
|
|
| 71 | 71 |
protected CQLQuery pQuery; |
| 72 | 72 |
|
| 73 | 73 |
protected QueryResult qresult; |
| ... | ... | |
| 81 | 81 |
public Subcorpus(CQPCorpus corpus) {
|
| 82 | 82 |
super(corpus); |
| 83 | 83 |
} |
| 84 |
|
|
| 84 |
|
|
| 85 | 85 |
/** |
| 86 | 86 |
* |
| 87 | 87 |
* @param partition |
| ... | ... | |
| 89 | 89 |
public Subcorpus(Partition partition) {
|
| 90 | 90 |
super(partition); |
| 91 | 91 |
} |
| 92 |
|
|
| 92 |
|
|
| 93 | 93 |
/** |
| 94 | 94 |
* |
| 95 | 95 |
* @param parametersNodePath |
| ... | ... | |
| 97 | 97 |
public Subcorpus(String parametersNodePath) {
|
| 98 | 98 |
super(parametersNodePath); |
| 99 | 99 |
} |
| 100 |
|
|
| 100 |
|
|
| 101 | 101 |
/** |
| 102 | 102 |
* Instantiates a new subcorpus. |
| 103 | 103 |
* |
| ... | ... | |
| 108 | 108 |
* @param parent the parent |
| 109 | 109 |
* @param query the query |
| 110 | 110 |
* @throws InvalidCqpIdException the invalid cqp id exception |
| 111 |
* {@link CQPCorpus#createSubcorpus(CQLQuery, String)}
|
|
| 111 |
* {@link CQPCorpus#createSubcorpus(CQLQuery, String)}
|
|
| 112 | 112 |
*/ |
| 113 |
@Override |
|
| 113 | 114 |
protected boolean _compute() throws Exception {
|
| 114 | 115 |
if (this.pQuery != null) {
|
| 115 | 116 |
this.qresult = null; // reset |
| ... | ... | |
| 120 | 121 |
} |
| 121 | 122 |
return qresult != null; |
| 122 | 123 |
} |
| 123 |
|
|
| 124 | 124 |
|
| 125 |
|
|
| 125 | 126 |
@Override |
| 126 | 127 |
public String getComputingStartMessage() {
|
| 127 | 128 |
return TXMCoreMessages.bind("Création du sous-corpus {0} dans le corpus {1}...", this.getSimpleName(), this.getParent().getName());
|
| 128 | 129 |
} |
| 129 |
|
|
| 130 |
|
|
| 130 | 131 |
@Override |
| 131 | 132 |
public String getComputingDoneMessage() {
|
| 132 | 133 |
if (hasBeenComputedOnce()) {
|
| 133 | 134 |
return getDetails(); |
| 134 |
} else {
|
|
| 135 |
} |
|
| 136 |
else {
|
|
| 135 | 137 |
return ""; |
| 136 | 138 |
} |
| 137 |
//return TXMCoreMessages.common_done; |
|
| 139 |
// return TXMCoreMessages.common_done;
|
|
| 138 | 140 |
} |
| 139 | 141 |
|
| 140 | 142 |
/* |
| 141 | 143 |
* retro compatibility method for import.xml file |
| 142 |
* |
|
| 143 | 144 |
* @see org.txm.searchengine.cqp.corpus.Corpus#getLocale() |
| 144 | 145 |
*/ |
| 145 | 146 |
@Override |
| 146 | 147 |
protected boolean _load(Element e) {
|
| 147 | 148 |
if (e != null) {
|
| 148 | 149 |
this.userName = e.getAttribute(NAME); |
| 149 |
this.pID = "S"+getNextSubcorpusCounter();
|
|
| 150 |
this.pID = "S" + getNextSubcorpusCounter();
|
|
| 150 | 151 |
this.pQuery = new CQLQuery(e.getAttribute("query")); //$NON-NLS-1$
|
| 151 |
|
|
| 152 |
|
|
| 152 | 153 |
NodeList subcorpusElems = e.getElementsByTagName("subcorpus"); //$NON-NLS-1$
|
| 153 |
for (int i = 0 ; i < subcorpusElems.getLength() ; i++) {
|
|
| 154 |
for (int i = 0; i < subcorpusElems.getLength(); i++) {
|
|
| 154 | 155 |
Element subcorpusElem = (Element) subcorpusElems.item(i); |
| 155 | 156 |
try {
|
| 156 | 157 |
Subcorpus subcorp = new Subcorpus(this); |
| 157 | 158 |
subcorp._load(subcorpusElem); |
| 158 |
} catch (Exception ex) {
|
|
| 159 |
} |
|
| 160 |
catch (Exception ex) {
|
|
| 159 | 161 |
Log.warning(this.pID + TXMCoreMessages.failedToRestoreSubcorpus + subcorpusElem.getAttribute("name")
|
| 160 |
+ " : " + ex); //$NON-NLS-1$ |
|
| 162 |
+ " : " + ex); //$NON-NLS-1$
|
|
| 161 | 163 |
} |
| 162 | 164 |
} |
| 163 |
|
|
| 165 |
|
|
| 164 | 166 |
NodeList partitionElems = e.getElementsByTagName("partition"); //$NON-NLS-1$
|
| 165 |
for (int i = 0 ; i < partitionElems.getLength() ; i++) {
|
|
| 167 |
for (int i = 0; i < partitionElems.getLength(); i++) {
|
|
| 166 | 168 |
Element partitionElem = (Element) partitionElems.item(i); |
| 167 | 169 |
String name = partitionElem.getAttribute("name"); //$NON-NLS-1$
|
| 168 |
List<String> names = new ArrayList<String>();
|
|
| 169 |
List<String> queries = new ArrayList<String>();
|
|
| 170 |
|
|
| 170 |
List<String> names = new ArrayList<>(); |
|
| 171 |
List<String> queries = new ArrayList<>(); |
|
| 172 |
|
|
| 171 | 173 |
NodeList partElems = partitionElem.getElementsByTagName("part"); //$NON-NLS-1$
|
| 172 | 174 |
for (int j = 0; j < partElems.getLength(); j++) {
|
| 173 | 175 |
Element part = (Element) partElems.item(j); |
| ... | ... | |
| 175 | 177 |
queries.add(part.getAttribute("query")); //$NON-NLS-1$
|
| 176 | 178 |
} |
| 177 | 179 |
try {
|
| 178 |
//System.out.println("Create Partition with corpus "+this+" : "+name+", queries "+queries+", names "+names);
|
|
| 180 |
// System.out.println("Create Partition with corpus "+this+" : "+name+", queries "+queries+", names "+names);
|
|
| 179 | 181 |
Partition partition = new Partition(this); |
| 180 | 182 |
partition.setParameters(name, queries, names); |
| 181 |
//partition.setSelfElement(partitionElem); |
|
| 183 |
// partition.setSelfElement(partitionElem);
|
|
| 182 | 184 |
partition._load(partitionElem); |
| 183 |
} catch (Exception ex) {
|
|
| 185 |
} |
|
| 186 |
catch (Exception ex) {
|
|
| 184 | 187 |
Log.warning(this.pID + TXMCoreMessages.failedToRestorePartition + name |
| 185 | 188 |
+ " : " + ex); //$NON-NLS-1$ |
| 186 |
//c.getParentNode().removeChild(c); |
|
| 189 |
// c.getParentNode().removeChild(c);
|
|
| 187 | 190 |
} |
| 188 | 191 |
} |
| 189 | 192 |
} |
| 190 | 193 |
initSelectionResult(e); |
| 191 |
//System.out.println("Subcorpus.load: Selection of "+this+" = "+selectionResult);
|
|
| 192 |
//return super.load(); |
|
| 194 |
// System.out.println("Subcorpus.load: Selection of "+this+" = "+selectionResult);
|
|
| 195 |
// return super.load();
|
|
| 193 | 196 |
return true; |
| 194 | 197 |
} |
| 195 |
|
|
| 198 |
|
|
| 196 | 199 |
@Override |
| 197 | 200 |
public boolean canCompute() {
|
| 198 | 201 |
return pID != null && userName != null && (pQuery != null || qresult != null); |
| ... | ... | |
| 200 | 203 |
|
| 201 | 204 |
/* |
| 202 | 205 |
* (non-Javadoc) |
| 203 |
* |
|
| 204 | 206 |
* @see |
| 205 | 207 |
* org.txm.searchengine.cqp.corpus.CqpObject#checkCqpId(java.lang |
| 206 | 208 |
* .String) |
| ... | ... | |
| 211 | 213 |
throw new InvalidCqpIdException(TXMCoreMessages.bind(CQPSearchEngineCoreMessages.p0IsNotAValidCQPIDForASubcorpusItMustBeAnUppercaseCharacterFollowedByLowercaseCharacters, pID)); |
| 212 | 214 |
return true; |
| 213 | 215 |
} |
| 214 |
|
|
| 216 |
|
|
| 215 | 217 |
@Override |
| 216 | 218 |
public void clean() {
|
| 217 | 219 |
super.clean(); |
| 218 | 220 |
|
| 219 | 221 |
// nothing to do if the corpus has not yet been computed |
| 220 |
if(this.hasBeenComputedOnce) {
|
|
| 222 |
if (this.hasBeenComputedOnce) {
|
|
| 221 | 223 |
try {
|
| 222 | 224 |
AbstractCqiClient CQI = CorpusManager.getCorpusManager().getCqiClient(); |
| 223 | 225 |
if (CQPSearchEngine.isInitialized()) {
|
| ... | ... | |
| 234 | 236 |
qresult = null; |
| 235 | 237 |
selectionResult = null; |
| 236 | 238 |
} |
| 237 |
|
|
| 239 |
|
|
| 238 | 240 |
@Override |
| 239 | 241 |
public String getDetails() {
|
| 240 | 242 |
try {
|
| 241 |
return "T "+this.getSize(); |
|
| 242 |
} catch (CqiClientException e) {
|
|
| 243 |
return this.getName()+ " (" + this.getQualifiedCqpId()+")";
|
|
| 243 |
return "T " + this.getSize(); |
|
| 244 | 244 |
} |
| 245 |
catch (CqiClientException e) {
|
|
| 246 |
return this.getName() + " (" + this.getQualifiedCqpId() + ")";
|
|
| 247 |
} |
|
| 245 | 248 |
} |
| 246 |
|
|
| 249 |
|
|
| 247 | 250 |
@Override |
| 248 | 251 |
public List<Match> getMatches() {
|
| 249 | 252 |
try {
|
| 250 | 253 |
return qresult.getMatches(); |
| 251 |
} catch (CqiClientException e) {
|
|
| 254 |
} |
|
| 255 |
catch (CqiClientException e) {
|
|
| 252 | 256 |
// TODO Auto-generated catch block |
| 253 | 257 |
org.txm.utils.logger.Log.printStackTrace(e); |
| 254 |
return new ArrayList<Match>();
|
|
| 258 |
return new ArrayList<>(); |
|
| 255 | 259 |
} |
| 256 | 260 |
} |
| 257 |
|
|
| 261 |
|
|
| 258 | 262 |
public int getNMatch() {
|
| 259 | 263 |
try {
|
| 260 | 264 |
return qresult.getNMatch(); |
| 261 |
} catch (CqiClientException e) {
|
|
| 265 |
} |
|
| 266 |
catch (CqiClientException e) {
|
|
| 262 | 267 |
// TODO Auto-generated catch block |
| 263 | 268 |
org.txm.utils.logger.Log.printStackTrace(e); |
| 264 | 269 |
return 0; |
| 265 | 270 |
} |
| 266 | 271 |
} |
| 267 |
|
|
| 272 |
|
|
| 268 | 273 |
/* |
| 269 | 274 |
* (non-Javadoc) |
| 270 |
* |
|
| 271 | 275 |
* @see org.txm.searchengine.cqp.corpus.Corpus#getProperties() |
| 272 | 276 |
*/ |
| 273 | 277 |
@Override |
| 274 | 278 |
public List<WordProperty> getProperties() throws CqiClientException {
|
| 275 | 279 |
return getMainCorpus().getProperties(); |
| 276 | 280 |
} |
| 277 |
|
|
| 281 |
|
|
| 278 | 282 |
/* |
| 279 | 283 |
* (non-Javadoc) |
| 280 |
* |
|
| 281 | 284 |
* @see |
| 282 | 285 |
* org.txm.searchengine.cqp.corpus.Corpus#getProperty(java.lang. |
| 283 | 286 |
* String) |
| ... | ... | |
| 286 | 289 |
public WordProperty getProperty(String name) throws CqiClientException {
|
| 287 | 290 |
return getMainCorpus().getProperty(name); |
| 288 | 291 |
} |
| 289 |
|
|
| 292 |
|
|
| 290 | 293 |
// protected void finalize() throws Throwable {
|
| 291 | 294 |
// try {
|
| 292 | 295 |
// // |
| 293 |
//
|
|
| 296 |
// |
|
| 294 | 297 |
// //As finalize is run in a separate thread, we have problems of concurrent |
| 295 | 298 |
// accesses to the cqi server if we drop the subcorpus here. |
| 296 | 299 |
// //It as to be done manually |
| ... | ... | |
| 301 | 304 |
// super.finalize(); |
| 302 | 305 |
// } |
| 303 | 306 |
// } |
| 304 |
|
|
| 307 |
|
|
| 305 | 308 |
/* |
| 306 | 309 |
* (non-Javadoc) |
| 307 |
* |
|
| 308 | 310 |
* @see |
| 309 | 311 |
* org.txm.searchengine.cqp.corpus.CqpObject#getQualifiedCqpId() |
| 310 | 312 |
*/ |
| ... | ... | |
| 312 | 314 |
public String getQualifiedCqpId() {
|
| 313 | 315 |
return getMainCorpus().getCqpId() + ":" + this.pID; //$NON-NLS-1$ |
| 314 | 316 |
} |
| 315 |
|
|
| 316 |
// /**
|
|
| 317 |
// * Register to parent.
|
|
| 318 |
// *
|
|
| 319 |
// * @param subcorpusName the subcorpus name
|
|
| 320 |
// * @param query the query
|
|
| 321 |
// */
|
|
| 322 |
// private void registerToParent(String subcorpusName, CQLQuery query) {
|
|
| 323 |
// if (this.getSelfElement() == null && this.getParent() != null)
|
|
| 324 |
// try {
|
|
| 325 |
// Element parentElem = (Element) this.getParent().getSelfElement();
|
|
| 326 |
// if (parentElem == null) return;
|
|
| 327 |
// Document doc = parentElem.getOwnerDocument();
|
|
| 328 |
// Element subcorpusElem = doc.createElement("subcorpus"); //$NON-NLS-1$
|
|
| 329 |
// subcorpusElem.setAttribute("name", subcorpusName); //$NON-NLS-1$
|
|
| 330 |
// subcorpusElem.setAttribute("desc", subcorpusName); //$NON-NLS-1$
|
|
| 331 |
// if (query != null)
|
|
| 332 |
// subcorpusElem.setAttribute("query", query.getQueryString()); //$NON-NLS-1$
|
|
| 333 |
// Element corporaElem;
|
|
| 334 |
// if (this.getParent() instanceof MainCorpus) {
|
|
| 335 |
// NodeList corporaList = parentElem.getElementsByTagName("preBuild"); //$NON-NLS-1$
|
|
| 336 |
// corporaElem = (Element) corporaList.item(0);
|
|
| 337 |
// } else {
|
|
| 338 |
// corporaElem = parentElem;
|
|
| 339 |
// }
|
|
| 340 |
// corporaElem.appendChild(subcorpusElem);
|
|
| 341 |
// this.setSelfElement(subcorpusElem);
|
|
| 342 |
// } catch (Exception e) {
|
|
| 343 |
// System.out.println(TXMCoreMessages.Subcorpus_2+Log.toString(e));
|
|
| 344 |
// }
|
|
| 317 |
|
|
| 318 |
// /**
|
|
| 319 |
// * Register to parent. |
|
| 320 |
// * |
|
| 321 |
// * @param subcorpusName the subcorpus name |
|
| 322 |
// * @param query the query |
|
| 323 |
// */ |
|
| 324 |
// private void registerToParent(String subcorpusName, CQLQuery query) {
|
|
| 325 |
// if (this.getSelfElement() == null && this.getParent() != null)
|
|
| 326 |
// try {
|
|
| 327 |
// Element parentElem = (Element) this.getParent().getSelfElement();
|
|
| 328 |
// if (parentElem == null) return;
|
|
| 329 |
// Document doc = parentElem.getOwnerDocument();
|
|
| 330 |
// Element subcorpusElem = doc.createElement("subcorpus"); //$NON-NLS-1$
|
|
| 331 |
// subcorpusElem.setAttribute("name", subcorpusName); //$NON-NLS-1$
|
|
| 332 |
// subcorpusElem.setAttribute("desc", subcorpusName); //$NON-NLS-1$
|
|
| 333 |
// if (query != null)
|
|
| 334 |
// subcorpusElem.setAttribute("query", query.getQueryString()); //$NON-NLS-1$
|
|
| 335 |
// Element corporaElem;
|
|
| 336 |
// if (this.getParent() instanceof MainCorpus) {
|
|
| 337 |
// NodeList corporaList = parentElem.getElementsByTagName("preBuild"); //$NON-NLS-1$
|
|
| 338 |
// corporaElem = (Element) corporaList.item(0);
|
|
| 339 |
// } else {
|
|
| 340 |
// corporaElem = parentElem;
|
|
| 341 |
// }
|
|
| 342 |
// corporaElem.appendChild(subcorpusElem);
|
|
| 343 |
// this.setSelfElement(subcorpusElem);
|
|
| 344 |
// } catch (Exception e) {
|
|
| 345 |
// System.out.println(TXMCoreMessages.Subcorpus_2+Log.toString(e));
|
|
| 346 |
// }
|
|
| 345 | 347 |
// |
| 346 |
// this._load();
|
|
| 347 |
// }
|
|
| 348 |
|
|
| 348 |
// this._load();
|
|
| 349 |
// }
|
|
| 350 |
|
|
| 349 | 351 |
/** |
| 350 | 352 |
* Gets the query. |
| 351 | 353 |
* |
| ... | ... | |
| 354 | 356 |
public CQLQuery getQuery() {
|
| 355 | 357 |
return pQuery; |
| 356 | 358 |
} |
| 357 |
|
|
| 359 |
|
|
| 358 | 360 |
@Override |
| 359 | 361 |
public String getResultType() {
|
| 360 | 362 |
return "Corpus"; |
| 361 | 363 |
} |
| 362 |
|
|
| 364 |
|
|
| 363 | 365 |
public SelectionResult getSelectionResult() {
|
| 364 | 366 |
return selectionResult; |
| 365 | 367 |
} |
| 366 |
|
|
| 368 |
|
|
| 367 | 369 |
@Override |
| 368 | 370 |
public String getSimpleName() {
|
| 369 | 371 |
return this.getName(); |
| 370 | 372 |
} |
| 371 |
|
|
| 373 |
|
|
| 372 | 374 |
/** |
| 373 | 375 |
* Returns the number of occurrences in the subcorpus. |
| 374 | 376 |
* |
| ... | ... | |
| 384 | 386 |
if (!hasBeenComputedOnce) return 0; |
| 385 | 387 |
|
| 386 | 388 |
if (this.size == -1) {
|
| 387 |
//Log.finest(TXMCoreMessages.bind(TXMCoreMessages.SUBCORPUS_SIZE, new Object[]{this.pName, "N/A"}));
|
|
| 389 |
// Log.finest(TXMCoreMessages.bind(TXMCoreMessages.SUBCORPUS_SIZE, new Object[]{this.pName, "N/A"}));
|
|
| 388 | 390 |
long start = System.currentTimeMillis(); |
| 389 |
|
|
| 391 |
|
|
| 390 | 392 |
try {
|
| 391 | 393 |
AbstractCqiClient cqiClient = CorpusManager.getCorpusManager().getCqiClient(); |
| 392 |
String qid= this.getQualifiedCqpId(); |
|
| 394 |
String qid = this.getQualifiedCqpId();
|
|
| 393 | 395 |
int nMatch = cqiClient.subCorpusSize(qid); |
| 394 |
//System.out.println("n matchs: "+nMatch);
|
|
| 396 |
// System.out.println("n matchs: "+nMatch);
|
|
| 395 | 397 |
if (nMatch == 0) {
|
| 396 | 398 |
this.size = 0; |
| 397 | 399 |
} |
| ... | ... | |
| 402 | 404 |
throw new UnexpectedAnswerException(); |
| 403 | 405 |
} |
| 404 | 406 |
this.size = 0; |
| 405 |
|
|
| 407 |
|
|
| 406 | 408 |
for (int i = 0; i < match.length; i++) {
|
| 407 | 409 |
size += matchend[i] - match[i] + 1; |
| 408 | 410 |
} |
| 409 | 411 |
} |
| 410 |
} catch (Exception e) {
|
|
| 412 |
} |
|
| 413 |
catch (Exception e) {
|
|
| 411 | 414 |
throw new CqiClientException(e); |
| 412 | 415 |
} |
| 413 | 416 |
long end = System.currentTimeMillis(); |
| 414 |
Log.finest(NLS.bind(SearchEngineCoreMessages.sizeOfSubcorpusP0P1ComputedInP2Ms, new Object[]{this.pID, this.size, (end - start)}));
|
|
| 417 |
Log.finest(NLS.bind(SearchEngineCoreMessages.sizeOfSubcorpusP0P1ComputedInP2Ms, new Object[] { this.pID, this.size, (end - start) }));
|
|
| 415 | 418 |
} |
| 416 | 419 |
return this.size; |
| 417 | 420 |
} |
| 418 |
|
|
| 421 |
|
|
| 419 | 422 |
@Override |
| 420 | 423 |
public int[] getStartLimits(String sup) throws IOException, |
| 421 |
CqiServerError, InvalidCqpIdException, CqiClientException {
|
|
| 424 |
CqiServerError, InvalidCqpIdException, CqiClientException {
|
|
| 422 | 425 |
return this.getMainCorpus().getStartLimits(sup); |
| 423 | 426 |
} |
| 424 |
|
|
| 427 |
|
|
| 425 | 428 |
/* |
| 426 | 429 |
* (non-Javadoc) |
| 427 |
* |
|
| 428 | 430 |
* @see |
| 429 | 431 |
* org.txm.searchengine.cqp.corpus.Corpus#getStructuralUnit(java |
| 430 | 432 |
* .lang.String) |
| ... | ... | |
| 434 | 436 |
throws CqiClientException {
|
| 435 | 437 |
return getMainCorpus().getStructuralUnit(name); |
| 436 | 438 |
} |
| 437 |
|
|
| 439 |
|
|
| 438 | 440 |
/* |
| 439 | 441 |
* (non-Javadoc) |
| 440 |
* |
|
| 441 | 442 |
* @see org.txm.searchengine.cqp.corpus.Corpus#getStructuralUnits() |
| 442 | 443 |
*/ |
| 443 | 444 |
@Override |
| 444 | 445 |
public List<StructuralUnit> getStructuralUnits() throws CqiClientException {
|
| 445 | 446 |
return getMainCorpus().getStructuralUnits(); |
| 446 | 447 |
} |
| 447 |
|
|
| 448 |
|
|
| 448 | 449 |
public SelectionResult initSelectionResult(Element e) {
|
| 449 |
|
|
| 450 |
|
|
| 450 | 451 |
if (e != null) {
|
| 451 | 452 |
selectionResult = new SelectionResult(); |
| 452 | 453 |
NodeList selectionList = e.getElementsByTagName("selection"); //$NON-NLS-1$
|
| 453 |
for (int i = 0 ; i < selectionList.getLength() ;) {
|
|
| 454 |
for (int i = 0; i < selectionList.getLength();) {
|
|
| 454 | 455 |
Element selection = (Element) selectionList.item(i); // the selection element |
| 455 |
|
|
| 456 |
|
|
| 456 | 457 |
NodeList textList = selection.getElementsByTagName("selText"); // get text ids //$NON-NLS-1$
|
| 457 |
for (int j = 0 ; j < textList.getLength() ; j++) {
|
|
| 458 |
for (int j = 0; j < textList.getLength(); j++) {
|
|
| 458 | 459 |
Element text = (Element) textList.item(j); |
| 459 | 460 |
selectionResult.add(text.getAttribute("id")); //$NON-NLS-1$
|
| 460 | 461 |
} |
| 461 |
|
|
| 462 |
|
|
| 462 | 463 |
NodeList critList = selection.getElementsByTagName("selCrit"); // get the criteria //$NON-NLS-1$
|
| 463 |
for (int j = 0 ; j < critList.getLength() ; j++) {
|
|
| 464 |
for (int j = 0; j < critList.getLength(); j++) {
|
|
| 464 | 465 |
Element crit = (Element) critList.item(j); |
| 465 | 466 |
String name = crit.getAttribute("id"); // the name of the critera //$NON-NLS-1$
|
| 466 | 467 |
if (!selectionResult.critera.containsKey(name)) // create new list if new critera |
| ... | ... | |
| 472 | 473 |
} |
| 473 | 474 |
return selectionResult; |
| 474 | 475 |
} |
| 475 |
|
|
| 476 |
|
|
| 476 | 477 |
@Override |
| 477 | 478 |
public boolean loadParameters() throws Exception {
|
| 478 | 479 |
String q = this.getStringParameterValue(TXMPreferences.QUERY); |
| ... | ... | |
| 482 | 483 |
|
| 483 | 484 |
return true; |
| 484 | 485 |
} |
| 485 |
|
|
| 486 |
|
|
| 486 | 487 |
@Override |
| 487 | 488 |
public boolean saveParameters() throws Exception {
|
| 488 | 489 |
if (pQuery != null) {
|
| ... | ... | |
| 490 | 491 |
} |
| 491 | 492 |
return true; |
| 492 | 493 |
} |
| 493 |
|
|
| 494 |
|
|
| 494 | 495 |
/** |
| 495 | 496 |
* Instantiates a new subcorpus. |
| 496 | 497 |
* |
| ... | ... | |
| 501 | 502 |
* @param parent the parent |
| 502 | 503 |
* @param query the query |
| 503 | 504 |
* @throws InvalidCqpIdException the invalid cqp id exception |
| 504 |
* {@link CQPCorpus#createSubcorpus(CQLQuery, String)}
|
|
| 505 |
* {@link CQPCorpus#createSubcorpus(CQLQuery, String)}
|
|
| 505 | 506 |
*/ |
| 506 | 507 |
protected void setParameters(String pID, String name, CQLQuery query) throws InvalidCqpIdException {
|
| 507 | 508 |
this.pID = pID; |
| 508 | 509 |
this.pQuery = query; |
| 509 | 510 |
this.userName = name; |
| 510 | 511 |
} |
| 511 |
|
|
| 512 |
|
|
| 512 | 513 |
/** |
| 513 | 514 |
* |
| 514 | 515 |
* @param cqpId |
| ... | ... | |
| 520 | 521 |
this.userName = name; |
| 521 | 522 |
this.pQuery = queryResult.getQuery(); |
| 522 | 523 |
this.qresult = queryResult; |
| 523 |
|
|
| 524 |
|
|
| 524 | 525 |
this.dirty = false; |
| 525 | 526 |
} |
| 526 |
|
|
| 527 |
|
|
| 527 | 528 |
@Override |
| 528 | 529 |
public boolean setParameters(TXMParameters parameters) {
|
| 529 | 530 |
// TODO Auto-generated method stub |
| 530 | 531 |
System.err.println("Subcorpus.setParameters(): not yet implemented.");
|
| 531 | 532 |
return true; |
| 532 | 533 |
} |
| 533 |
|
|
| 534 |
|
|
| 534 | 535 |
/** |
| 535 | 536 |
* Sets the query to use. |
| 537 |
* |
|
| 536 | 538 |
* @param query |
| 537 | 539 |
*/ |
| 538 | 540 |
public void setQuery(CQLQuery query) {
|
| 539 | 541 |
this.pQuery = query; |
| 540 | 542 |
} |
| 541 |
|
|
| 542 |
|
|
| 543 |
//TODO merge SelectionResult and Subcorpus ? |
|
| 543 |
|
|
| 544 |
|
|
| 545 |
// TODO merge SelectionResult and Subcorpus ?
|
|
| 544 | 546 |
public void setSelectionResult(SelectionResult selectionResult) {
|
| 545 | 547 |
this.selectionResult = selectionResult; |
| 546 |
//TODO persists the selectionResult ? |
|
| 547 |
// if (this.getSelfElement() != null) {
|
|
| 548 |
// //remove old node if any
|
|
| 549 |
// NodeList selectionList = getSelfElement().getElementsByTagName("selection"); //$NON-NLS-1$
|
|
| 550 |
// for (int i = 0 ; i < selectionList.getLength() ; i++)
|
|
| 551 |
// getSelfElement().removeChild(selectionList.item(i));
|
|
| 548 |
// TODO persists the selectionResult ?
|
|
| 549 |
// if (this.getSelfElement() != null) {
|
|
| 550 |
// //remove old node if any
|
|
| 551 |
// NodeList selectionList = getSelfElement().getElementsByTagName("selection"); //$NON-NLS-1$
|
|
| 552 |
// for (int i = 0 ; i < selectionList.getLength() ; i++)
|
|
| 553 |
// getSelfElement().removeChild(selectionList.item(i));
|
|
| 552 | 554 |
// |
| 553 |
// Document doc = this.getParent().getSelfElement().getOwnerDocument();
|
|
| 554 |
// Element selectionElem = doc.createElement("selection"); // the selection element //$NON-NLS-1$
|
|
| 555 |
// for (String txt : selectionResult) // the text ids
|
|
| 556 |
// {
|
|
| 557 |
// Element textElem = doc.createElement("selText"); //$NON-NLS-1$
|
|
| 558 |
// textElem.setAttribute("id", txt); //$NON-NLS-1$
|
|
| 559 |
// selectionElem.appendChild(textElem);
|
|
| 560 |
// }
|
|
| 561 |
// for (String crit : selectionResult.critera.keySet()) // the critera
|
|
| 562 |
// {
|
|
| 563 |
// ArrayList<String> values = selectionResult.critera.get(crit);
|
|
| 564 |
// for (String value : values) {
|
|
| 565 |
// Element criteraElem = doc.createElement("selCrit"); //$NON-NLS-1$
|
|
| 566 |
// criteraElem.setAttribute("id", crit); //$NON-NLS-1$
|
|
| 567 |
// criteraElem.setAttribute("value", value); //$NON-NLS-1$
|
|
| 568 |
// selectionElem.appendChild(criteraElem);
|
|
| 569 |
// }
|
|
| 570 |
// }
|
|
| 571 |
// getSelfElement().appendChild(selectionElem);// append selection elem to self
|
|
| 572 |
// }
|
|
| 555 |
// Document doc = this.getParent().getSelfElement().getOwnerDocument();
|
|
| 556 |
// Element selectionElem = doc.createElement("selection"); // the selection element //$NON-NLS-1$
|
|
| 557 |
// for (String txt : selectionResult) // the text ids
|
|
| 558 |
// {
|
|
| 559 |
// Element textElem = doc.createElement("selText"); //$NON-NLS-1$
|
|
| 560 |
// textElem.setAttribute("id", txt); //$NON-NLS-1$
|
|
| 561 |
// selectionElem.appendChild(textElem);
|
|
| 562 |
// }
|
|
| 563 |
// for (String crit : selectionResult.critera.keySet()) // the critera
|
|
| 564 |
// {
|
|
| 565 |
// ArrayList<String> values = selectionResult.critera.get(crit);
|
|
| 566 |
// for (String value : values) {
|
|
| 567 |
// Element criteraElem = doc.createElement("selCrit"); //$NON-NLS-1$
|
|
| 568 |
// criteraElem.setAttribute("id", crit); //$NON-NLS-1$
|
|
| 569 |
// criteraElem.setAttribute("value", value); //$NON-NLS-1$
|
|
| 570 |
// selectionElem.appendChild(criteraElem);
|
|
| 571 |
// }
|
|
| 572 |
// }
|
|
| 573 |
// getSelfElement().appendChild(selectionElem);// append selection elem to self
|
|
| 574 |
// }
|
|
| 573 | 575 |
} |
| 574 |
|
|
| 575 |
/* (non-Javadoc) |
|
| 576 |
|
|
| 577 |
/* |
|
| 578 |
* (non-Javadoc) |
|
| 576 | 579 |
* @see java.lang.Object#toString() |
| 577 | 580 |
*/ |
| 578 | 581 |
@Override |
| 579 | 582 |
public String toString() {
|
| 580 | 583 |
return this.getName(); |
| 581 | 584 |
} |
| 582 |
|
|
| 585 |
|
|
| 583 | 586 |
@Override |
| 584 | 587 |
@Deprecated |
| 585 | 588 |
public boolean toTxt(File outfile, String encoding, String colseparator, String txtseparator) throws Exception {
|
| 586 | 589 |
return false; |
| 587 | 590 |
} |
| 588 |
|
|
| 591 |
|
|
| 589 | 592 |
@Override |
| 590 | 593 |
public void setIsModified(boolean b) {
|
| 591 | 594 |
// nothing |
| 592 | 595 |
} |
| 593 |
|
|
| 596 |
|
|
| 594 | 597 |
@Override |
| 595 | 598 |
public CorpusBuild getRootCorpusBuild() {
|
| 596 | 599 |
return getMainCorpus(); |
| ... | ... | |
| 599 | 602 |
/** |
| 600 | 603 |
* Return the CQP START positions of the main corpus texts |
| 601 | 604 |
*/ |
| 605 |
@Override |
|
| 602 | 606 |
public int[] getTextStartLimits() throws CqiClientException, IOException, CqiServerError, InvalidCqpIdException {
|
| 603 | 607 |
if (textLimits == null) {
|
| 604 | 608 |
textLimits = getStartLimits("[text] expand left to text");
|
| ... | ... | |
| 609 | 613 |
/** |
| 610 | 614 |
* Return the CQP END positions of the main corpus texts |
| 611 | 615 |
*/ |
| 616 |
@Override |
|
| 612 | 617 |
public int[] getTextEndLimits() throws CqiClientException, IOException, CqiServerError, InvalidCqpIdException {
|
| 613 | 618 |
if (textEndLimits == null) {
|
| 614 | 619 |
String queryResultId = queryResultNamePrefix + UUID.randomUUID().toString(); |
| 615 | 620 |
CorpusManager.getCorpusManager().getCqiClient().cqpQuery( |
| 616 | 621 |
this.getQualifiedCqpId(), queryResultId, "[text] expand right to text"); //$NON-NLS-1$ |
| 617 |
QueryResult queryResult = new QueryResult(queryResultId, queryResultId, this, new CQLQuery("[text] expand right to text")); //$NON-NLS-1$
|
|
| 622 |
QueryResult queryResult = new QueryResult(queryResultId, queryResultId, this, new CQLQuery("[text] expand right to text")); //$NON-NLS-1$
|
|
| 618 | 623 |
textEndLimits = queryResult.getEnds(); |
| 619 | 624 |
} |
| 620 | 625 |
return textEndLimits; |
| ... | ... | |
| 625 | 630 |
* |
| 626 | 631 |
* @return the nb texts |
| 627 | 632 |
* @throws CqiClientException the cqi client exception |
| 628 |
* @throws CqiServerError
|
|
| 629 |
* @throws IOException
|
|
| 633 |
* @throws CqiServerError |
|
| 634 |
* @throws IOException |
|
| 630 | 635 |
*/ |
| 631 |
public int getNbTexts() throws CqiClientException, IOException, CqiServerError
|
|
| 632 |
{
|
|
| 636 |
@Override
|
|
| 637 |
public int getNbTexts() throws CqiClientException, IOException, CqiServerError {
|
|
| 633 | 638 |
if (nbtext == -1) {
|
| 634 | 639 |
|
| 635 | 640 |
List<Integer> structsIncorpus = getTextNumberInCorpus(); |
| ... | ... | |
| 638 | 643 |
} |
| 639 | 644 |
return nbtext; |
| 640 | 645 |
} |
| 641 |
|
|
| 646 |
|
|
| 642 | 647 |
private List<Integer> getTextNumberInCorpus() throws CqiClientException, IOException, CqiServerError {
|
| 643 | 648 |
|
| 644 | 649 |
StructuralUnit text_su = this.getStructuralUnit("text"); //$NON-NLS-1$
|
| ... | ... | |
| 647 | 652 |
|
| 648 | 653 |
int[] structs = new int[nbtext]; |
| 649 | 654 |
int[][] structs_positions = new int[structs.length][2]; |
| 650 |
for (int i = 0 ; i < nbtext ; i++) {
|
|
| 655 |
for (int i = 0; i < nbtext; i++) {
|
|
| 651 | 656 |
structs[i] = i; |
| 652 | 657 |
structs_positions[i] = CorpusManager.getCorpusManager().getCqiClient().struc2Cpos(text_id_sup.getQualifiedName(), i); |
| 653 | 658 |
} |
| 654 | 659 |
|
| 655 |
List<Integer> structsIncorpus = new ArrayList<Integer>();
|
|
| 656 |
//filter structs with matches |
|
| 660 |
List<Integer> structsIncorpus = new ArrayList<>(); |
|
| 661 |
// filter structs with matches
|
|
| 657 | 662 |
List<Match> matches = this.getMatches(); |
| 658 | 663 |
int iText = 0; |
| 659 | 664 |
int iMatch = 0; |
| ... | ... | |
| 661 | 666 |
while (iText < structs.length && iMatch < matches.size()) {
|
| 662 | 667 |
if (structs_positions[iText][1] < matches.get(iMatch).getStart()) { // match before text
|
| 663 | 668 |
iText++; |
| 664 |
} else if (matches.get(iMatch).getEnd() < structs_positions[iText][0]) { // text before match
|
|
| 669 |
} |
|
| 670 |
else if (matches.get(iMatch).getEnd() < structs_positions[iText][0]) { // text before match
|
|
| 665 | 671 |
iMatch++; |
| 666 |
} else { // a match in the text
|
|
| 672 |
} |
|
| 673 |
else { // a match in the text
|
|
| 667 | 674 |
structsIncorpus.add(iText); |
| 668 | 675 |
iText++; |
| 669 | 676 |
} |
| ... | ... | |
| 678 | 685 |
* @throws IOException |
| 679 | 686 |
* @throws CqiServerError |
| 680 | 687 |
*/ |
| 688 |
@Override |
|
| 681 | 689 |
public String[] getCorpusTextIdsList() throws CqiClientException, IOException, CqiServerError {
|
| 682 |
|
|
| 690 |
|
|
| 683 | 691 |
StructuralUnit text_su = this.getStructuralUnit("text"); //$NON-NLS-1$
|
| 684 | 692 |
StructuralUnitProperty text_id_sup = text_su.getProperty("id"); //$NON-NLS-1$
|
| 685 | 693 |
|
| 686 | 694 |
List<Integer> list = getTextNumberInCorpus(); |
| 687 | 695 |
int[] structs = new int[list.size()]; |
| 688 |
for (int i = 0 ; i < list.size() ; i++) {
|
|
| 696 |
for (int i = 0; i < list.size(); i++) {
|
|
| 689 | 697 |
structs[i] = list.get(i); |
| 690 | 698 |
} |
| 691 |
|
|
| 699 |
|
|
| 692 | 700 |
return CorpusManager.getCorpusManager().getCqiClient().struc2Str(text_id_sup.getQualifiedName(), structs); |
| 693 | 701 |
} |
| 694 | 702 |
|
| ... | ... | |
| 700 | 708 |
* @throws CqiServerError the cqi server error |
| 701 | 709 |
* @throws CqiClientException the cqi client exception |
| 702 | 710 |
*/ |
| 703 |
public HashMap<String, Integer> getTextsIds() throws IOException, CqiServerError, CqiClientException
|
|
| 704 |
{
|
|
| 711 |
@Override
|
|
| 712 |
public HashMap<String, Integer> getTextsIds() throws IOException, CqiServerError, CqiClientException {
|
|
| 705 | 713 |
if (textids == null) {
|
| 706 |
textids = new HashMap<String, Integer>();
|
|
| 714 |
textids = new HashMap<>(); |
|
| 707 | 715 |
List<Integer> list = getTextNumberInCorpus(); |
| 708 | 716 |
int[] structs = new int[list.size()]; |
| 709 |
for (int i = 0 ; i < list.size() ; i++) {
|
|
| 717 |
for (int i = 0; i < list.size(); i++) {
|
|
| 710 | 718 |
structs[i] = list.get(i); |
| 711 | 719 |
} |
| 712 |
|
|
| 720 |
|
|
| 713 | 721 |
StructuralUnit text_su = this.getStructuralUnit("text"); //$NON-NLS-1$
|
| 714 | 722 |
StructuralUnitProperty text_id_sup = text_su.getProperty("id"); //$NON-NLS-1$
|
| 715 |
|
|
| 723 |
|
|
| 716 | 724 |
String[] ids = CorpusManager.getCorpusManager().getCqiClient().struc2Str(text_id_sup.getQualifiedName(), structs); |
| 717 |
for(int i = 0 ; i < ids.length ; i++) {
|
|
| 725 |
for (int i = 0; i < ids.length; i++) {
|
|
| 718 | 726 |
textids.put(ids[i], structs[i]); |
| 719 | 727 |
} |
| 720 | 728 |
} |
| 721 | 729 |
return textids; |
| 722 | 730 |
} |
| 723 |
} |
|
| 731 |
} |
|
Formats disponibles : Unified diff