Révision 2488
| tmp/org.txm.progression.core/src/org/txm/progression/core/functions/Progression.java (revision 2488) | ||
|---|---|---|
| 64 | 64 |
|
| 65 | 65 |
/** |
| 66 | 66 |
* Progression similar to the progression of Weblex software. |
| 67 |
*
|
|
| 67 |
* |
|
| 68 | 68 |
* @author mdecorde |
| 69 | 69 |
* @author sjacquot |
| 70 | 70 |
*/ |
| 71 |
public class Progression extends ChartResult {
|
|
| 72 |
|
|
| 73 |
|
|
| 71 |
public class Progression extends ChartResult {
|
|
| 72 |
|
|
| 73 |
|
|
| 74 | 74 |
/** The norep. */ |
| 75 | 75 |
public static int norep = 1; |
| 76 |
|
|
| 76 |
|
|
| 77 | 77 |
/** The prefix r. */ |
| 78 | 78 |
public static String prefixR = "Progression"; //$NON-NLS-1$ |
| 79 |
|
|
| 79 |
|
|
| 80 | 80 |
public static final String DENSITY_CHART_TYPE = "density"; //$NON-NLS-1$ |
| 81 |
|
|
| 81 |
|
|
| 82 | 82 |
/** The Xmin corpus. */ |
| 83 | 83 |
protected int XminCorpus; |
| 84 |
|
|
| 84 |
|
|
| 85 | 85 |
/** The Xmax corpus. */ |
| 86 | 86 |
protected int XmaxCorpus; |
| 87 |
|
|
| 87 |
|
|
| 88 | 88 |
/** The writer. */ |
| 89 | 89 |
private BufferedWriter writer; |
| 90 |
|
|
| 90 |
|
|
| 91 | 91 |
/** List of token position in the text. */ |
| 92 | 92 |
protected List<int[]> allPositions; |
| 93 |
|
|
| 93 |
|
|
| 94 | 94 |
/** The structurepositions. */ |
| 95 | 95 |
protected int[] structurePositions; |
| 96 |
|
|
| 96 |
|
|
| 97 | 97 |
/** The structurenames. */ |
| 98 | 98 |
protected String[] structureNames; |
| 99 |
|
|
| 99 |
|
|
| 100 | 100 |
/** graph boundaries. */ |
| 101 | 101 |
protected int maxY; |
| 102 |
|
|
| 102 |
|
|
| 103 | 103 |
/** The max x. */ |
| 104 | 104 |
protected int maxX; |
| 105 |
|
|
| 106 |
|
|
| 105 |
|
|
| 106 |
|
|
| 107 | 107 |
/** The bande. */ |
| 108 | 108 |
protected float bande = 0.0f; |
| 109 |
|
|
| 109 |
|
|
| 110 | 110 |
/** |
| 111 | 111 |
* set this to skip the QueryResult build step |
| 112 | 112 |
*/ |
| 113 | 113 |
protected List<QueryResult> queryResults; |
| 114 |
|
|
| 115 |
|
|
| 114 |
|
|
| 115 |
|
|
| 116 | 116 |
/** |
| 117 | 117 |
* Queries. |
| 118 | 118 |
*/ |
| 119 |
@Parameter(key=TXMPreferences.QUERIES)
|
|
| 119 |
@Parameter(key = TXMPreferences.QUERIES)
|
|
| 120 | 120 |
protected List<CQLQuery> queries; |
| 121 |
|
|
| 121 |
|
|
| 122 | 122 |
/** |
| 123 | 123 |
* Structural unit. |
| 124 | 124 |
*/ |
| 125 |
@Parameter(key=TXMPreferences.STRUCTURAL_UNIT, type=Parameter.RENDERING)
|
|
| 125 |
@Parameter(key = TXMPreferences.STRUCTURAL_UNIT, type = Parameter.RENDERING)
|
|
| 126 | 126 |
protected StructuralUnit structuralUnit; |
| 127 |
|
|
| 127 |
|
|
| 128 | 128 |
/** |
| 129 | 129 |
* Structural unit property. |
| 130 | 130 |
*/ |
| 131 |
@Parameter(key=TXMPreferences.STRUCTURAL_UNIT_PROPERTY, type=Parameter.RENDERING)
|
|
| 131 |
@Parameter(key = TXMPreferences.STRUCTURAL_UNIT_PROPERTY, type = Parameter.RENDERING)
|
|
| 132 | 132 |
protected StructuralUnitProperty structuralUnitProperty; |
| 133 |
|
|
| 133 |
|
|
| 134 | 134 |
/** |
| 135 | 135 |
* Repeats or not the same value when displaying the section markers. |
| 136 | 136 |
*/ |
| 137 |
@Parameter(key=ProgressionPreferences.REPEAT_SAME_VALUES, type=Parameter.RENDERING)
|
|
| 137 |
@Parameter(key = ProgressionPreferences.REPEAT_SAME_VALUES, type = Parameter.RENDERING)
|
|
| 138 | 138 |
protected boolean repeatValues; |
| 139 |
|
|
| 139 |
|
|
| 140 | 140 |
/** |
| 141 | 141 |
* Only used in Density mode. |
| 142 | 142 |
*/ |
| 143 |
@Parameter(key=ProgressionPreferences.BANDE_MULTIPLIER)
|
|
| 143 |
@Parameter(key = ProgressionPreferences.BANDE_MULTIPLIER)
|
|
| 144 | 144 |
protected float bandeMultiplier; |
| 145 |
|
|
| 145 |
|
|
| 146 | 146 |
/** |
| 147 | 147 |
* Regex to display only the property that match it. |
| 148 | 148 |
*/ |
| 149 |
@Parameter(key=ProgressionPreferences.PROPERTY_REGEX, type=Parameter.RENDERING)
|
|
| 149 |
@Parameter(key = ProgressionPreferences.PROPERTY_REGEX, type = Parameter.RENDERING)
|
|
| 150 | 150 |
protected String propertyRegex; |
| 151 |
|
|
| 152 |
|
|
| 151 |
|
|
| 152 |
|
|
| 153 | 153 |
/** |
| 154 | 154 |
* Creates a not computed progression |
| 155 |
* |
|
| 155 | 156 |
* @param parent |
| 156 | 157 |
*/ |
| 157 |
public Progression(CQPCorpus parent) {
|
|
| 158 |
public Progression(CQPCorpus parent) {
|
|
| 158 | 159 |
this(null, parent); |
| 159 | 160 |
} |
| 160 |
|
|
| 161 |
|
|
| 161 | 162 |
/** |
| 162 | 163 |
* Creates a not computed progression |
| 164 |
* |
|
| 163 | 165 |
* @param parametersNodePath |
| 164 | 166 |
*/ |
| 165 |
public Progression(String parametersNodePath) {
|
|
| 167 |
public Progression(String parametersNodePath) {
|
|
| 166 | 168 |
super(parametersNodePath, null); |
| 167 | 169 |
} |
| 168 |
|
|
| 170 |
|
|
| 169 | 171 |
/** |
| 170 | 172 |
* Creates a not computed progression. |
| 173 |
* |
|
| 171 | 174 |
* @param parametersNodePath |
| 172 | 175 |
* @param parent |
| 173 | 176 |
*/ |
| ... | ... | |
| 176 | 179 |
|
| 177 | 180 |
this.domainGridLinesVisible = false; |
| 178 | 181 |
} |
| 179 |
|
|
| 180 | 182 |
|
| 181 |
|
|
| 183 |
|
|
| 184 |
|
|
| 182 | 185 |
@Override |
| 183 | 186 |
public boolean loadParameters() {
|
| 184 |
if(!this.getStringParameterValue(TXMPreferences.QUERIES).isEmpty()) {
|
|
| 187 |
if (!this.getStringParameterValue(TXMPreferences.QUERIES).isEmpty()) {
|
|
| 185 | 188 |
this.queries = CQLQuery.stringToQueries(this.getStringParameterValue(TXMPreferences.QUERIES)); |
| 186 | 189 |
} |
| 187 |
|
|
| 188 |
if(!this.getStringParameterValue(TXMPreferences.STRUCTURAL_UNIT).isEmpty()) {
|
|
| 190 |
|
|
| 191 |
if (!this.getStringParameterValue(TXMPreferences.STRUCTURAL_UNIT).isEmpty()) {
|
|
| 189 | 192 |
try {
|
| 190 | 193 |
this.structuralUnit = this.getCorpus().getStructuralUnit(this.getStringParameterValue(TXMPreferences.STRUCTURAL_UNIT)); |
| 191 |
if(!this.getStringParameterValue(TXMPreferences.STRUCTURAL_UNIT_PROPERTY).isEmpty()) {
|
|
| 194 |
if (!this.getStringParameterValue(TXMPreferences.STRUCTURAL_UNIT_PROPERTY).isEmpty()) {
|
|
| 192 | 195 |
this.structuralUnitProperty = this.structuralUnit.getProperty(this.getStringParameterValue(TXMPreferences.STRUCTURAL_UNIT_PROPERTY)); |
| 193 | 196 |
} |
| 194 | 197 |
} |
| ... | ... | |
| 197 | 200 |
e.printStackTrace(); |
| 198 | 201 |
} |
| 199 | 202 |
} |
| 200 |
|
|
| 201 |
|
|
| 203 |
|
|
| 204 |
|
|
| 202 | 205 |
// generate the chart type according to cumulative boolean preference |
| 203 |
if(!this.getBooleanParameterValue(ProgressionPreferences.CHART_CUMULATIVE)) {
|
|
| 206 |
if (!this.getBooleanParameterValue(ProgressionPreferences.CHART_CUMULATIVE)) {
|
|
| 204 | 207 |
this.chartType = Progression.DENSITY_CHART_TYPE; |
| 205 | 208 |
} |
| 206 |
|
|
| 209 |
|
|
| 207 | 210 |
return true; |
| 208 | 211 |
} |
| 209 |
|
|
| 210 |
|
|
| 212 |
|
|
| 213 |
|
|
| 211 | 214 |
@Override |
| 212 | 215 |
public boolean saveParameters() {
|
| 213 |
|
|
| 214 |
if(this.queries != null && !this.queries.isEmpty()) {
|
|
| 216 |
|
|
| 217 |
if (this.queries != null && !this.queries.isEmpty()) {
|
|
| 215 | 218 |
this.saveParameter(TXMPreferences.QUERIES, CQLQuery.queriesToString(this.queries)); |
| 216 | 219 |
} |
| 217 |
|
|
| 218 |
if(this.structuralUnit != null) {
|
|
| 220 |
|
|
| 221 |
if (this.structuralUnit != null) {
|
|
| 219 | 222 |
this.saveParameter(TXMPreferences.STRUCTURAL_UNIT, this.structuralUnit.getName()); |
| 220 | 223 |
} |
| 221 |
|
|
| 222 |
if(this.structuralUnitProperty != null) {
|
|
| 224 |
|
|
| 225 |
if (this.structuralUnitProperty != null) {
|
|
| 223 | 226 |
this.saveParameter(TXMPreferences.STRUCTURAL_UNIT_PROPERTY, this.structuralUnitProperty.getName()); |
| 224 | 227 |
} |
| 225 |
|
|
| 228 |
|
|
| 226 | 229 |
return true; |
| 227 | 230 |
} |
| 228 |
|
|
| 231 |
|
|
| 229 | 232 |
@Override |
| 230 | 233 |
public boolean canCompute() {
|
| 231 |
|
|
| 234 |
|
|
| 232 | 235 |
// null or empty queries |
| 233 |
if (this.queries == null || this.queries.isEmpty()) {
|
|
| 236 |
if (this.queries == null || this.queries.isEmpty()) {
|
|
| 234 | 237 |
Log.fine("Progression.canCompute(): can not compute with no query."); //$NON-NLS-1$
|
| 235 | 238 |
return false; |
| 236 | 239 |
} |
| 237 |
|
|
| 240 |
|
|
| 238 | 241 |
// wrong REGEX |
| 239 |
if (!this.propertyRegex.isEmpty()) {
|
|
| 242 |
if (!this.propertyRegex.isEmpty()) {
|
|
| 240 | 243 |
try {
|
| 241 | 244 |
Pattern.compile(this.propertyRegex); |
| 242 | 245 |
} |
| ... | ... | |
| 246 | 249 |
return false; |
| 247 | 250 |
} |
| 248 | 251 |
} |
| 249 |
|
|
| 252 |
|
|
| 250 | 253 |
return true; |
| 251 | 254 |
} |
| 252 |
|
|
| 255 |
|
|
| 253 | 256 |
@Override |
| 254 | 257 |
public boolean setParameters(TXMParameters parameters) {
|
| 255 | 258 |
// TODO Auto-generated method stub |
| 256 | 259 |
System.err.println("Progression.setParameters(): not yet implemented."); //$NON-NLS-1$
|
| 257 | 260 |
return true; |
| 258 | 261 |
} |
| 259 |
|
|
| 260 |
|
|
| 261 |
|
|
| 262 |
|
|
| 263 |
|
|
| 264 |
|
|
| 262 | 265 |
/** |
| 263 | 266 |
* Sets the structural unit. |
| 267 |
* |
|
| 264 | 268 |
* @param structuralUnit |
| 265 | 269 |
*/ |
| 266 |
public void setStructuralUnit(StructuralUnit structuralUnit) {
|
|
| 270 |
public void setStructuralUnit(StructuralUnit structuralUnit) {
|
|
| 267 | 271 |
this.structuralUnit = structuralUnit; |
| 268 | 272 |
this.setStructuralUnitProperty(null); |
| 269 | 273 |
} |
| 270 |
|
|
| 274 |
|
|
| 271 | 275 |
/** |
| 272 | 276 |
* Sets the structural unit property. |
| 277 |
* |
|
| 273 | 278 |
* @param structuralUnitProperty |
| 274 | 279 |
*/ |
| 275 |
public void setStructuralUnitProperty(StructuralUnitProperty structuralUnitProperty) {
|
|
| 280 |
public void setStructuralUnitProperty(StructuralUnitProperty structuralUnitProperty) {
|
|
| 276 | 281 |
this.structuralUnitProperty = structuralUnitProperty; |
| 277 | 282 |
} |
| 278 | 283 |
|
| 279 | 284 |
public void setQueryResults(List<QueryResult> queryResults) {
|
| 280 | 285 |
this.queryResults = queryResults; |
| 281 | 286 |
} |
| 282 |
|
|
| 287 |
|
|
| 283 | 288 |
/** |
| 284 | 289 |
* |
| 285 | 290 |
* @param queries |
| ... | ... | |
| 292 | 297 |
* @param bandeMultiplier |
| 293 | 298 |
*/ |
| 294 | 299 |
// FIXME : SJ: all parameters related to rendering should be removed from this class and centralized in the charts engine system, eg. line width |
| 295 |
public void setParameters(List<CQLQuery> queries,
|
|
| 296 |
StructuralUnit structuralUnit, StructuralUnitProperty property, String propertyregex,
|
|
| 297 |
boolean doCumulative, int lineWidth, boolean repeatValues, float bandeMultiplier) {
|
|
| 300 |
public void setParameters(List<CQLQuery> queries, |
|
| 301 |
StructuralUnit structuralUnit, StructuralUnitProperty property, String propertyregex, |
|
| 302 |
boolean doCumulative, int lineWidth, boolean repeatValues, float bandeMultiplier) {
|
|
| 298 | 303 |
this.queries = queries; // the query of the selection |
| 299 | 304 |
this.structuralUnit = structuralUnit; |
| 300 | 305 |
this.structuralUnitProperty = property; |
| 301 | 306 |
this.propertyRegex = propertyregex; |
| 302 | 307 |
this.repeatValues = repeatValues; |
| 303 | 308 |
this.bandeMultiplier = bandeMultiplier; |
| 304 |
|
|
| 305 |
for (int i = 0 ; i < DefaultTheme.monoWidths.length ; i++) {
|
|
| 309 |
|
|
| 310 |
for (int i = 0; i < DefaultTheme.monoWidths.length; i++) {
|
|
| 306 | 311 |
DefaultTheme.monoWidths[i] = lineWidth; |
| 307 | 312 |
} |
| 308 | 313 |
} |
| 309 |
|
|
| 314 |
|
|
| 310 | 315 |
/** |
| 311 | 316 |
* Computes. |
| 312 | 317 |
* |
| 313 | 318 |
* @return true, if successful |
| 314 | 319 |
* @throws CqiClientException the cqi client exception |
| 315 |
* @throws CqiServerError
|
|
| 316 |
* @throws IOException
|
|
| 320 |
* @throws CqiServerError |
|
| 321 |
* @throws IOException |
|
| 317 | 322 |
*/ |
| 323 |
@Override |
|
| 318 | 324 |
protected boolean __compute() throws Exception {
|
| 319 |
|
|
| 325 |
|
|
| 320 | 326 |
// Queries |
| 321 |
if(this.hasParameterChanged(ProgressionPreferences.QUERIES)) {
|
|
| 322 |
this.allPositions = new ArrayList<int[]>();
|
|
| 327 |
if (this.hasParameterChanged(ProgressionPreferences.QUERIES)) {
|
|
| 328 |
this.allPositions = new ArrayList<>(); |
|
| 323 | 329 |
this.structurePositions = new int[0]; |
| 324 | 330 |
this.structureNames = new String[0]; |
| 325 |
|
|
| 331 |
|
|
| 326 | 332 |
this.subTask("Processing queries...");
|
| 327 | 333 |
if (!this.stepQueries()) {
|
| 328 | 334 |
return true; |
| 329 | 335 |
} |
| 330 |
|
|
| 331 |
if (this.isCanceled()) {
|
|
| 336 |
|
|
| 337 |
if (this.isCanceled()) {
|
|
| 332 | 338 |
return false; |
| 333 | 339 |
} |
| 334 | 340 |
this.worked(10); |
| 335 | 341 |
} |
| 336 |
|
|
| 342 |
|
|
| 337 | 343 |
// Structural units |
| 338 | 344 |
this.subTask("Processing structural units...");
|
| 339 |
if (!this.stepStructuralUnits()) {
|
|
| 345 |
if (!this.stepStructuralUnits()) {
|
|
| 340 | 346 |
return false; |
| 341 | 347 |
} |
| 342 | 348 |
this.worked(20); |
| 343 |
|
|
| 344 |
|
|
| 349 |
|
|
| 350 |
|
|
| 345 | 351 |
// Finalization steps |
| 346 |
if(this.hasParameterChanged(ProgressionPreferences.QUERIES)) {
|
|
| 352 |
if (this.hasParameterChanged(ProgressionPreferences.QUERIES)) {
|
|
| 347 | 353 |
this.subTask("Finalizing...");
|
| 348 |
if (!stepFinalize()) {
|
|
| 354 |
if (!stepFinalize()) {
|
|
| 349 | 355 |
return false; |
| 350 | 356 |
} |
| 351 | 357 |
this.worked(20); |
| 352 | 358 |
} |
| 353 |
|
|
| 359 |
|
|
| 354 | 360 |
return true; |
| 355 | 361 |
} |
| 356 |
|
|
| 362 |
|
|
| 357 | 363 |
/** |
| 358 | 364 |
* Test discontinuity of a corpus. If the corpus is a subcorpus then test if the start-end of matches make no hole |
| 365 |
* |
|
| 359 | 366 |
* @param corpus |
| 360 | 367 |
* @return |
| 361 | 368 |
*/ |
| ... | ... | |
| 370 | 377 |
previousMatch = m; |
| 371 | 378 |
continue; |
| 372 | 379 |
} |
| 373 |
|
|
| 374 |
if (m.getStart() == previousMatch.getEnd()+1) {
|
|
| 380 |
|
|
| 381 |
if (m.getStart() == previousMatch.getEnd() + 1) {
|
|
| 375 | 382 |
previousMatch = m; |
| 376 | 383 |
continue; |
| 377 | 384 |
} |
| ... | ... | |
| 382 | 389 |
} |
| 383 | 390 |
return true; |
| 384 | 391 |
} |
| 385 |
|
|
| 392 |
|
|
| 386 | 393 |
/** |
| 387 | 394 |
* compute using already resolved QueryResults |
| 395 |
* |
|
| 388 | 396 |
* @param queryResults |
| 389 | 397 |
* @return |
| 390 | 398 |
* @throws CqiClientException |
| 391 | 399 |
*/ |
| 392 | 400 |
public boolean stepQueries(List<QueryResult> queryResults) throws CqiClientException {
|
| 393 | 401 |
if (queryResults.size() != queries.size()) return false; |
| 394 |
|
|
| 402 |
|
|
| 395 | 403 |
maxX = getCorpus().getSize(); |
| 396 | 404 |
int npositions = 0; |
| 397 |
for (int iQuery = 0 ; iQuery < queries.size() ; iQuery++) {
|
|
| 405 |
for (int iQuery = 0; iQuery < queries.size(); iQuery++) {
|
|
| 398 | 406 |
QueryResult result = queryResults.get(iQuery); |
| 399 |
|
|
| 407 |
|
|
| 400 | 408 |
int nLines = result.getNMatch(); |
| 401 | 409 |
if (maxY < nLines) {
|
| 402 | 410 |
maxY = nLines; |
| 403 | 411 |
} |
| 404 |
|
|
| 412 |
|
|
| 405 | 413 |
List<Match> matches = null; |
| 406 | 414 |
if (nLines > 0) {
|
| 407 | 415 |
matches = result.getMatches(0, nLines - 1); // get the indexes sequences of result's tokens |
| 408 | 416 |
} |
| 409 | 417 |
else {
|
| 410 |
matches = new ArrayList<Match>();
|
|
| 418 |
matches = new ArrayList<>(); |
|
| 411 | 419 |
} |
| 412 |
//System.out.println("matches: "+matches);
|
|
| 420 |
// System.out.println("matches: "+matches);
|
|
| 413 | 421 |
int[] positions = new int[matches.size()]; |
| 414 | 422 |
// System.out.println("nb positions "+matches.size());
|
| 415 | 423 |
// if (matches.size() > 0) |
| ... | ... | |
| 422 | 430 |
} |
| 423 | 431 |
return npositions > 0; |
| 424 | 432 |
} |
| 425 |
|
|
| 433 |
|
|
| 426 | 434 |
/** |
| 427 | 435 |
* Step queries. |
| 428 | 436 |
* |
| 429 | 437 |
* @throws CqiClientException the cqi client exception |
| 430 | 438 |
*/ |
| 431 |
public boolean stepQueries() throws CqiClientException {
|
|
| 439 |
public boolean stepQueries() throws CqiClientException {
|
|
| 432 | 440 |
maxX = getCorpus().getSize(); |
| 433 | 441 |
int npositions = 0; |
| 434 |
|
|
| 435 |
for (int iQuery = 0 ; iQuery < queries.size() ; iQuery++) {
|
|
| 442 |
|
|
| 443 |
for (int iQuery = 0; iQuery < queries.size(); iQuery++) {
|
|
| 436 | 444 |
QueryResult result = null; |
| 437 | 445 |
if (queryResults != null) {
|
| 438 | 446 |
result = queryResults.get(iQuery); |
| 439 |
} else {
|
|
| 447 |
} |
|
| 448 |
else {
|
|
| 440 | 449 |
CQLQuery query = queries.get(iQuery); |
| 441 |
// System.out.println("query "+query.getQueryString());
|
|
| 450 |
// System.out.println("query "+query.getQueryString());
|
|
| 442 | 451 |
result = getCorpus().query(query, query.getQueryString() |
| 443 | 452 |
.replace(" ", "_") + "_progression", true); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
| 444 | 453 |
} |
| 445 | 454 |
int nLines = result.getNMatch(); |
| 446 | 455 |
if (maxY < nLines) |
| 447 | 456 |
maxY = nLines; |
| 448 |
|
|
| 457 |
|
|
| 449 | 458 |
List<Match> matches = null; |
| 450 | 459 |
if (nLines > 0) {
|
| 451 | 460 |
matches = result.getMatches(0, nLines - 1); // get the indexes |
| 452 | 461 |
// sequences of result's tokens |
| 453 |
} else {
|
|
| 454 |
matches = new ArrayList<Match>(); |
|
| 455 | 462 |
} |
| 456 |
//System.out.println("matches: "+matches);
|
|
| 463 |
else {
|
|
| 464 |
matches = new ArrayList<>(); |
|
| 465 |
} |
|
| 466 |
// System.out.println("matches: "+matches);
|
|
| 457 | 467 |
int[] positions = new int[matches.size()]; |
| 458 | 468 |
// System.out.println("nb positions "+matches.size());
|
| 459 | 469 |
// if (matches.size() > 0) |
| ... | ... | |
| 467 | 477 |
} |
| 468 | 478 |
return npositions > 0; |
| 469 | 479 |
} |
| 470 |
|
|
| 480 |
|
|
| 471 | 481 |
/** |
| 472 | 482 |
* Step structural units. |
| 473 | 483 |
* |
| 474 | 484 |
* @throws CqiClientException the cqi client exception |
| 475 |
* @throws CqiServerError
|
|
| 476 |
* @throws IOException
|
|
| 485 |
* @throws CqiServerError |
|
| 486 |
* @throws IOException |
|
| 477 | 487 |
*/ |
| 478 |
public boolean stepStructuralUnits() throws CqiClientException, IOException, CqiServerError {
|
|
| 488 |
public boolean stepStructuralUnits() throws CqiClientException, IOException, CqiServerError {
|
|
| 479 | 489 |
// structure query |
| 480 | 490 |
if (structuralUnit != null) {
|
| 481 | 491 |
CQLQuery query = new CQLQuery("<" + structuralUnit.getName() + ">[]"); //$NON-NLS-1$ //$NON-NLS-2$
|
| ... | ... | |
| 483 | 493 |
if (!propertyRegex.isEmpty()) {
|
| 484 | 494 |
query = new CQLQuery("<" + structuralUnitProperty.getFullName() + "=\"" + propertyRegex + "\">[]"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
| 485 | 495 |
} |
| 486 |
else {
|
|
| 496 |
else {
|
|
| 487 | 497 |
query = new CQLQuery("<" + structuralUnitProperty.getFullName() + ">[]"); //$NON-NLS-1$ //$NON-NLS-2$
|
| 488 | 498 |
} |
| 489 | 499 |
} |
| 490 |
|
|
| 500 |
|
|
| 491 | 501 |
// System.out.println("Struct: "+ structure+" property: "+property);
|
| 492 | 502 |
// System.out.println(query.getQueryString()); |
| 493 | 503 |
QueryResult result = getCorpus().query(query, query.getQueryString().replace(" ", "_") + "_progression", true); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
| 494 | 504 |
int[] starts = result.getStarts(); |
| 495 | 505 |
result.drop(); |
| 496 | 506 |
int nLines = starts.length; |
| 497 |
|
|
| 498 |
// List<Match> matches = new ArrayList<Match>();
|
|
| 499 |
// if (nLines > 0)
|
|
| 500 |
// matches = result.getMatches(0, nLines - 1); // get the indexes
|
|
| 507 |
|
|
| 508 |
// List<Match> matches = new ArrayList<Match>();
|
|
| 509 |
// if (nLines > 0)
|
|
| 510 |
// matches = result.getMatches(0, nLines - 1); // get the indexes
|
|
| 501 | 511 |
// sequences of result's tokens |
| 502 |
|
|
| 512 |
|
|
| 503 | 513 |
structurePositions = new int[nLines]; |
| 504 | 514 |
structureNames = new String[nLines]; |
| 505 |
|
|
| 506 |
|
|
| 507 |
|
|
| 508 |
Map<Property, List<String>> refValues = new HashMap<Property, List<String>>();
|
|
| 515 |
|
|
| 516 |
|
|
| 517 |
|
|
| 518 |
Map<Property, List<String>> refValues = new HashMap<>(); |
|
| 509 | 519 |
if (structuralUnitProperty != null) {
|
| 510 | 520 |
List<String> values = CorpusManager.getCorpusManager().getCqiClient().getSingleData(structuralUnitProperty, starts); |
| 511 | 521 |
refValues.put(structuralUnitProperty, values); |
| 512 | 522 |
} |
| 513 |
|
|
| 523 |
|
|
| 514 | 524 |
int i = 0; |
| 515 | 525 |
String previousname = ""; //$NON-NLS-1$ |
| 516 | 526 |
String currentname = ""; //$NON-NLS-1$ |
| ... | ... | |
| 523 | 533 |
else {
|
| 524 | 534 |
currentname = structuralUnit.getName() + i; |
| 525 | 535 |
} |
| 526 |
|
|
| 527 |
// if (repeatValues) {
|
|
| 536 |
|
|
| 537 |
// if (repeatValues) {
|
|
| 528 | 538 |
structureNames[i] = currentname; |
| 529 | 539 |
structurePositions[i] = m; |
| 530 |
// }
|
|
| 531 |
// else {
|
|
| 532 |
// if (!previousname.equals(currentname)) {
|
|
| 533 |
// structureNames[i] = currentname;
|
|
| 534 |
// structurePositions[i] = m;
|
|
| 535 |
// }
|
|
| 536 |
// }
|
|
| 540 |
// }
|
|
| 541 |
// else {
|
|
| 542 |
// if (!previousname.equals(currentname)) {
|
|
| 543 |
// structureNames[i] = currentname;
|
|
| 544 |
// structurePositions[i] = m;
|
|
| 545 |
// }
|
|
| 546 |
// }
|
|
| 537 | 547 |
if (i == 0) {
|
| 538 | 548 |
structurePositions[i] += 1; |
| 539 | 549 |
} |
| 540 |
|
|
| 550 |
|
|
| 541 | 551 |
if (m - previousposition < distmin) {
|
| 542 | 552 |
distmin = m - previousposition; |
| 543 | 553 |
} |
| 544 |
|
|
| 554 |
|
|
| 545 | 555 |
previousname = currentname; |
| 546 | 556 |
previousposition = m; |
| 547 |
|
|
| 557 |
|
|
| 548 | 558 |
i++; |
| 549 |
|
|
| 559 |
|
|
| 550 | 560 |
this.worked(1); |
| 551 | 561 |
} |
| 552 | 562 |
this.bande = distmin * this.bandeMultiplier; |
| 553 | 563 |
} |
| 554 | 564 |
return true; |
| 555 | 565 |
} |
| 556 |
|
|
| 566 |
|
|
| 557 | 567 |
/** |
| 558 | 568 |
* Step finalize. |
| 559 | 569 |
* |
| 560 | 570 |
* @throws CqiClientException the cqi client exception |
| 561 | 571 |
*/ |
| 562 |
public boolean stepFinalize() throws CqiClientException {
|
|
| 572 |
public boolean stepFinalize() throws CqiClientException {
|
|
| 563 | 573 |
if (getCorpus() instanceof Subcorpus) {
|
| 564 | 574 |
List<Match> matches = ((Subcorpus) getCorpus()).getMatches(); |
| 565 | 575 |
if (matches.size() == 0) {
|
| 566 | 576 |
System.out.println(ProgressionCoreMessages.errorColonSubcorpusWithSize0); |
| 567 | 577 |
return false; |
| 568 | 578 |
} |
| 569 |
|
|
| 579 |
|
|
| 570 | 580 |
XminCorpus = matches.get(0).getStart(); |
| 571 |
XmaxCorpus = matches.get(matches.size()-1).getEnd();
|
|
| 581 |
XmaxCorpus = matches.get(matches.size() - 1).getEnd();
|
|
| 572 | 582 |
} |
| 573 | 583 |
else {
|
| 574 | 584 |
XminCorpus = 0; |
| 575 | 585 |
XmaxCorpus = getCorpus().getSize(); |
| 576 | 586 |
} |
| 577 |
//System.out.println("min: "+XminCorpus+ "max: "+XmaxCorpus);
|
|
| 578 |
|
|
| 587 |
// System.out.println("min: "+XminCorpus+ "max: "+XmaxCorpus);
|
|
| 588 |
|
|
| 579 | 589 |
if (bande <= 0) {
|
| 580 | 590 |
bande = ((XmaxCorpus - XminCorpus) / 100) * bandeMultiplier; |
| 581 | 591 |
if (bande == 0) {
|
| ... | ... | |
| 584 | 594 |
} |
| 585 | 595 |
return true; |
| 586 | 596 |
} |
| 587 |
|
|
| 588 |
|
|
| 589 |
|
|
| 590 |
|
|
| 591 |
|
|
| 597 |
|
|
| 598 |
|
|
| 599 |
|
|
| 600 |
|
|
| 601 |
|
|
| 592 | 602 |
/** |
| 593 | 603 |
* Gets the max x. |
| 594 | 604 |
* |
| ... | ... | |
| 597 | 607 |
public int getMaxX() {
|
| 598 | 608 |
return maxX; |
| 599 | 609 |
} |
| 600 |
|
|
| 610 |
|
|
| 601 | 611 |
/** |
| 602 | 612 |
* Gets the max y. |
| 603 | 613 |
* |
| ... | ... | |
| 606 | 616 |
public int getMaxY() {
|
| 607 | 617 |
return maxY; |
| 608 | 618 |
} |
| 609 |
|
|
| 619 |
|
|
| 610 | 620 |
/** |
| 611 | 621 |
* Gets the corpus. |
| 612 | 622 |
* |
| ... | ... | |
| 615 | 625 |
public CQPCorpus getCorpus() {
|
| 616 | 626 |
return (CQPCorpus) this.parent; |
| 617 | 627 |
} |
| 618 |
|
|
| 628 |
|
|
| 619 | 629 |
/** |
| 620 | 630 |
* Gets the queries. |
| 621 | 631 |
* |
| ... | ... | |
| 624 | 634 |
public List<CQLQuery> getQueries() {
|
| 625 | 635 |
return this.queries; |
| 626 | 636 |
} |
| 627 |
|
|
| 637 |
|
|
| 628 | 638 |
/** |
| 629 | 639 |
* @return the allPositions |
| 630 | 640 |
*/ |
| 631 | 641 |
public List<int[]> getAllPositions() {
|
| 632 | 642 |
return allPositions; |
| 633 | 643 |
} |
| 634 |
|
|
| 635 |
|
|
| 644 |
|
|
| 645 |
|
|
| 636 | 646 |
/** |
| 637 | 647 |
* Gets the positions counts. |
| 638 | 648 |
* |
| 639 | 649 |
* @return the positions counts |
| 640 | 650 |
*/ |
| 641 | 651 |
public List<Integer> getPositionsCounts() {
|
| 642 |
ArrayList<Integer> counts = new ArrayList<Integer>();
|
|
| 652 |
ArrayList<Integer> counts = new ArrayList<>(); |
|
| 643 | 653 |
for (int[] list : allPositions) {
|
| 644 | 654 |
counts.add(list.length); |
| 645 | 655 |
} |
| 646 | 656 |
return counts; |
| 647 | 657 |
} |
| 648 |
|
|
| 649 |
|
|
| 658 |
|
|
| 659 |
|
|
| 650 | 660 |
/** |
| 651 | 661 |
* Checks if is repeatvalues. |
| 652 | 662 |
* |
| ... | ... | |
| 655 | 665 |
public boolean isRepeatingValues() {
|
| 656 | 666 |
return repeatValues; |
| 657 | 667 |
} |
| 658 |
|
|
| 659 |
|
|
| 660 |
|
|
| 668 |
|
|
| 669 |
|
|
| 670 |
|
|
| 661 | 671 |
@Override |
| 662 | 672 |
public void clean() {
|
| 663 | 673 |
try {
|
| ... | ... | |
| 671 | 681 |
org.txm.utils.logger.Log.printStackTrace(e); |
| 672 | 682 |
} |
| 673 | 683 |
} |
| 674 |
|
|
| 684 |
|
|
| 675 | 685 |
@Override |
| 676 | 686 |
public boolean toTxt(File outfile, String encoding, String colseparator, |
| 677 | 687 |
String txtseparator) throws Exception {
|
| 678 | 688 |
return toTxt(outfile, encoding); |
| 679 | 689 |
} |
| 680 |
|
|
| 681 |
|
|
| 690 |
|
|
| 691 |
|
|
| 682 | 692 |
/** |
| 683 | 693 |
* |
| 684 | 694 |
* @param outfile |
| ... | ... | |
| 690 | 700 |
// NK: writer declared as class attribute to perform a clean if the operation is interrupted |
| 691 | 701 |
tryAcquireSemaphore(); |
| 692 | 702 |
this.writer = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(outfile), |
| 693 |
encoding));
|
|
| 694 |
for (int i = 0 ; i < queries.size() ; i++) {
|
|
| 695 |
writer.write("* "+queries.get(i)+"\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 703 |
encoding)); |
|
| 704 |
for (int i = 0; i < queries.size(); i++) {
|
|
| 705 |
writer.write("* " + queries.get(i) + "\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 696 | 706 |
for (int p : allPositions.get(i)) {
|
| 697 |
writer.write(""+p+"\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 707 |
writer.write("" + p + "\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 698 | 708 |
} |
| 699 | 709 |
} |
| 700 | 710 |
writer.close(); |
| 701 | 711 |
releaseSemaphore(); |
| 702 | 712 |
return true; |
| 703 | 713 |
} |
| 704 |
|
|
| 705 |
|
|
| 714 |
|
|
| 715 |
|
|
| 706 | 716 |
/** |
| 707 | 717 |
* @return the structurePositions |
| 708 | 718 |
*/ |
| 709 | 719 |
public int[] getStructurePositions() {
|
| 710 | 720 |
return structurePositions; |
| 711 | 721 |
} |
| 712 |
|
|
| 722 |
|
|
| 713 | 723 |
/** |
| 714 | 724 |
* @return the structureNames |
| 715 | 725 |
*/ |
| 716 | 726 |
public String[] getStructureNames() {
|
| 717 | 727 |
return structureNames; |
| 718 | 728 |
} |
| 719 |
|
|
| 720 |
|
|
| 729 |
|
|
| 730 |
|
|
| 721 | 731 |
/** |
| 722 | 732 |
* @return the propertyRegex |
| 723 | 733 |
*/ |
| 724 | 734 |
public String getPropertyRegex() {
|
| 725 | 735 |
return propertyRegex; |
| 726 | 736 |
} |
| 727 |
|
|
| 737 |
|
|
| 728 | 738 |
/** |
| 729 | 739 |
* @return the bande |
| 730 | 740 |
*/ |
| 731 | 741 |
public float getBande() {
|
| 732 | 742 |
return bande; |
| 733 | 743 |
} |
| 734 |
|
|
| 744 |
|
|
| 735 | 745 |
/** |
| 736 | 746 |
* @return the xminCorpus |
| 737 | 747 |
*/ |
| 738 | 748 |
public int getXminCorpus() {
|
| 739 | 749 |
return XminCorpus; |
| 740 | 750 |
} |
| 741 |
|
|
| 751 |
|
|
| 742 | 752 |
/** |
| 743 | 753 |
* @return the xmaxCorpus |
| 744 | 754 |
*/ |
| 745 | 755 |
public int getXmaxCorpus() {
|
| 746 | 756 |
return XmaxCorpus; |
| 747 | 757 |
} |
| 748 |
|
|
| 749 |
|
|
| 758 |
|
|
| 759 |
|
|
| 750 | 760 |
@Override |
| 751 | 761 |
public String[] getExportTXTExtensions() {
|
| 752 |
return new String[]{"*.txt"};
|
|
| 762 |
return new String[] { "*.txt" };
|
|
| 753 | 763 |
} |
| 754 |
|
|
| 755 |
|
|
| 764 |
|
|
| 765 |
|
|
| 756 | 766 |
/** |
| 757 | 767 |
* Computes and gets a query constructed from all the queries in alternative mode (with pipe |). |
| 768 |
* |
|
| 758 | 769 |
* @return |
| 759 | 770 |
*/ |
| 760 | 771 |
// FIXME: not used? even in synchronous editors (concordance and edition?)? |
| 761 |
public String getQueriesString() {
|
|
| 762 |
ArrayList<Integer> allQueriesIndexes = new ArrayList<Integer>(this.queries.size());
|
|
| 763 |
for(int i = 0; i < this.queries.size(); i++) {
|
|
| 772 |
public String getQueriesString() {
|
|
| 773 |
ArrayList<Integer> allQueriesIndexes = new ArrayList<>(this.queries.size()); |
|
| 774 |
for (int i = 0; i < this.queries.size(); i++) {
|
|
| 764 | 775 |
allQueriesIndexes.add(i); |
| 765 | 776 |
} |
| 766 |
return this.getQueriesString(allQueriesIndexes) ;
|
|
| 777 |
return this.getQueriesString(allQueriesIndexes); |
|
| 767 | 778 |
} |
| 768 |
|
|
| 779 |
|
|
| 769 | 780 |
/** |
| 770 | 781 |
* Computes and gets a query constructed from all the queries specified by their index in alternative mode (with pipe |). |
| 782 |
* |
|
| 771 | 783 |
* @param queriesIndexes |
| 772 | 784 |
* @return |
| 773 | 785 |
*/ |
| 774 |
public String getQueriesString(ArrayList<Integer> queriesIndexes) {
|
|
| 775 |
|
|
| 786 |
public String getQueriesString(ArrayList<Integer> queriesIndexes) {
|
|
| 787 |
|
|
| 776 | 788 |
String query = ""; |
| 777 |
|
|
| 778 |
for(int i = 0; i < this.queries.size(); i++) {
|
|
| 779 |
if(queriesIndexes.contains(i)) {
|
|
| 789 |
|
|
| 790 |
for (int i = 0; i < this.queries.size(); i++) {
|
|
| 791 |
if (queriesIndexes.contains(i)) {
|
|
| 780 | 792 |
String tmpQuery = this.queries.get(i).getQueryString(); |
| 781 |
// Remove "[" and "]" only if it's a simple query as "[look]"
|
|
| 782 |
if(tmpQuery.startsWith("[") && !tmpQuery.contains("=")) {
|
|
| 793 |
// Remove "[" and "]" only if it's a simple query as "[look]" |
|
| 794 |
if (tmpQuery.startsWith("[") && !tmpQuery.contains("=")) {
|
|
| 783 | 795 |
int length = tmpQuery.length(); |
| 784 |
if(tmpQuery.endsWith("]")) {
|
|
| 796 |
if (tmpQuery.endsWith("]")) {
|
|
| 785 | 797 |
length--; |
| 786 | 798 |
} |
| 787 | 799 |
tmpQuery = tmpQuery.substring(1, length); |
| 788 | 800 |
} |
| 789 |
if(query.length() > 0) {
|
|
| 801 |
if (query.length() > 0) {
|
|
| 790 | 802 |
query += " | "; |
| 791 | 803 |
} |
| 792 | 804 |
query += tmpQuery; |
| 793 | 805 |
} |
| 794 | 806 |
} |
| 795 |
|
|
| 807 |
|
|
| 796 | 808 |
return query; |
| 797 |
|
|
| 809 |
|
|
| 798 | 810 |
} |
| 799 |
|
|
| 800 |
|
|
| 811 |
|
|
| 812 |
|
|
| 801 | 813 |
@Override |
| 802 | 814 |
public String getDetails() {
|
| 803 | 815 |
return this.getName(); |
| 804 | 816 |
} |
| 805 |
|
|
| 817 |
|
|
| 806 | 818 |
@Override |
| 807 | 819 |
public String getName() {
|
| 808 | 820 |
try {
|
| 809 | 821 |
return this.parent.getSimpleName() + ": " + this.getSimpleName(); //$NON-NLS-1$ |
| 810 | 822 |
} |
| 811 |
catch(Exception e) {
|
|
| 823 |
catch (Exception e) {
|
|
| 812 | 824 |
} |
| 813 | 825 |
return ""; //$NON-NLS-1$ |
| 814 | 826 |
} |
| 815 |
|
|
| 827 |
|
|
| 816 | 828 |
@Override |
| 817 | 829 |
public String getSimpleName() {
|
| 818 | 830 |
try {
|
| 819 | 831 |
if (this.queries == null || this.queries.isEmpty()) {
|
| 820 | 832 |
return getEmptyName(); |
| 821 | 833 |
} |
| 822 |
String output = StringUtils.join(this.queries, ","); //$NON-NLS-1$
|
|
| 834 |
String output = StringUtils.join(this.queries, ","); //$NON-NLS-1$ |
|
| 823 | 835 |
if (this.propertyRegex != null) {
|
| 824 | 836 |
output += " " + this.propertyRegex; //$NON-NLS-1$ |
| 825 | 837 |
} |
| ... | ... | |
| 829 | 841 |
} |
| 830 | 842 |
return this.getEmptyName(); |
| 831 | 843 |
} |
| 832 |
|
|
| 833 |
|
|
| 844 |
|
|
| 845 |
|
|
| 834 | 846 |
@Override |
| 835 | 847 |
public String getComputingStartMessage() {
|
| 836 |
return TXMCoreMessages.bind(ProgressionCoreMessages.progressionOfP0InTheP1Corpus, this.queries.toString(), this.getCorpus().getName());
|
|
| 848 |
return TXMCoreMessages.bind(ProgressionCoreMessages.progressionOfP0InTheP1Corpus, this.queries, this.getCorpus().getName()); |
|
| 837 | 849 |
} |
| 838 |
|
|
| 839 |
|
|
| 850 |
|
|
| 851 |
|
|
| 840 | 852 |
@Override |
| 841 | 853 |
public String getComputingDoneMessage() {
|
| 842 |
if (this.getAllPositions() == null || this.getAllPositions().isEmpty()) {
|
|
| 854 |
if (this.getAllPositions() == null || this.getAllPositions().isEmpty()) {
|
|
| 843 | 855 |
return TXMCoreMessages.common_noResults; |
| 844 | 856 |
} |
| 845 | 857 |
else {
|
| ... | ... | |
| 850 | 862 |
return TXMCoreMessages.bind(TXMCoreMessages.common_P0Positions, positions); |
| 851 | 863 |
} |
| 852 | 864 |
} |
| 853 |
|
|
| 865 |
|
|
| 854 | 866 |
/** |
| 855 | 867 |
* @return the bandeMultiplier |
| 856 | 868 |
*/ |
| 857 | 869 |
public float getBandeMultiplier() {
|
| 858 | 870 |
return bandeMultiplier; |
| 859 | 871 |
} |
| 860 |
|
|
| 872 |
|
|
| 861 | 873 |
/** |
| 862 | 874 |
* @param bandeMultiplier the bandeMultiplier to set |
| 863 | 875 |
*/ |
| 864 | 876 |
public void setBandeMultiplier(float bandeMultiplier) {
|
| 865 | 877 |
this.bandeMultiplier = bandeMultiplier; |
| 866 | 878 |
} |
| 867 |
|
|
| 868 |
|
|
| 869 |
|
|
| 879 |
|
|
| 880 |
|
|
| 881 |
|
|
| 870 | 882 |
/** |
| 871 | 883 |
* @param repeatValues the repeatValues to set |
| 872 | 884 |
*/ |
| 873 | 885 |
public void setRepeatValues(boolean repeatValues) {
|
| 874 | 886 |
this.repeatValues = repeatValues; |
| 875 | 887 |
} |
| 876 |
|
|
| 877 |
|
|
| 878 |
|
|
| 879 |
|
|
| 888 |
|
|
| 889 |
|
|
| 890 |
|
|
| 891 |
|
|
| 880 | 892 |
/** |
| 881 | 893 |
* @return the structuralUnit |
| 882 | 894 |
*/ |
| 883 | 895 |
public StructuralUnit getStructuralUnit() {
|
| 884 | 896 |
return structuralUnit; |
| 885 | 897 |
} |
| 886 |
|
|
| 887 |
|
|
| 888 |
|
|
| 898 |
|
|
| 899 |
|
|
| 900 |
|
|
| 889 | 901 |
/** |
| 890 | 902 |
* @return the structuralUnitProperty |
| 891 | 903 |
*/ |
| 892 | 904 |
public StructuralUnitProperty getStructuralUnitProperty() {
|
| 893 | 905 |
return structuralUnitProperty; |
| 894 | 906 |
} |
| 895 |
|
|
| 896 |
|
|
| 897 |
|
|
| 907 |
|
|
| 908 |
|
|
| 909 |
|
|
| 898 | 910 |
/** |
| 899 | 911 |
* @param queries the queries to set |
| 900 | 912 |
*/ |
| 901 | 913 |
public void setQueries(List<CQLQuery> queries) {
|
| 902 | 914 |
this.queries = queries; |
| 903 | 915 |
} |
| 904 |
|
|
| 916 |
|
|
| 905 | 917 |
@Override |
| 906 | 918 |
public String getResultType() {
|
| 907 | 919 |
return ProgressionCoreMessages.RESULT_TYPE; |
| 908 | 920 |
} |
| 909 |
|
|
| 910 |
|
|
| 911 |
} |
|
| 921 |
|
|
| 922 |
|
|
| 923 |
} |
|
Formats disponibles : Unified diff