Révision 3198
| tmp/org.txm.cooccurrence.core/src/org/txm/cooccurrence/core/functions/Cooccurrence.java (revision 3198) | ||
|---|---|---|
| 41 | 41 |
import java.util.Set; |
| 42 | 42 |
|
| 43 | 43 |
import org.eclipse.osgi.util.NLS; |
| 44 |
import org.rosuda.REngine.REXPMismatchException; |
|
| 45 | 44 |
import org.txm.concordance.core.functions.Concordance; |
| 46 | 45 |
import org.txm.concordance.core.functions.Line; |
| 47 | 46 |
import org.txm.cooccurrence.core.functions.comparators.CLineComparator; |
| ... | ... | |
| 57 | 56 |
import org.txm.searchengine.cqp.clientExceptions.CqiClientException; |
| 58 | 57 |
import org.txm.searchengine.cqp.clientExceptions.UnexpectedAnswerException; |
| 59 | 58 |
import org.txm.searchengine.cqp.corpus.CQPCorpus; |
| 60 |
import org.txm.searchengine.cqp.corpus.CorpusManager; |
|
| 61 | 59 |
import org.txm.searchengine.cqp.corpus.Property; |
| 62 | 60 |
import org.txm.searchengine.cqp.corpus.QueryResult; |
| 63 | 61 |
import org.txm.searchengine.cqp.corpus.StructuralUnit; |
| 64 |
import org.txm.searchengine.cqp.corpus.StructuralUnitProperty; |
|
| 65 |
import org.txm.searchengine.cqp.corpus.VirtualProperty; |
|
| 66 | 62 |
import org.txm.searchengine.cqp.corpus.WordProperty; |
| 67 |
import org.txm.searchengine.cqp.corpus.query.Match; |
|
| 68 | 63 |
import org.txm.searchengine.cqp.corpus.query.CQLQuery; |
| 64 |
import org.txm.searchengine.cqp.corpus.query.Match; |
|
| 69 | 65 |
import org.txm.searchengine.cqp.serverException.CqiServerError; |
| 70 | 66 |
import org.txm.specificities.core.statsengine.r.function.SpecificitiesR; |
| 71 | 67 |
import org.txm.statsengine.core.StatException; |
| ... | ... | |
| 76 | 72 |
import org.txm.utils.logger.Log; |
| 77 | 73 |
|
| 78 | 74 |
/** |
| 79 |
* Compute a cooccurrence from a concordance.
|
|
| 75 |
* Compute a cooccurrence from a pivot query.
|
|
| 80 | 76 |
* |
| 81 | 77 |
* @author mdecorde |
| 82 | 78 |
* |
| 83 | 79 |
*/ |
| 84 | 80 |
public class Cooccurrence extends TXMResult {
|
| 85 | 81 |
|
| 82 |
/** |
|
| 83 |
* The Class CLine. |
|
| 84 |
*/ |
|
| 85 |
public class CLine {
|
|
| 86 |
|
|
| 87 |
/** The cooc. */ |
|
| 88 |
Cooccurrence cooc; |
|
| 89 |
|
|
| 90 |
/** The distmoyenne. */ |
|
| 91 |
public float distmoyenne; |
|
| 92 |
|
|
| 93 |
/** The freq. */ |
|
| 94 |
public int freq; |
|
| 95 |
|
|
| 96 |
/** The id. */ |
|
| 97 |
public int id; |
|
| 98 |
|
|
| 99 |
/** The mode. */ |
|
| 100 |
public long mode; |
|
| 101 |
|
|
| 102 |
/** The nbocc. */ |
|
| 103 |
public int nbocc; |
|
| 104 |
|
|
| 105 |
/** The occ. */ |
|
| 106 |
public String occ; |
|
| 107 |
|
|
| 108 |
/** The props. */ |
|
| 109 |
public List<String> props; |
|
| 110 |
|
|
| 111 |
/** The score. */ |
|
| 112 |
public double score; |
|
| 113 |
|
|
| 114 |
/** |
|
| 115 |
* Instantiates a new c line. |
|
| 116 |
* |
|
| 117 |
* @param cooc the cooc |
|
| 118 |
* @param occ the occ |
|
| 119 |
* @param props the props |
|
| 120 |
* @param nbocc the nbocc |
|
| 121 |
* @param freq the freq |
|
| 122 |
* @param score the score |
|
| 123 |
* @param distmoyenne the distmoyenne |
|
| 124 |
* @param mode the mode |
|
| 125 |
*/ |
|
| 126 |
public CLine(Cooccurrence cooc, String occ, List<String> props, |
|
| 127 |
int nbocc, int freq, double score, Float distmoyenne, long mode) {
|
|
| 128 |
this.occ = occ; |
|
| 129 |
this.props = props; |
|
| 130 |
this.freq = freq; |
|
| 131 |
this.nbocc = nbocc; |
|
| 132 |
this.score = score; |
|
| 133 |
this.distmoyenne = distmoyenne; |
|
| 134 |
this.mode = mode; |
|
| 135 |
this.cooc = cooc; |
|
| 136 |
} |
|
| 137 |
|
|
| 138 |
/** |
|
| 139 |
* Adds the txt sep. |
|
| 140 |
* |
|
| 141 |
* @param str the str |
|
| 142 |
* @param sep the sep |
|
| 143 |
* @return the string |
|
| 144 |
*/ |
|
| 145 |
private String addTxtSep(String str, String sep) {
|
|
| 146 |
return sep + str.replace(sep, sep + sep) + sep; |
|
| 147 |
} |
|
| 148 |
|
|
| 149 |
/** |
|
| 150 |
* Gets the cooc. |
|
| 151 |
* |
|
| 152 |
* @return the cooc |
|
| 153 |
*/ |
|
| 154 |
public Cooccurrence getCooc() {
|
|
| 155 |
return cooc; |
|
| 156 |
} |
|
| 157 |
|
|
| 158 |
/** |
|
| 159 |
* Resume. |
|
| 160 |
* |
|
| 161 |
* @param colseparator the colseparator |
|
| 162 |
* @param txtseparator the txtseparator |
|
| 163 |
* @return the string |
|
| 164 |
*/ |
|
| 165 |
public String resume(String colseparator, String txtseparator) {
|
|
| 166 |
return addTxtSep("" + occ, txtseparator) //$NON-NLS-1$
|
|
| 167 |
+ colseparator + freq + colseparator + nbocc + colseparator + score + colseparator + distmoyenne; |
|
| 168 |
} |
|
| 169 |
|
|
| 170 |
/** |
|
| 171 |
* Sets the count and dist. |
|
| 172 |
* |
|
| 173 |
* @param count the count |
|
| 174 |
* @param dist the dist |
|
| 175 |
*/ |
|
| 176 |
public void setCountAndDist(int count, int dist) {
|
|
| 177 |
this.nbocc = count; |
|
| 178 |
this.distmoyenne = dist; |
|
| 179 |
} |
|
| 180 |
|
|
| 181 |
/** |
|
| 182 |
* Sets the freq. |
|
| 183 |
* |
|
| 184 |
* @param freq the new freq |
|
| 185 |
*/ |
|
| 186 |
public void setFreq(int freq) {
|
|
| 187 |
this.freq = freq; |
|
| 188 |
} |
|
| 189 |
|
|
| 190 |
/** |
|
| 191 |
* Sets the score. |
|
| 192 |
*/ |
|
| 193 |
public void setScore() {// FB == freq, R = nbocc
|
|
| 194 |
this.score = (this.nbocc + this.freq + this.distmoyenne); |
|
| 195 |
} |
|
| 196 |
|
|
| 197 |
@Override |
|
| 198 |
public String toString() {
|
|
| 199 |
return occ + CooccurrenceCoreMessages.fColon + freq + CooccurrenceCoreMessages.occColon + nbocc + CooccurrenceCoreMessages.scoreColon + score + CooccurrenceCoreMessages.meanDistColon |
|
| 200 |
+ distmoyenne + CooccurrenceCoreMessages.propertiesColon + props; |
|
| 201 |
} |
|
| 202 |
} |
|
| 203 |
|
|
| 86 | 204 |
/** The nocooc. */ |
| 87 | 205 |
protected static int nocooc = 1; |
| 88 | 206 |
|
| ... | ... | |
| 92 | 210 |
/** The allsignaturesstr. */ |
| 93 | 211 |
private HashMap<Integer, String> allsignaturesstr; |
| 94 | 212 |
|
| 95 |
/** The anticontextquery. */
|
|
| 213 |
/** The anti-context query used to trim the nearest contexts cooccurrents. */
|
|
| 96 | 214 |
private CQLQuery anticontextquery; |
| 97 | 215 |
|
| 98 | 216 |
private boolean buildLexicalTableWithCooccurrents; |
| ... | ... | |
| 243 | 361 |
super(parametersNodePath); |
| 244 | 362 |
} |
| 245 | 363 |
|
| 246 |
|
|
| 247 |
|
|
| 248 | 364 |
@Override |
| 249 | 365 |
protected boolean _compute(TXMProgressMonitor monitor) throws CqiClientException, IOException, CqiServerError, StatException {
|
| 250 | 366 |
// System.out.println("cooc: "+corpus+" "+query+" "+properties+" "+limit+" "+maxLeft+" "+minLeft+" "+minRight+" "+maxRight+" "+minFreq+" "+minCof+" "+minScore+" "+includeXpivot);
|
| ... | ... | |
| 307 | 423 |
return true; |
| 308 | 424 |
} |
| 309 | 425 |
|
| 426 |
/** |
|
| 427 |
* To txt. |
|
| 428 |
* |
|
| 429 |
* @param outfile the outfile |
|
| 430 |
* @param encoding the encoding |
|
| 431 |
* @param colseparator the colseparator |
|
| 432 |
* @param txtseparator the txtseparator |
|
| 433 |
* @return true, if successful |
|
| 434 |
*/ |
|
| 310 | 435 |
@Override |
| 311 |
public boolean loadParameters() throws CqiClientException {
|
|
| 312 |
pProperties = (List<WordProperty>) Property.stringToProperties(getCorpus(), this.getStringParameterValue(TXMPreferences.UNIT_PROPERTIES)); |
|
| 313 |
pQuery = new CQLQuery(this.getStringParameterValue(TXMPreferences.QUERY)); |
|
| 314 |
pStructuralUnitLimit = this.getCorpus().getStructuralUnit(this.getStringParameterValue(CooccurrencePreferences.STRUCTURAL_UNIT_LIMIT)); |
|
| 315 |
return true; |
|
| 316 |
} |
|
| 317 |
|
|
| 318 |
@Override |
|
| 319 |
public boolean saveParameters() {
|
|
| 320 |
this.saveParameter(TXMPreferences.UNIT_PROPERTIES, Property.propertiesToString(this.pProperties)); |
|
| 321 |
|
|
| 322 |
if (pQuery != null) {
|
|
| 323 |
this.saveParameter(TXMPreferences.QUERY, pQuery.getQueryString()); |
|
| 436 |
public boolean _toTxt(File outfile, String encoding, String colseparator, String txtseparator) {
|
|
| 437 |
try {
|
|
| 438 |
// NK: writer declared as class attribute to perform a clean if the operation is interrupted |
|
| 439 |
this.writer = new BufferedWriter(new OutputStreamWriter( |
|
| 440 |
new FileOutputStream(outfile), encoding)); |
|
| 441 |
// if ("UTF-8".equals(encoding)) writer.write('\ufeff'); // UTF-8 BOM
|
|
| 442 |
toTxt(writer, colseparator, txtseparator); |
|
| 324 | 443 |
} |
| 325 |
|
|
| 326 |
if (pStructuralUnitLimit != null) {
|
|
| 327 |
this.saveParameter(CooccurrencePreferences.STRUCTURAL_UNIT_LIMIT, this.pStructuralUnitLimit.getName());
|
|
| 444 |
catch (Exception e) {
|
|
| 445 |
Log.severe(TXMCoreMessages.bind(TXMCoreMessages.error_error, e));
|
|
| 446 |
return false;
|
|
| 328 | 447 |
} |
| 329 |
|
|
| 330 | 448 |
return true; |
| 331 | 449 |
} |
| 332 | 450 |
|
| ... | ... | |
| 493 | 611 |
// } |
| 494 | 612 |
|
| 495 | 613 |
/** |
| 496 |
* Gets the corpus.
|
|
| 614 |
* Creates a CQL query string from the specified lines.
|
|
| 497 | 615 |
* |
| 498 |
* @return the corpus |
|
| 616 |
* @param lines |
|
| 617 |
* @return the query |
|
| 499 | 618 |
*/ |
| 500 |
public CQPCorpus getCorpus() {
|
|
| 501 |
return (CQPCorpus) this.getParent(); |
|
| 502 |
} |
|
| 619 |
public String createQuery(List<CLine> lines) {
|
|
| 503 | 620 |
|
| 504 |
@Override |
|
| 505 |
public String getDetails() {
|
|
| 506 |
Object[] params = new Object[] { this.getParent(), this.pQuery, this.pProperties, this.pStructuralUnitLimit, (this.pMinLeftContextSize - 1), (this.pMaxLeftContextSize - 1),
|
|
| 507 |
(this.pMinRightContextSize - 1), |
|
| 508 |
(this.pMaxRightContextSize - 1), this.pFminFilter, this.pFCoocFilter, this.pScoreMinFilter }; |
|
| 509 |
return NLS.bind(CooccurrenceCoreMessages.info_details, params); |
|
| 510 |
} |
|
| 511 |
|
|
| 512 |
@Override |
|
| 513 |
public String getName() {
|
|
| 514 |
try {
|
|
| 515 |
return this.getParent().getName() + TXMPreferences.PARENT_NAME_SEPARATOR + this.getSimpleName(); |
|
| 621 |
if (this.getQuery().isEmpty()) {
|
|
| 622 |
return ""; //$NON-NLS-1$ |
|
| 516 | 623 |
} |
| 517 |
catch (Exception e) {
|
|
| 518 |
return this.getSimpleName(); |
|
| 519 |
} |
|
| 520 |
} |
|
| 521 | 624 |
|
| 522 | 625 |
|
| 523 |
@Override |
|
| 524 |
public String getSimpleName() {
|
|
| 525 |
if (this.pQuery != null && !this.pQuery.isEmpty()) {
|
|
| 526 |
StringBuffer output = new StringBuffer(); |
|
| 527 |
output.append(this.pQuery.asString()); |
|
| 528 |
output.append(WordProperty.asString(this.pProperties)); |
|
| 626 |
int nbProps = this.getProperties().size(); |
|
| 627 |
List<WordProperty> props = this.getProperties(); |
|
| 529 | 628 |
|
| 530 |
if (this.pMaxLeftContextSize > 0 && this.pMaxRightContextSize > 0) {
|
|
| 531 |
output.append(" " + (this.pMaxLeftContextSize - 1) + " " + (this.pMaxRightContextSize - 1)); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 629 |
String query = "@["; //$NON-NLS-1$ |
|
| 630 |
for (int p = 0; p < nbProps; p++) {
|
|
| 631 |
|
|
| 632 |
if (props.get(p) instanceof WordProperty) {
|
|
| 633 |
ArrayList<String> values = new ArrayList<String>(); |
|
| 634 |
for (int l = 0; l < lines.size(); l++) {
|
|
| 635 |
CLine line = lines.get(l); |
|
| 636 |
String s = line.props.get(p); |
|
| 637 |
values.add(s); |
|
| 638 |
} |
|
| 639 |
String test = ((WordProperty)props.get(p)).getCQLTest(values); |
|
| 640 |
if (test != null) {
|
|
| 641 |
query += test; |
|
| 642 |
} |
|
| 532 | 643 |
} |
| 533 |
output.append(TXMCoreMessages.formatMinFilter(this.pFminFilter)); |
|
| 534 |
output.append(TXMCoreMessages.formatMinFilter(this.pFCoocFilter)); |
|
| 535 |
output.append(TXMCoreMessages.formatMinFilter(this.pScoreMinFilter)); |
|
| 644 |
if (p < nbProps-1) {
|
|
| 645 |
query += " & "; //$NON-NLS-1$ |
|
| 646 |
} |
|
| 647 |
} |
|
| 648 |
query += "] "; //$NON-NLS-1$ |
|
| 536 | 649 |
|
| 537 |
// TODO: SJ: improve the hiding or display of value according to the default preferences values |
|
| 538 |
// output.append(TXMCoreMessages.formatMinFilter(this.pFminFilter, CooccurrencePreferences.getInstance().getInt(CooccurrencePreferences.F_MIN) + 1)); |
|
| 539 |
// output.append(TXMCoreMessages.formatMinFilter(this.pFCoocFilter, CooccurrencePreferences.getInstance().getInt(CooccurrencePreferences.MIN_COUNT) + 1)); |
|
| 540 |
// output.append(TXMCoreMessages.formatMinFilter(this.pScoreMinFilter, CooccurrencePreferences.getInstance().getDouble(CooccurrencePreferences.MIN_SCORE) + 1)); |
|
| 650 |
int maxempan = Math.max(this.getMaxLeft(), this.getMaxRight()); |
|
| 651 |
if (this.getIncludeXPivot() && maxempan == 0) maxempan = 1; |
|
| 541 | 652 |
|
| 653 |
String maxempanstr = "within " + maxempan + " "; //$NON-NLS-1$ //$NON-NLS-2$ |
|
| 654 |
if (this.getStructuralUnitLimit() != null) maxempanstr += this.getStructuralUnitLimit().getName(); |
|
| 542 | 655 |
|
| 543 |
return output.toString(); |
|
| 656 |
|
|
| 657 |
String pquery = CQLQuery.fixQuery(this.getQuery().getQueryString()); |
|
| 658 |
if (this.getMaxLeft() == 0) {
|
|
| 659 |
query = "" + pquery + " []* " + query + " " + maxempanstr; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ |
|
| 544 | 660 |
} |
| 661 |
else if (this.getMaxRight() == 0) {
|
|
| 662 |
query = "" + query + " []* " + pquery + " " + maxempanstr; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ |
|
| 663 |
} |
|
| 545 | 664 |
else {
|
| 546 |
return this.getEmptyName();
|
|
| 665 |
query = "(" + pquery + " []* " + query + ") | (" + query + " []* " + pquery + ") " + maxempanstr; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$ //$NON-NLS-5$
|
|
| 547 | 666 |
} |
| 548 |
} |
|
| 549 | 667 |
|
| 550 |
|
|
| 551 |
@Override |
|
| 552 |
public String getComputingStartMessage() {
|
|
| 553 |
return TXMCoreMessages.bind(CooccurrenceCoreMessages.cooccurrentsOfP0PropertieP1InTheP2Corpus, (this.pQuery != null?this.pQuery.asString():"<no query>"), WordProperty.asString(this.pProperties), |
|
| 554 |
(this.pMaxLeftContextSize - 1), (this.pMaxRightContextSize - 1), this.pFminFilter, this.pFCoocFilter, this.pScoreMinFilter, this.getCorpus().getName()); |
|
| 668 |
return query; |
|
| 555 | 669 |
} |
| 556 | 670 |
|
| 557 |
|
|
| 558 | 671 |
@Override |
| 559 | 672 |
public String getComputingDoneMessage() {
|
| 560 | 673 |
if (this.lines.isEmpty()) {
|
| ... | ... | |
| 565 | 678 |
} |
| 566 | 679 |
} |
| 567 | 680 |
|
| 681 |
@Override |
|
| 682 |
public String getComputingStartMessage() {
|
|
| 683 |
return TXMCoreMessages.bind(CooccurrenceCoreMessages.cooccurrentsOfP0PropertieP1InTheP2Corpus, (this.pQuery != null?this.pQuery.asString():"<no query>"), WordProperty.asString(this.pProperties), |
|
| 684 |
(this.pMaxLeftContextSize - 1), (this.pMaxRightContextSize - 1), this.pFminFilter, this.pFCoocFilter, this.pScoreMinFilter, this.getCorpus().getName()); |
|
| 685 |
} |
|
| 568 | 686 |
|
| 569 | 687 |
|
| 570 | 688 |
/** |
| 689 |
* Gets the corpus. |
|
| 690 |
* |
|
| 691 |
* @return the corpus |
|
| 692 |
*/ |
|
| 693 |
public CQPCorpus getCorpus() {
|
|
| 694 |
return (CQPCorpus) this.getParent(); |
|
| 695 |
} |
|
| 696 |
|
|
| 697 |
|
|
| 698 |
@Override |
|
| 699 |
public String getDetails() {
|
|
| 700 |
Object[] params = new Object[] { this.getParent(), this.pQuery, this.pProperties, this.pStructuralUnitLimit, (this.pMinLeftContextSize - 1), (this.pMaxLeftContextSize - 1),
|
|
| 701 |
(this.pMinRightContextSize - 1), |
|
| 702 |
(this.pMaxRightContextSize - 1), this.pFminFilter, this.pFCoocFilter, this.pScoreMinFilter }; |
|
| 703 |
return NLS.bind(CooccurrenceCoreMessages.info_details, params); |
|
| 704 |
} |
|
| 705 |
|
|
| 706 |
|
|
| 707 |
/** |
|
| 571 | 708 |
* Gets the fA. |
| 572 | 709 |
* |
| 573 | 710 |
* @return the fA |
| ... | ... | |
| 625 | 762 |
return pMinRightContextSize; |
| 626 | 763 |
} |
| 627 | 764 |
|
| 765 |
@Override |
|
| 766 |
public String getName() {
|
|
| 767 |
try {
|
|
| 768 |
return this.getParent().getName() + TXMPreferences.PARENT_NAME_SEPARATOR + this.getSimpleName(); |
|
| 769 |
} |
|
| 770 |
catch (Exception e) {
|
|
| 771 |
return this.getSimpleName(); |
|
| 772 |
} |
|
| 773 |
} |
|
| 774 |
|
|
| 628 | 775 |
/** |
| 629 | 776 |
* Gets the lines. |
| 630 | 777 |
* |
| ... | ... | |
| 689 | 836 |
} |
| 690 | 837 |
|
| 691 | 838 |
|
| 839 |
@Override |
|
| 840 |
public String getResultType() {
|
|
| 841 |
return CooccurrenceCoreMessages.RESULT_TYPE; |
|
| 842 |
} |
|
| 843 |
|
|
| 844 |
@Override |
|
| 845 |
public String getSimpleName() {
|
|
| 846 |
if (this.pQuery != null && !this.pQuery.isEmpty()) {
|
|
| 847 |
StringBuffer output = new StringBuffer(); |
|
| 848 |
output.append(this.pQuery.asString()); |
|
| 849 |
output.append(WordProperty.asString(this.pProperties)); |
|
| 850 |
|
|
| 851 |
if (this.pMaxLeftContextSize > 0 && this.pMaxRightContextSize > 0) {
|
|
| 852 |
output.append(" " + (this.pMaxLeftContextSize - 1) + " " + (this.pMaxRightContextSize - 1)); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 853 |
} |
|
| 854 |
output.append(TXMCoreMessages.formatMinFilter(this.pFminFilter)); |
|
| 855 |
output.append(TXMCoreMessages.formatMinFilter(this.pFCoocFilter)); |
|
| 856 |
output.append(TXMCoreMessages.formatMinFilter(this.pScoreMinFilter)); |
|
| 857 |
|
|
| 858 |
// TODO: SJ: improve the hiding or display of value according to the default preferences values |
|
| 859 |
// output.append(TXMCoreMessages.formatMinFilter(this.pFminFilter, CooccurrencePreferences.getInstance().getInt(CooccurrencePreferences.F_MIN) + 1)); |
|
| 860 |
// output.append(TXMCoreMessages.formatMinFilter(this.pFCoocFilter, CooccurrencePreferences.getInstance().getInt(CooccurrencePreferences.MIN_COUNT) + 1)); |
|
| 861 |
// output.append(TXMCoreMessages.formatMinFilter(this.pScoreMinFilter, CooccurrencePreferences.getInstance().getDouble(CooccurrencePreferences.MIN_SCORE) + 1)); |
|
| 862 |
|
|
| 863 |
|
|
| 864 |
return output.toString(); |
|
| 865 |
} |
|
| 866 |
else {
|
|
| 867 |
return this.getEmptyName(); |
|
| 868 |
} |
|
| 869 |
} |
|
| 870 |
|
|
| 692 | 871 |
/** |
| 693 | 872 |
* Gets the structural unit limit. |
| 694 | 873 |
* |
| ... | ... | |
| 772 | 951 |
return false; |
| 773 | 952 |
} |
| 774 | 953 |
|
| 954 |
@Override |
|
| 955 |
public boolean loadParameters() throws CqiClientException {
|
|
| 956 |
pProperties = (List<WordProperty>) Property.stringToProperties(getCorpus(), this.getStringParameterValue(TXMPreferences.UNIT_PROPERTIES)); |
|
| 957 |
pQuery = new CQLQuery(this.getStringParameterValue(TXMPreferences.QUERY)); |
|
| 958 |
pStructuralUnitLimit = this.getCorpus().getStructuralUnit(this.getStringParameterValue(CooccurrencePreferences.STRUCTURAL_UNIT_LIMIT)); |
|
| 959 |
return true; |
|
| 960 |
} |
|
| 961 |
|
|
| 775 | 962 |
/** |
| 776 | 963 |
* Prints the. |
| 777 | 964 |
*/ |
| ... | ... | |
| 784 | 971 |
System.out.println(line.resume("\t", "")); //$NON-NLS-1$ //$NON-NLS-2$
|
| 785 | 972 |
} |
| 786 | 973 |
|
| 974 |
@Override |
|
| 975 |
public boolean saveParameters() {
|
|
| 976 |
this.saveParameter(TXMPreferences.UNIT_PROPERTIES, Property.propertiesToString(this.pProperties)); |
|
| 977 |
|
|
| 978 |
if (pQuery != null) {
|
|
| 979 |
this.saveParameter(TXMPreferences.QUERY, pQuery.getQueryString()); |
|
| 980 |
} |
|
| 981 |
|
|
| 982 |
if (pStructuralUnitLimit != null) {
|
|
| 983 |
this.saveParameter(CooccurrencePreferences.STRUCTURAL_UNIT_LIMIT, this.pStructuralUnitLimit.getName()); |
|
| 984 |
} |
|
| 985 |
|
|
| 986 |
return true; |
|
| 987 |
} |
|
| 988 |
|
|
| 787 | 989 |
public void setCoocQuery(String q) {
|
| 788 | 990 |
pCooccurentQueryFilter = q; |
| 789 | 991 |
} |
| 790 | 992 |
|
| 993 |
public void setIncludeXpivot(boolean b) {
|
|
| 994 |
pIncludeXpivot = b; |
|
| 995 |
} |
|
| 996 |
|
|
| 791 | 997 |
/** |
| 792 | 998 |
* Sets the max left. |
| 793 | 999 |
* |
| ... | ... | |
| 841 | 1047 |
this.buildLexicalTableWithCooccurrents = buildLexicalTableWithCooccurrents; |
| 842 | 1048 |
} |
| 843 | 1049 |
|
| 844 |
public void setIncludeXpivot(boolean b) {
|
|
| 845 |
pIncludeXpivot = b; |
|
| 846 |
} |
|
| 847 |
|
|
| 848 | 1050 |
@Override |
| 849 | 1051 |
public boolean setParameters(TXMParameters parameters) {
|
| 850 | 1052 |
try {
|
| ... | ... | |
| 888 | 1090 |
return true; |
| 889 | 1091 |
} |
| 890 | 1092 |
|
| 1093 |
/** |
|
| 1094 |
* Sets the query. |
|
| 1095 |
* |
|
| 1096 |
* @param query |
|
| 1097 |
*/ |
|
| 1098 |
public void setQuery(CQLQuery query) {
|
|
| 1099 |
this.pQuery = query; |
|
| 1100 |
} |
|
| 1101 |
|
|
| 891 | 1102 |
public void setReferenceCorpus(String symbol) {
|
| 892 | 1103 |
referenceCorpus = symbol; |
| 893 | 1104 |
} |
| ... | ... | |
| 1448 | 1659 |
return toTxt(outfile, encoding, "\t", ""); //$NON-NLS-1$ //$NON-NLS-2$ |
| 1449 | 1660 |
} |
| 1450 | 1661 |
|
| 1451 |
/** |
|
| 1452 |
* To txt. |
|
| 1453 |
* |
|
| 1454 |
* @param outfile the outfile |
|
| 1455 |
* @param encoding the encoding |
|
| 1456 |
* @param colseparator the colseparator |
|
| 1457 |
* @param txtseparator the txtseparator |
|
| 1458 |
* @return true, if successful |
|
| 1459 |
*/ |
|
| 1460 |
@Override |
|
| 1461 |
public boolean _toTxt(File outfile, String encoding, String colseparator, String txtseparator) {
|
|
| 1462 |
try {
|
|
| 1463 |
// NK: writer declared as class attribute to perform a clean if the operation is interrupted |
|
| 1464 |
this.writer = new BufferedWriter(new OutputStreamWriter( |
|
| 1465 |
new FileOutputStream(outfile), encoding)); |
|
| 1466 |
// if ("UTF-8".equals(encoding)) writer.write('\ufeff'); // UTF-8 BOM
|
|
| 1467 |
toTxt(writer, colseparator, txtseparator); |
|
| 1468 |
} |
|
| 1469 |
catch (Exception e) {
|
|
| 1470 |
Log.severe(TXMCoreMessages.bind(TXMCoreMessages.error_error, e)); |
|
| 1471 |
return false; |
|
| 1472 |
} |
|
| 1473 |
return true; |
|
| 1474 |
} |
|
| 1475 | 1662 |
|
| 1476 | 1663 |
/** |
| 1477 | 1664 |
* To txt. |
| ... | ... | |
| 1502 | 1689 |
} |
| 1503 | 1690 |
|
| 1504 | 1691 |
|
| 1505 |
/** |
|
| 1506 |
* Sets the query. |
|
| 1507 |
* |
|
| 1508 |
* @param query |
|
| 1509 |
*/ |
|
| 1510 |
public void setQuery(CQLQuery query) {
|
|
| 1511 |
this.pQuery = query; |
|
| 1512 |
} |
|
| 1513 |
|
|
| 1514 |
/** |
|
| 1515 |
* Creates a CQL query string from the specified lines. |
|
| 1516 |
* |
|
| 1517 |
* @param lines |
|
| 1518 |
* @return the query |
|
| 1519 |
*/ |
|
| 1520 |
public String createQuery(List<CLine> lines) {
|
|
| 1521 |
|
|
| 1522 |
if (this.getQuery().isEmpty()) {
|
|
| 1523 |
return ""; //$NON-NLS-1$ |
|
| 1524 |
} |
|
| 1525 |
|
|
| 1526 |
|
|
| 1527 |
int nbProps = this.getProperties().size(); |
|
| 1528 |
List<WordProperty> props = this.getProperties(); |
|
| 1529 |
|
|
| 1530 |
String query = "@["; //$NON-NLS-1$ |
|
| 1531 |
for (int p = 0; p < nbProps; p++) {
|
|
| 1532 |
|
|
| 1533 |
if (props.get(p) instanceof WordProperty) {
|
|
| 1534 |
ArrayList<String> values = new ArrayList<String>(); |
|
| 1535 |
for (int l = 0; l < lines.size(); l++) {
|
|
| 1536 |
CLine line = lines.get(l); |
|
| 1537 |
String s = line.props.get(p); |
|
| 1538 |
values.add(s); |
|
| 1539 |
} |
|
| 1540 |
String test = ((WordProperty)props.get(p)).getCQLTest(values); |
|
| 1541 |
if (test != null) {
|
|
| 1542 |
query += test; |
|
| 1543 |
} |
|
| 1544 |
} |
|
| 1545 |
if (p < nbProps-1) {
|
|
| 1546 |
query += " & "; //$NON-NLS-1$ |
|
| 1547 |
} |
|
| 1548 |
} |
|
| 1549 |
query += "] "; //$NON-NLS-1$ |
|
| 1550 |
|
|
| 1551 |
int maxempan = Math.max(this.getMaxLeft(), this.getMaxRight()); |
|
| 1552 |
if (this.getIncludeXPivot() && maxempan == 0) maxempan = 1; |
|
| 1553 |
|
|
| 1554 |
String maxempanstr = "within " + maxempan + " "; //$NON-NLS-1$ //$NON-NLS-2$ |
|
| 1555 |
if (this.getStructuralUnitLimit() != null) maxempanstr += this.getStructuralUnitLimit().getName(); |
|
| 1556 |
|
|
| 1557 |
|
|
| 1558 |
String pquery = CQLQuery.fixQuery(this.getQuery().getQueryString()); |
|
| 1559 |
if (this.getMaxLeft() == 0) {
|
|
| 1560 |
query = "" + pquery + " []* " + query + " " + maxempanstr; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ |
|
| 1561 |
} |
|
| 1562 |
else if (this.getMaxRight() == 0) {
|
|
| 1563 |
query = "" + query + " []* " + pquery + " " + maxempanstr; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ |
|
| 1564 |
} |
|
| 1565 |
else {
|
|
| 1566 |
query = "(" + pquery + " []* " + query + ") | (" + query + " []* " + pquery + ") " + maxempanstr; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$ //$NON-NLS-5$
|
|
| 1567 |
} |
|
| 1568 |
|
|
| 1569 |
return query; |
|
| 1570 |
} |
|
| 1571 |
|
|
| 1572 |
|
|
| 1573 |
/** |
|
| 1574 |
* The Class CLine. |
|
| 1575 |
*/ |
|
| 1576 |
public class CLine {
|
|
| 1577 |
|
|
| 1578 |
/** The cooc. */ |
|
| 1579 |
Cooccurrence cooc; |
|
| 1580 |
|
|
| 1581 |
/** The distmoyenne. */ |
|
| 1582 |
public float distmoyenne; |
|
| 1583 |
|
|
| 1584 |
/** The freq. */ |
|
| 1585 |
public int freq; |
|
| 1586 |
|
|
| 1587 |
/** The id. */ |
|
| 1588 |
public int id; |
|
| 1589 |
|
|
| 1590 |
/** The mode. */ |
|
| 1591 |
public long mode; |
|
| 1592 |
|
|
| 1593 |
/** The nbocc. */ |
|
| 1594 |
public int nbocc; |
|
| 1595 |
|
|
| 1596 |
/** The occ. */ |
|
| 1597 |
public String occ; |
|
| 1598 |
|
|
| 1599 |
/** The props. */ |
|
| 1600 |
public List<String> props; |
|
| 1601 |
|
|
| 1602 |
/** The score. */ |
|
| 1603 |
public double score; |
|
| 1604 |
|
|
| 1605 |
/** |
|
| 1606 |
* Instantiates a new c line. |
|
| 1607 |
* |
|
| 1608 |
* @param cooc the cooc |
|
| 1609 |
* @param occ the occ |
|
| 1610 |
* @param props the props |
|
| 1611 |
* @param nbocc the nbocc |
|
| 1612 |
* @param freq the freq |
|
| 1613 |
* @param score the score |
|
| 1614 |
* @param distmoyenne the distmoyenne |
|
| 1615 |
* @param mode the mode |
|
| 1616 |
*/ |
|
| 1617 |
public CLine(Cooccurrence cooc, String occ, List<String> props, |
|
| 1618 |
int nbocc, int freq, double score, Float distmoyenne, long mode) {
|
|
| 1619 |
this.occ = occ; |
|
| 1620 |
this.props = props; |
|
| 1621 |
this.freq = freq; |
|
| 1622 |
this.nbocc = nbocc; |
|
| 1623 |
this.score = score; |
|
| 1624 |
this.distmoyenne = distmoyenne; |
|
| 1625 |
this.mode = mode; |
|
| 1626 |
this.cooc = cooc; |
|
| 1627 |
} |
|
| 1628 |
|
|
| 1629 |
/** |
|
| 1630 |
* Adds the txt sep. |
|
| 1631 |
* |
|
| 1632 |
* @param str the str |
|
| 1633 |
* @param sep the sep |
|
| 1634 |
* @return the string |
|
| 1635 |
*/ |
|
| 1636 |
private String addTxtSep(String str, String sep) {
|
|
| 1637 |
return sep + str.replace(sep, sep + sep) + sep; |
|
| 1638 |
} |
|
| 1639 |
|
|
| 1640 |
/** |
|
| 1641 |
* Gets the cooc. |
|
| 1642 |
* |
|
| 1643 |
* @return the cooc |
|
| 1644 |
*/ |
|
| 1645 |
public Cooccurrence getCooc() {
|
|
| 1646 |
return cooc; |
|
| 1647 |
} |
|
| 1648 |
|
|
| 1649 |
/** |
|
| 1650 |
* Resume. |
|
| 1651 |
* |
|
| 1652 |
* @param colseparator the colseparator |
|
| 1653 |
* @param txtseparator the txtseparator |
|
| 1654 |
* @return the string |
|
| 1655 |
*/ |
|
| 1656 |
public String resume(String colseparator, String txtseparator) {
|
|
| 1657 |
return addTxtSep("" + occ, txtseparator) //$NON-NLS-1$
|
|
| 1658 |
+ colseparator + freq + colseparator + nbocc + colseparator + score + colseparator + distmoyenne; |
|
| 1659 |
} |
|
| 1660 |
|
|
| 1661 |
/** |
|
| 1662 |
* Sets the count and dist. |
|
| 1663 |
* |
|
| 1664 |
* @param count the count |
|
| 1665 |
* @param dist the dist |
|
| 1666 |
*/ |
|
| 1667 |
public void setCountAndDist(int count, int dist) {
|
|
| 1668 |
this.nbocc = count; |
|
| 1669 |
this.distmoyenne = dist; |
|
| 1670 |
} |
|
| 1671 |
|
|
| 1672 |
/** |
|
| 1673 |
* Sets the freq. |
|
| 1674 |
* |
|
| 1675 |
* @param freq the new freq |
|
| 1676 |
*/ |
|
| 1677 |
public void setFreq(int freq) {
|
|
| 1678 |
this.freq = freq; |
|
| 1679 |
} |
|
| 1680 |
|
|
| 1681 |
/** |
|
| 1682 |
* Sets the score. |
|
| 1683 |
*/ |
|
| 1684 |
public void setScore() {// FB == freq, R = nbocc
|
|
| 1685 |
this.score = (this.nbocc + this.freq + this.distmoyenne); |
|
| 1686 |
} |
|
| 1687 |
|
|
| 1688 |
@Override |
|
| 1689 |
public String toString() {
|
|
| 1690 |
return occ + CooccurrenceCoreMessages.fColon + freq + CooccurrenceCoreMessages.occColon + nbocc + CooccurrenceCoreMessages.scoreColon + score + CooccurrenceCoreMessages.meanDistColon |
|
| 1691 |
+ distmoyenne + CooccurrenceCoreMessages.propertiesColon + props; |
|
| 1692 |
} |
|
| 1693 |
} |
|
| 1694 |
|
|
| 1695 |
|
|
| 1696 |
@Override |
|
| 1697 |
public String getResultType() {
|
|
| 1698 |
return CooccurrenceCoreMessages.RESULT_TYPE; |
|
| 1699 |
} |
|
| 1700 |
|
|
| 1701 |
|
|
| 1702 | 1692 |
} |
Formats disponibles : Unified diff