Révision 3230
| tmp/org.txm.searchengine.cqp.core/src/org/txm/importer/cwb/CwbMakeAll.java (revision 3230) | ||
|---|---|---|
| 35 | 35 |
import org.txm.libs.cqp.CQPLibPreferences; |
| 36 | 36 |
import org.txm.searchengine.cqp.clientExceptions.ServerNotFoundException; |
| 37 | 37 |
import org.txm.utils.OSDetector; |
| 38 |
import org.txm.utils.io.IOUtils; |
|
| 38 | 39 |
import org.txm.utils.logger.Log; |
| 39 | 40 |
|
| 40 | 41 |
// TODO: Auto-generated Javadoc |
| ... | ... | |
| 160 | 161 |
if (debug) {
|
| 161 | 162 |
cmd.add(1, "-D"); //$NON-NLS-1$ |
| 162 | 163 |
} |
| 163 |
return run(cmd, monitorOutput, true); // always wait for cwb-makeall to end |
|
| 164 |
boolean ret = run(cmd, monitorOutput, true); // always wait for cwb-makeall to end |
|
| 165 |
|
|
| 166 |
// if (ret) {
|
|
| 167 |
// File regFile = new File(pathToRegistry); |
|
| 168 |
// ReadRegistryFile registry = new ReadRegistryFile(regFile); |
|
| 169 |
// registry.read(); |
|
| 170 |
// |
|
| 171 |
// File dataDirectory = registry.getDataDirectory(); |
|
| 172 |
// |
|
| 173 |
// for (String struct : registry.getSAttributesMap().keySet()) {
|
|
| 174 |
// for (String attr : registry.getSAttributesMap().get(struct)) {
|
|
| 175 |
// File avxFile = new File(dataDirectory, attr+".avx"); |
|
| 176 |
// File avsFile = new File(dataDirectory, attr+".avs"); |
|
| 177 |
// File rngFile = new File(dataDirectory, attr+".rng"); |
|
| 178 |
// File cpos2idFile = new File(dataDirectory, attr+".corpus.idx"); |
|
| 179 |
// File id2StrFile = new File(dataDirectory, attr+".lexicon.idx"); |
|
| 180 |
// |
|
| 181 |
// if (avsFile.exists()) {
|
|
| 182 |
// System.out.println("Building id -> cpos: "+id2StrFile);
|
|
| 183 |
// ArrayList<String> strings = new ArrayList<String>(); |
|
| 184 |
// String dataString = IOUtils.getText(avsFile); |
|
| 185 |
// |
|
| 186 |
// System.out.println("Building cpos -> id: "+cpos2idFile);
|
|
| 187 |
// } |
|
| 188 |
// } |
|
| 189 |
// } |
|
| 190 |
// |
|
| 191 |
// return ret; |
|
| 192 |
// } else {
|
|
| 193 |
// return false; |
|
| 194 |
// } |
|
| 195 |
return ret; |
|
| 164 | 196 |
} |
| 165 | 197 |
|
| 166 | 198 |
/** |
| tmp/org.txm.searchengine.cqp.core/src/org/txm/importer/cwb/ReadRegistryFile.java (revision 3230) | ||
|---|---|---|
| 20 | 20 |
*/ |
| 21 | 21 |
public class ReadRegistryFile {
|
| 22 | 22 |
|
| 23 |
File registryFile; |
|
| 23 |
File registryFile, dataDirectory;
|
|
| 24 | 24 |
|
| 25 |
String id; |
|
| 26 |
|
|
| 25 | 27 |
ArrayList<String> pAttributes; |
| 26 | 28 |
|
| 27 | 29 |
ArrayList<String> sAttributes; |
| ... | ... | |
| 47 | 49 |
for (String line : IOUtils.getLines(registryFile, System.getProperty("file.encoding"))) {
|
| 48 | 50 |
line = line.trim(); // remove first tab |
| 49 | 51 |
|
| 50 |
if (line.startsWith("ATTRIBUTE ")) {
|
|
| 52 |
if (line.startsWith("HOME ")) {
|
|
| 53 |
line = line.substring(5); // remove 'ATTRIBUTE ' |
|
| 54 |
if (line.startsWith("\"") && line.endsWith("\"")) {
|
|
| 55 |
line = line.substring(1, line.length() - 1); |
|
| 56 |
} |
|
| 57 |
dataDirectory = new File(line); |
|
| 58 |
} else if (line.startsWith("ID ")) {
|
|
| 59 |
line = line.substring(3); // remove 'ATTRIBUTE ' |
|
| 60 |
id = line; |
|
| 61 |
} else if (line.startsWith("ATTRIBUTE ")) {
|
|
| 51 | 62 |
line = line.substring(10); // remove 'ATTRIBUTE ' |
| 52 | 63 |
pAttributes.add(line); |
| 53 | 64 |
} |
| ... | ... | |
| 90 | 101 |
} |
| 91 | 102 |
|
| 92 | 103 |
/** |
| 104 |
* |
|
| 105 |
* @return the value of the ID field |
|
| 106 |
*/ |
|
| 107 |
public String getID() {
|
|
| 108 |
return id; |
|
| 109 |
} |
|
| 110 |
|
|
| 111 |
/** |
|
| 112 |
* |
|
| 113 |
* @return the data directory pointed by the value of the HOME field |
|
| 114 |
*/ |
|
| 115 |
public File getDataDirectory() {
|
|
| 116 |
return dataDirectory; |
|
| 117 |
} |
|
| 118 |
|
|
| 119 |
/** |
|
| 93 | 120 |
* Test the CQP index files of each p-attribute and s-attribute properties |
| 94 | 121 |
* |
| 95 | 122 |
* @param dataDirectory the directory where the binary files should be found |
| tmp/org.txm.searchengine.cqp.core/src/org/txm/searchengine/cqp/core/preferences/SubcorpusPreferences.java (revision 3230) | ||
|---|---|---|
| 9 | 9 |
|
| 10 | 10 |
public static String OPTIMIZED_MODE = "optimized_mode"; |
| 11 | 11 |
|
| 12 |
public static final String TARGET_STRATEGY = "target_strategy"; |
|
| 13 |
|
|
| 12 | 14 |
public SubcorpusPreferences() {
|
| 13 | 15 |
// TODO Auto-generated constructor stub |
| 14 | 16 |
} |
| tmp/org.txm.searchengine.cqp.core/src/org/txm/searchengine/cqp/core/functions/summary/Summary.java (revision 3230) | ||
|---|---|---|
| 75 | 75 |
|
| 76 | 76 |
QueryResult result = corpus.query(new CQLQuery("<" + properties.get(j).getFullName() + ">[]"), "summary", false); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
| 77 | 77 |
int[] starts = result.getStarts(); |
| 78 |
result.drop(); |
|
| 78 |
//result.drop();
|
|
| 79 | 79 |
result = corpus.query(new CQLQuery("[]</" + properties.get(j).getFullName() + ">"), "summary", false); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
| 80 | 80 |
int[] ends = result.getStarts(); |
| 81 |
result.drop(); |
|
| 81 |
//result.drop();
|
|
| 82 | 82 |
// System.out.println("starts: "+Arrays.toString(starts));
|
| 83 | 83 |
// System.out.println("ends: "+Arrays.toString(ends));
|
| 84 | 84 |
int[] strcutpos = CQPSearchEngine.getCqiClient().cpos2Struc(properties.get(j).getQualifiedName(), starts); |
| tmp/org.txm.searchengine.cqp.core/src/org/txm/searchengine/cqp/corpus/StructuralUnitProperty.java (revision 3230) | ||
|---|---|---|
| 337 | 337 |
List<String> values = getValues(); |
| 338 | 338 |
return values.toArray(new String[values.size()]); |
| 339 | 339 |
} |
| 340 |
|
|
| 341 |
@Override |
|
| 342 |
public int[] cpos2Id(int[] positions) throws Exception {
|
|
| 343 |
|
|
| 344 |
// TODO Auto-generated method stub |
|
| 345 |
return null; |
|
| 346 |
} |
|
| 347 |
|
|
| 348 |
@Override |
|
| 349 |
public String[] id2Str(int[] positions) throws Exception {
|
|
| 350 |
|
|
| 351 |
// TODO Auto-generated method stub |
|
| 352 |
return null; |
|
| 353 |
} |
|
| 340 | 354 |
} |
| tmp/org.txm.searchengine.cqp.core/src/org/txm/searchengine/cqp/corpus/CQPCorpus.java (revision 3230) | ||
|---|---|---|
| 27 | 27 |
// |
| 28 | 28 |
package org.txm.searchengine.cqp.corpus; |
| 29 | 29 |
|
| 30 |
import java.io.File; |
|
| 30 | 31 |
import java.io.IOException; |
| 31 | 32 |
import java.util.ArrayList; |
| 32 | 33 |
import java.util.Collection; |
| ... | ... | |
| 546 | 547 |
return null; |
| 547 | 548 |
} |
| 548 | 549 |
|
| 549 |
CorpusManager.getCorpusManager().getCqiClient().cqpQuery(this.getQualifiedCqpId(), subcorpusCqpId, CQLQuery.fixQuery(query.getQueryString())); |
|
| 550 |
//CorpusManager.getCorpusManager().getCqiClient().cqpQuery(this.getQualifiedCqpId(), subcorpusCqpId, CQLQuery.fixQuery(query.getQueryString()));
|
|
| 550 | 551 |
// System.out.println("SUBCORPUS: "+subcorpusCqpId+"
|
| 551 | 552 |
// q="+query.getQueryString()); |
| 552 | 553 |
subcorpus = new Subcorpus(this); |
| ... | ... | |
| 607 | 608 |
* @param elem the elem |
| 608 | 609 |
* @return the subcorpus |
| 609 | 610 |
* @throws CqiClientException the cqi client exception |
| 611 |
* @throws InterruptedException |
|
| 610 | 612 |
*/ |
| 611 |
public Subcorpus createSubcorpus(String name, SelectionResult selectionResult) throws CqiClientException {
|
|
| 613 |
public Subcorpus createSubcorpus(String name, SelectionResult selectionResult) throws CqiClientException, InterruptedException {
|
|
| 612 | 614 |
Subcorpus sub = createSubcorpus(this.getStructuralUnit("text"), this.getStructuralUnit("text").getProperty("id"), selectionResult, name); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
| 613 | 615 |
sub.setSelectionResult(selectionResult); |
| 616 |
sub.compute(); |
|
| 614 | 617 |
return sub; |
| 615 | 618 |
} |
| 616 | 619 |
|
| ... | ... | |
| 1432 | 1435 |
QueryResult queryResult = null; |
| 1433 | 1436 |
// String queryResultId = queryResultNamePrefix + UUID.randomUUID().toString(); |
| 1434 | 1437 |
String queryResultId = queryResultNamePrefix + getNextQueryCounter(); |
| 1435 |
String fixedQuery = CQLQuery.fixQuery(query.getQueryString()); |
|
| 1436 | 1438 |
|
| 1437 |
if (fixedQuery.length() > CQLQuery.MAX_CQL_LENGTH) {
|
|
| 1439 |
|
|
| 1440 |
if (query.getQueryString().length() > CQLQuery.MAX_CQL_LENGTH) {
|
|
| 1438 | 1441 |
try {
|
| 1439 |
Log.warning(NLS.bind("Error: CQL length limit reached ({0}) with {1} length.", CQLQuery.MAX_CQL_LENGTH, fixedQuery.length()));
|
|
| 1442 |
Log.warning(NLS.bind("Error: CQL length limit reached ({0}) with {1} length.", CQLQuery.MAX_CQL_LENGTH, query.getQueryString().length()));
|
|
| 1440 | 1443 |
return new FakeQueryResult(queryResultName, this, query, new int[0], new int[0], new int[0]); |
| 1441 | 1444 |
} |
| 1442 | 1445 |
catch (InvalidCqpIdException e) {
|
| ... | ... | |
| 1445 | 1448 |
} |
| 1446 | 1449 |
} |
| 1447 | 1450 |
|
| 1448 |
Log.finest(NLS.bind(TXMCoreMessages.queryOnP0ColonP1InfP2, new String[] { this.getQualifiedCqpId(), queryResultId, fixedQuery })); // $NON-NLS-1$
|
|
| 1451 |
Log.finest(NLS.bind(TXMCoreMessages.queryOnP0ColonP1InfP2, new String[] { this.getQualifiedCqpId(), queryResultId, query.getQueryString() })); // $NON-NLS-1$
|
|
| 1449 | 1452 |
// long start = System.currentTimeMillis(); |
| 1450 | 1453 |
try {
|
| 1451 |
CorpusManager.getCorpusManager().getCqiClient().cqpQuery(this.getQualifiedCqpId(), queryResultId, fixedQuery); |
|
| 1454 |
//CorpusManager.getCorpusManager().getCqiClient().cqpQuery(this.getQualifiedCqpId(), queryResultId, fixedQuery); |
|
| 1455 |
|
|
| 1456 |
String q = query.getQueryString(); |
|
| 1457 |
if (q.matches("undump \"[^ ]+\";")) { // undump query
|
|
| 1458 |
String path = q.substring(8, q.length() - 2); |
|
| 1459 |
if (!new File(path).exists()) {
|
|
| 1460 |
Log.severe(NLS.bind("Error: could not compute subcorpus from the dump file: {0}.", path));
|
|
| 1461 |
return null; |
|
| 1462 |
} |
|
| 1463 |
CorpusManager.getCorpusManager().getCqiClient().query(";");
|
|
| 1464 |
CorpusManager.getCorpusManager().getCqiClient().query("undump "+queryResultId+" < "+q.substring(7));
|
|
| 1465 |
} else if (q.matches("(union|join|intersect|inter|difference|diff) [^ ]+ [^ ]+;")) {
|
|
| 1466 |
CorpusManager.getCorpusManager().getCqiClient().query(";");
|
|
| 1467 |
CorpusManager.getCorpusManager().getCqiClient().query(queryResultId+" = "+q); |
|
| 1468 |
} else {
|
|
| 1469 |
//String fixedQuery = CQLQuery.fixQuery(query.getQueryString()); |
|
| 1470 |
CorpusManager.getCorpusManager().getCqiClient().cqpQuery(this.getQualifiedCqpId(), queryResultId, CQLQuery.fixQuery(query.getQueryString())); |
|
| 1471 |
} |
|
| 1472 |
|
|
| 1452 | 1473 |
queryResult = new QueryResult(queryResultId, queryResultName, this, query); |
| 1453 | 1474 |
|
| 1454 | 1475 |
// if (save) {
|
| tmp/org.txm.searchengine.cqp.core/src/org/txm/searchengine/cqp/corpus/Property.java (revision 3230) | ||
|---|---|---|
| 33 | 33 |
import org.apache.commons.lang.StringUtils; |
| 34 | 34 |
import org.txm.core.messages.TXMCoreMessages; |
| 35 | 35 |
import org.txm.core.preferences.TXMPreferences; |
| 36 |
import org.txm.searchengine.core.SearchEngine; |
|
| 36 | 37 |
import org.txm.searchengine.core.SearchEngineProperty; |
| 38 |
import org.txm.searchengine.core.SearchEnginesManager; |
|
| 37 | 39 |
import org.txm.searchengine.cqp.clientExceptions.CqiClientException; |
| 38 | 40 |
import org.txm.utils.logger.Log; |
| 39 | 41 |
|
| ... | ... | |
| 45 | 47 |
* |
| 46 | 48 |
* @author Jean-Philippe Magué, mdecorde |
| 47 | 49 |
*/ |
| 48 |
public abstract class Property implements Comparable<Property>, SearchEngineProperty {
|
|
| 49 |
|
|
| 50 |
/** The name. */ |
|
| 51 |
protected String name; |
|
| 52 |
|
|
| 50 |
public abstract class Property extends org.txm.searchengine.core.Property implements Comparable<Property>, SearchEngineProperty {
|
|
| 51 |
|
|
| 53 | 52 |
/** The corpus. */ |
| 54 | 53 |
protected CQPCorpus corpus; |
| 55 | 54 |
|
| ... | ... | |
| 64 | 63 |
* the corpus |
| 65 | 64 |
*/ |
| 66 | 65 |
protected Property(String name, CQPCorpus corpus) {
|
| 67 |
this.name = name;
|
|
| 66 |
super(name);
|
|
| 68 | 67 |
this.corpus = corpus; |
| 69 | 68 |
} |
| 70 | 69 |
|
| 70 |
public SearchEngine getSearchEngine() {
|
|
| 71 |
return SearchEnginesManager.getCQPSearchEngine(); |
|
| 72 |
} |
|
| 73 |
|
|
| 71 | 74 |
/** |
| 72 | 75 |
* Gets the qualified name = corpus name + "." + this.getFullName() . |
| 73 | 76 |
* |
| ... | ... | |
| 84 | 87 |
@Override |
| 85 | 88 |
public abstract String getFullName(); |
| 86 | 89 |
|
| 87 |
/** |
|
| 88 |
* Gets the name. |
|
| 89 |
* |
|
| 90 |
* @return the name |
|
| 91 |
*/ |
|
| 92 |
@Override |
|
| 93 |
public String getName() {
|
|
| 94 |
return name; |
|
| 95 |
} |
|
| 96 | 90 |
|
| 97 | 91 |
/** |
| 98 | 92 |
* Gets the corpus over which the property is defined. |
| tmp/org.txm.searchengine.cqp.core/src/org/txm/searchengine/cqp/corpus/Subcorpus.java (revision 3230) | ||
|---|---|---|
| 80 | 80 |
* CQP text ids cache |
| 81 | 81 |
*/ |
| 82 | 82 |
protected String[] cqpTextIDS = null; |
| 83 |
|
|
| 84 |
|
|
| 83 | 85 |
|
| 84 | 86 |
/** |
| 85 | 87 |
* |
| ... | ... | |
| 130 | 132 |
Log.warning(NLS.bind("Error: CQL length limit reached ({0}) with {1} length.", CQLQuery.MAX_CQL_LENGTH, pQuery.getQueryString().length()));
|
| 131 | 133 |
return false; |
| 132 | 134 |
} |
| 133 |
CorpusManager.getCorpusManager().getCqiClient().cqpQuery(parent_id, this.pID, CQLQuery.fixQuery(this.pQuery.getQueryString())); |
|
| 134 | 135 |
|
| 136 |
String q = this.pQuery.getQueryString(); |
|
| 137 |
if (q.matches("undump [^ ]+;")) { // undump query
|
|
| 138 |
String path = q.substring(8, q.length() - 2); |
|
| 139 |
if (!new File(path).exists()) {
|
|
| 140 |
Log.severe(NLS.bind("Error: could not compute subcorpus from the dump file: {0}.", path));
|
|
| 141 |
return false; |
|
| 142 |
} |
|
| 143 |
CorpusManager.getCorpusManager().getCqiClient().query(";");
|
|
| 144 |
CorpusManager.getCorpusManager().getCqiClient().query("undump "+this.pID+" < "+q.substring(7));
|
|
| 145 |
} else if (q.matches("(union|join|intersect|inter|difference|diff) [^ ]+ [^ ]+;")) {
|
|
| 146 |
CorpusManager.getCorpusManager().getCqiClient().query(";");
|
|
| 147 |
CorpusManager.getCorpusManager().getCqiClient().query(this.pID+" = "+q); |
|
| 148 |
} else {
|
|
| 149 |
CorpusManager.getCorpusManager().getCqiClient().cqpQuery(parent_id, this.pID, CQLQuery.fixQuery(q)); |
|
| 150 |
// if (q.contains("@[") || q.contains("@\"")) {
|
|
| 151 |
// Log.fine("Target detected in "+q+", recentering on target.");
|
|
| 152 |
// CorpusManager.getCorpusManager().getCqiClient().cqpQuery(this.getQualifiedCqpId(), this.pID, "<target>[]"); |
|
| 153 |
// } |
|
| 154 |
} |
|
| 155 |
|
|
| 135 | 156 |
this.qresult = new QueryResult(this.pID, this.getUserName(), this.getCorpusParent(), this.pQuery); // getCorpusParent().query(pQuery, this.pID, true); |
| 136 | 157 |
} |
| 137 | 158 |
|
Formats disponibles : Unified diff