Révision 3883
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/repository/TypedValue.java (revision 3883) | ||
|---|---|---|
| 46 | 46 |
} |
| 47 | 47 |
|
| 48 | 48 |
public String toString() {
|
| 49 |
return PK.toString() + "(" + name + ")";
|
|
| 49 |
return PK.toString() + "(" + name + ")"; //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 50 | 50 |
} |
| 51 | 51 |
|
| 52 | 52 |
public String toHumanString() {
|
| 53 |
return name + "(" + PK.getId() + ")";
|
|
| 53 |
return name + "(" + PK.getId() + ")"; //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 54 | 54 |
} |
| 55 | 55 |
} |
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/repository/AnnotationEffect.java (revision 3883) | ||
|---|---|---|
| 17 | 17 |
* @return Effet suffix used to display the annotation type in messages |
| 18 | 18 |
*/ |
| 19 | 19 |
public String toSuffix() {
|
| 20 |
if (this.equals(TOKEN)) return " (T)"; |
|
| 21 |
else if (this.equals(ANNOTATION)) return " (A)"; |
|
| 22 |
else return ""; |
|
| 20 |
if (this.equals(TOKEN)) return " (T)"; //$NON-NLS-1$
|
|
| 21 |
else if (this.equals(ANNOTATION)) return " (A)"; //$NON-NLS-1$
|
|
| 22 |
else return ""; //$NON-NLS-1$
|
|
| 23 | 23 |
} |
| 24 | 24 |
} |
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/repository/TypedValuePK.java (revision 3883) | ||
|---|---|---|
| 44 | 44 |
} |
| 45 | 45 |
|
| 46 | 46 |
public String toString() {
|
| 47 |
return typeId + "=" + id; |
|
| 47 |
return typeId + "=" + id; //$NON-NLS-1$
|
|
| 48 | 48 |
} |
| 49 | 49 |
} |
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/repository/KnowledgeRepository.java (revision 3883) | ||
|---|---|---|
| 19 | 19 |
|
| 20 | 20 |
public abstract class KnowledgeRepository { // extends Vector<AnnotationType>
|
| 21 | 21 |
|
| 22 |
protected String values_base_url = ""; |
|
| 22 |
protected String values_base_url = ""; //$NON-NLS-1$
|
|
| 23 | 23 |
|
| 24 |
protected String types_base_url = ""; |
|
| 24 |
protected String types_base_url = ""; //$NON-NLS-1$
|
|
| 25 | 25 |
|
| 26 |
protected String url = ""; |
|
| 26 |
protected String url = ""; //$NON-NLS-1$
|
|
| 27 | 27 |
|
| 28 |
protected String dbPath = ""; // the KR database path |
|
| 28 |
protected String dbPath = ""; // the KR database path //$NON-NLS-1$
|
|
| 29 | 29 |
|
| 30 | 30 |
protected String name; |
| 31 | 31 |
|
| ... | ... | |
| 45 | 45 |
|
| 46 | 46 |
private EntityManagerFactory emf; |
| 47 | 47 |
|
| 48 |
public final static String NAME = "name"; |
|
| 48 |
public final static String NAME = "name"; //$NON-NLS-1$
|
|
| 49 | 49 |
|
| 50 |
public final static String TYPE_URL = "url"; |
|
| 50 |
public final static String TYPE_URL = "url"; //$NON-NLS-1$
|
|
| 51 | 51 |
|
| 52 |
public final static String TYPE_RESURL = "resurl"; |
|
| 52 |
public final static String TYPE_RESURL = "resurl"; //$NON-NLS-1$
|
|
| 53 | 53 |
|
| 54 |
public final static String TYPE_TYPEURL = "typeurl"; |
|
| 54 |
public final static String TYPE_TYPEURL = "typeurl"; //$NON-NLS-1$
|
|
| 55 | 55 |
|
| 56 |
public final static String TYPE_ID = "id"; |
|
| 56 |
public final static String TYPE_ID = "id"; //$NON-NLS-1$
|
|
| 57 | 57 |
|
| 58 |
public final static String TYPE_SIZE = "size"; |
|
| 58 |
public final static String TYPE_SIZE = "size"; //$NON-NLS-1$
|
|
| 59 | 59 |
|
| 60 |
public final static String TYPE_EFFECT = "effect"; |
|
| 60 |
public final static String TYPE_EFFECT = "effect"; //$NON-NLS-1$
|
|
| 61 | 61 |
|
| 62 |
public final static String TYPE_SQLFIELD_NAME = "sql_name"; |
|
| 62 |
public final static String TYPE_SQLFIELD_NAME = "sql_name"; //$NON-NLS-1$
|
|
| 63 | 63 |
|
| 64 |
public final static String TYPE_SQLFIELD_ID = "sql_id"; |
|
| 64 |
public final static String TYPE_SQLFIELD_ID = "sql_id"; //$NON-NLS-1$
|
|
| 65 | 65 |
|
| 66 |
public final static String TYPE_SQLFIELD_QUERY = "sql_query"; |
|
| 66 |
public final static String TYPE_SQLFIELD_QUERY = "sql_query"; //$NON-NLS-1$
|
|
| 67 | 67 |
|
| 68 |
public final static String LOGIN_KEY = "org.txm.repository.credencies.login."; |
|
| 68 |
public final static String LOGIN_KEY = "org.txm.repository.credencies.login."; //$NON-NLS-1$
|
|
| 69 | 69 |
|
| 70 |
public final static String PASSWORD_KEY = "org.txm.repository.credencies.password."; |
|
| 70 |
public final static String PASSWORD_KEY = "org.txm.repository.credencies.password."; //$NON-NLS-1$
|
|
| 71 | 71 |
|
| 72 | 72 |
public static final boolean[] FALSES = { false, false };
|
| 73 | 73 |
|
| ... | ... | |
| 85 | 85 |
* @return |
| 86 | 86 |
*/ |
| 87 | 87 |
public EntityManager initializeEntityManager() {
|
| 88 |
this.dbPath = new File(Toolbox.getTxmHomePath(), "repositories/" + name).getAbsolutePath(); |
|
| 88 |
this.dbPath = new File(Toolbox.getTxmHomePath(), "repositories/" + name).getAbsolutePath(); //$NON-NLS-1$
|
|
| 89 | 89 |
// System.out.println("KnowledgeRepository.initializeEntityManager [DB @ "+dbPath+"]");
|
| 90 | 90 |
|
| 91 | 91 |
HashMap<String, Object> properties = new HashMap<String, Object>(); |
| ... | ... | |
| 95 | 95 |
// |
| 96 | 96 |
// properties.put(PersistenceUnitProperties.ECLIPSELINK_PERSISTENCE_XML, "/home/mdecorde/workspace079/org.txm.annotation.core/META-INF"); |
| 97 | 97 |
// //properties.put("javax.persistence.jdbc.driver", "org.hsqldb.jdbcDriver");
|
| 98 |
String urlProperty = "jdbc:hsqldb:file:" + dbPath + "/db;shutdown=true;hsqldb.lock_file=false;"; |
|
| 98 |
String urlProperty = "jdbc:hsqldb:file:" + dbPath + "/db;shutdown=true;hsqldb.lock_file=false;"; //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 99 | 99 |
if (Log.getLevel().intValue() < Level.INFO.intValue()) {
|
| 100 |
urlProperty += ""; |
|
| 100 |
urlProperty += ""; //$NON-NLS-1$
|
|
| 101 | 101 |
} |
| 102 | 102 |
else {
|
| 103 |
urlProperty += "hsqldb.applog=0;hsqldb.sqllog=0"; |
|
| 103 |
urlProperty += "hsqldb.applog=0;hsqldb.sqllog=0"; //$NON-NLS-1$
|
|
| 104 | 104 |
} |
| 105 |
properties.put("javax.persistence.jdbc.url", urlProperty);
|
|
| 105 |
properties.put("javax.persistence.jdbc.url", urlProperty); //$NON-NLS-1$
|
|
| 106 | 106 |
// properties.put("javax.persistence.jdbc.username", "SA");
|
| 107 | 107 |
// properties.put(PersistenceUnitProperties.DDL_GENERATION_MODE, "database"); |
| 108 |
properties.put(PersistenceUnitProperties.DDL_GENERATION, "create-or-extend-tables"); // create&update table if needed |
|
| 108 |
properties.put(PersistenceUnitProperties.DDL_GENERATION, "create-or-extend-tables"); // create&update table if needed //$NON-NLS-1$
|
|
| 109 | 109 |
// properties.put(PersistenceUnitProperties.DDL_GENERATION, "drop-and-create-tables"); // drop all and recreate |
| 110 | 110 |
// if (Log.getLevel().intValue() < Level.WARNING.intValue()) {
|
| 111 | 111 |
// properties.put(PersistenceUnitProperties.LOGGING_LEVEL, "WARNING"); |
| ... | ... | |
| 119 | 119 |
PersistenceProvider pp = new PersistenceProvider(); |
| 120 | 120 |
emf = pp.createEntityManagerFactory(DatabasePersistenceManager.PERSISTENCE_UNIT_NAME, properties); |
| 121 | 121 |
if (emf == null) {
|
| 122 |
Log.warning(NLS.bind("Error: could not create the ''{0}'' EntityManagerFactory and PersistenceProvider={1}.", DatabasePersistenceManager.PERSISTENCE_UNIT_NAME, pp));
|
|
| 122 |
Log.warning(NLS.bind("Error: could not create the ''{0}'' EntityManagerFactory and PersistenceProvider={1}.", DatabasePersistenceManager.PERSISTENCE_UNIT_NAME, pp)); //$NON-NLS-1$
|
|
| 123 | 123 |
return null; |
| 124 | 124 |
} |
| 125 | 125 |
EntityManager entityManager = (EntityManager) emf.createEntityManager(); |
| ... | ... | |
| 132 | 132 |
public void setAccessProperties(HashMap<String, String> accessProperties) {
|
| 133 | 133 |
this.accessProperties = accessProperties; |
| 134 | 134 |
|
| 135 |
String sVersion = accessProperties.get("version");
|
|
| 135 |
String sVersion = accessProperties.get("version"); //$NON-NLS-1$
|
|
| 136 | 136 |
if (sVersion.length() > 0) {
|
| 137 | 137 |
this.version = Integer.parseInt(sVersion); |
| 138 | 138 |
} |
| ... | ... | |
| 218 | 218 |
|
| 219 | 219 |
public String toHumanString() {
|
| 220 | 220 |
StringBuffer buffer = new StringBuffer(); |
| 221 |
buffer.append(this.getName() + " version=" + version + " (" + this.getURL() + "): " + this.getAccessType() + "\n");
|
|
| 221 |
buffer.append(this.getName() + " version=" + version + " (" + this.getURL() + "): " + this.getAccessType() + "\n"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
|
|
| 222 | 222 |
for (AnnotationType type : getAllAnnotationTypes()) {
|
| 223 |
buffer.append("\t" + type.toHumanString() + "\n");
|
|
| 223 |
buffer.append("\t" + type.toHumanString() + "\n"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 224 | 224 |
} |
| 225 | 225 |
return buffer.toString(); |
| 226 | 226 |
} |
| 227 | 227 |
|
| 228 | 228 |
public void reload() {
|
| 229 |
Log.fine("Reloading KR " + name);
|
|
| 229 |
Log.fine("Reloading KR " + name); //$NON-NLS-1$
|
|
| 230 | 230 |
|
| 231 | 231 |
clearTypes(); |
| 232 | 232 |
buildTypesAndValues(); |
| ... | ... | |
| 240 | 240 |
*/ |
| 241 | 241 |
public List<TypedValue> getValues(AnnotationType type, int limit) throws Exception {
|
| 242 | 242 |
|
| 243 |
TypedQuery<TypedValue> query = jpaem.createQuery("SELECT t FROM TypedValue AS t "
|
|
| 244 |
+ "WHERE t.PK.typeId LIKE '" + type.getId() + "'", TypedValue.class); |
|
| 243 |
TypedQuery<TypedValue> query = jpaem.createQuery("SELECT t FROM TypedValue AS t " //$NON-NLS-1$
|
|
| 244 |
+ "WHERE t.PK.typeId LIKE '" + type.getId() + "'", TypedValue.class); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 245 | 245 |
if (limit != -1) {
|
| 246 | 246 |
query.setMaxResults(limit); |
| 247 | 247 |
} |
| ... | ... | |
| 256 | 256 |
* @throws Exception |
| 257 | 257 |
*/ |
| 258 | 258 |
public boolean hasValues(AnnotationType type) throws Exception {
|
| 259 |
TypedQuery<TypedValue> query = jpaem.createQuery("SELECT t.name FROM TypedValue AS t "
|
|
| 260 |
+ "WHERE t.PK.typeId LIKE '" + type.getId() + "'", TypedValue.class); |
|
| 259 |
TypedQuery<TypedValue> query = jpaem.createQuery("SELECT t.name FROM TypedValue AS t " //$NON-NLS-1$
|
|
| 260 |
+ "WHERE t.PK.typeId LIKE '" + type.getId() + "'", TypedValue.class); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 261 | 261 |
query.setMaxResults(1); |
| 262 | 262 |
|
| 263 | 263 |
return query.getResultList().size() > 0; |
| ... | ... | |
| 280 | 280 |
// clear all persisted values |
| 281 | 281 |
if (jpaem != null) {
|
| 282 | 282 |
jpaem.getTransaction().begin(); |
| 283 |
Query query = jpaem.createQuery("DELETE FROM TypedValue");
|
|
| 283 |
Query query = jpaem.createQuery("DELETE FROM TypedValue"); //$NON-NLS-1$
|
|
| 284 | 284 |
query.executeUpdate(); |
| 285 | 285 |
jpaem.getTransaction().commit(); |
| 286 | 286 |
} |
| ... | ... | |
| 329 | 329 |
String url = typeFieldsContent.get(KnowledgeRepository.TYPE_URL); |
| 330 | 330 |
String size = typeFieldsContent.get(KnowledgeRepository.TYPE_SIZE); |
| 331 | 331 |
String s_effect = typeFieldsContent.get(KnowledgeRepository.TYPE_EFFECT); |
| 332 |
if (s_effect == null) s_effect = "SEGMENT"; |
|
| 332 |
if (s_effect == null) s_effect = "SEGMENT"; //$NON-NLS-1$
|
|
| 333 | 333 |
|
| 334 | 334 |
for (String typeField : typeFieldsContent.keySet()) {
|
| 335 | 335 |
// System.out.println("Type field : "+typeField+" value : "+typeFieldsContent.get(typeField));
|
| ... | ... | |
| 360 | 360 |
* @return |
| 361 | 361 |
*/ |
| 362 | 362 |
public AnnotationType addType(String name, String id) {
|
| 363 |
return addType(name, id, this.url + "/" + id); |
|
| 363 |
return addType(name, id, this.url + "/" + id); //$NON-NLS-1$
|
|
| 364 | 364 |
} |
| 365 | 365 |
|
| 366 | 366 |
/** |
| ... | ... | |
| 403 | 403 |
} |
| 404 | 404 |
|
| 405 | 405 |
public List<AnnotationType> getAllAnnotationTypes() {
|
| 406 |
TypedQuery<AnnotationType> query = jpaem.createQuery("SELECT t FROM AnnotationType AS t ", AnnotationType.class);
|
|
| 406 |
TypedQuery<AnnotationType> query = jpaem.createQuery("SELECT t FROM AnnotationType AS t ", AnnotationType.class); //$NON-NLS-1$
|
|
| 407 | 407 |
return query.getResultList(); |
| 408 | 408 |
} |
| 409 | 409 |
|
| 410 | 410 |
public List<TypedValue> getAllTypedValuesForType(String idType) {
|
| 411 |
TypedQuery<TypedValue> query = jpaem.createQuery("SELECT t FROM TypedValue AS t WHERE t.PK.typeId LIKE '" + idType + "'", TypedValue.class);
|
|
| 411 |
TypedQuery<TypedValue> query = jpaem.createQuery("SELECT t FROM TypedValue AS t WHERE t.PK.typeId LIKE '" + idType + "'", TypedValue.class); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 412 | 412 |
return query.getResultList(); |
| 413 | 413 |
} |
| 414 | 414 |
|
| ... | ... | |
| 419 | 419 |
* @throws Exception |
| 420 | 420 |
*/ |
| 421 | 421 |
public List<TypedValue> findTypedValuesWithPrefix(AnnotationType type, String prefix) throws Exception {
|
| 422 |
TypedQuery<TypedValue> query = jpaem.createQuery("SELECT t FROM TypedValue AS t "
|
|
| 423 |
+ "WHERE t.PK.typeId LIKE '" + type.getId() + "'" |
|
| 424 |
+ " AND t.name LIKE '" + prefix + "%'", TypedValue.class); |
|
| 422 |
TypedQuery<TypedValue> query = jpaem.createQuery("SELECT t FROM TypedValue AS t " //$NON-NLS-1$
|
|
| 423 |
+ "WHERE t.PK.typeId LIKE '" + type.getId() + "'" //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 424 |
+ " AND t.name LIKE '" + prefix + "%'", TypedValue.class); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 425 | 425 |
// System.out.println("Query for "+prefix+" has nb results = "+query.getResultList().size());
|
| 426 | 426 |
return query.getResultList(); |
| 427 | 427 |
} |
| ... | ... | |
| 478 | 478 |
values = strings.get(lang); |
| 479 | 479 |
} |
| 480 | 480 |
if (values == null) { // get default lang = "en"
|
| 481 |
values = strings.get("en");
|
|
| 481 |
values = strings.get("en"); //$NON-NLS-1$
|
|
| 482 | 482 |
} |
| 483 | 483 |
if (values == null) { // fail
|
| 484 | 484 |
return null; |
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/temporary/DeleteAnnotation.java (revision 3883) | ||
|---|---|---|
| 8 | 8 |
|
| 9 | 9 |
@Test |
| 10 | 10 |
public void test() {
|
| 11 |
fail("Not yet implemented");
|
|
| 11 |
fail("Not yet implemented"); //$NON-NLS-1$
|
|
| 12 | 12 |
} |
| 13 | 13 |
|
| 14 | 14 |
} |
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/temporary/UpdateAnnotation.java (revision 3883) | ||
|---|---|---|
| 8 | 8 |
|
| 9 | 9 |
@Test |
| 10 | 10 |
public void test() {
|
| 11 |
fail("Not yet implemented");
|
|
| 11 |
fail("Not yet implemented"); //$NON-NLS-1$
|
|
| 12 | 12 |
} |
| 13 | 13 |
|
| 14 | 14 |
} |
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/temporary/CreateAnnotation.java (revision 3883) | ||
|---|---|---|
| 16 | 16 |
|
| 17 | 17 |
@Test |
| 18 | 18 |
public void test() throws CqiClientException, InvalidCqpIdException {
|
| 19 |
if (!Toolbox.isInitialized()) fail("Toolbox not initialized.");
|
|
| 20 |
MainCorpus corpus = CorpusManager.getCorpusManager().getCorpus("VOEUX");
|
|
| 21 |
if (corpus == null) fail("Corpus Voeux not loaded.");
|
|
| 19 |
if (!Toolbox.isInitialized()) fail("Toolbox not initialized."); //$NON-NLS-1$
|
|
| 20 |
MainCorpus corpus = CorpusManager.getCorpusManager().getCorpus("VOEUX"); //$NON-NLS-1$
|
|
| 21 |
if (corpus == null) fail("Corpus Voeux not loaded."); //$NON-NLS-1$
|
|
| 22 | 22 |
|
| 23 | 23 |
HashMap<String, Object> properties = TemporaryAnnotationManager.getInitialisationProperties(this.getClass(), corpus); |
| 24 |
properties.put("eclipselink.persistencexml", System.getProperty("user.home") + "/workspace442/org.txm.core/META-INF/persistence.xml");
|
|
| 24 |
properties.put("eclipselink.persistencexml", System.getProperty("user.home") + "/workspace442/org.txm.core/META-INF/persistence.xml"); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
|
| 25 | 25 |
|
| 26 | 26 |
TemporaryAnnotationManager tam = new TemporaryAnnotationManager(corpus, properties); |
| 27 | 27 |
System.out.println(tam); |
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/conversion/XMLTXMFileRuledConversion.java (revision 3883) | ||
|---|---|---|
| 26 | 26 |
|
| 27 | 27 |
protected String newType; |
| 28 | 28 |
|
| 29 |
public static final String DELETE = "supprimer"; |
|
| 29 |
public static final String DELETE = "supprimer"; //$NON-NLS-1$
|
|
| 30 | 30 |
|
| 31 |
public static final String COPY = "copier"; |
|
| 31 |
public static final String COPY = "copier"; //$NON-NLS-1$
|
|
| 32 | 32 |
|
| 33 |
public static final String ABANDON = "abandon"; |
|
| 33 |
public static final String ABANDON = "abandon"; //$NON-NLS-1$
|
|
| 34 | 34 |
|
| 35 | 35 |
HashSet<String> noMatchValues = new HashSet<>(); |
| 36 | 36 |
|
| ... | ... | |
| 58 | 58 |
} |
| 59 | 59 |
|
| 60 | 60 |
if (elementPath == null) { // where fixing txm:words
|
| 61 |
if (!this.newType.startsWith("#")) this.newType = "#" + this.newType;
|
|
| 62 |
if (!this.oldType.startsWith("#")) this.oldType = "#" + this.oldType;
|
|
| 61 |
if (!this.newType.startsWith("#")) this.newType = "#" + this.newType; //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 62 |
if (!this.oldType.startsWith("#")) this.oldType = "#" + this.oldType; //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 63 | 63 |
} |
| 64 | 64 |
} |
| 65 | 65 |
|
| ... | ... | |
| 99 | 99 |
String value = parser.getAttributeValue(i); |
| 100 | 100 |
if (oldType != null && oldType.equals(att)) {
|
| 101 | 101 |
if (converter != null) {
|
| 102 |
if (debug) Log.finer("CALL CONVERTER with attribute "+att);
|
|
| 102 |
if (debug) Log.finer("CALL CONVERTER with attribute "+att); //$NON-NLS-1$
|
|
| 103 | 103 |
newValue = converter.getValue(parser, localname, att, value); |
| 104 | 104 |
} |
| 105 | 105 |
else {
|
| ... | ... | |
| 112 | 112 |
} |
| 113 | 113 |
|
| 114 | 114 |
if (oldType == null && converter != null) {
|
| 115 |
if (debug) Log.finer("CALL CONVERTER with no attribute set");
|
|
| 115 |
if (debug) Log.finer("CALL CONVERTER with no attribute set"); //$NON-NLS-1$
|
|
| 116 | 116 |
newValue = converter.getValue(parser, localname, null, null); |
| 117 | 117 |
} |
| 118 | 118 |
|
| 119 | 119 |
if (newValue != null) {
|
| 120 |
if (debug) Log.finer("WRITE NEWTYPE: " + newType + "=" + newValue);
|
|
| 120 |
if (debug) Log.finer("WRITE NEWTYPE: " + newType + "=" + newValue); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 121 | 121 |
writeAttribute(null, newType, newValue); |
| 122 | 122 |
} |
| 123 | 123 |
} |
| ... | ... | |
| 142 | 142 |
|
| 143 | 143 |
if (!inW) super.processStartElement(); // don't write W content |
| 144 | 144 |
|
| 145 |
if (localname.equals("w")) {
|
|
| 145 |
if (localname.equals("w")) { //$NON-NLS-1$
|
|
| 146 | 146 |
inW = true; |
| 147 | 147 |
anaValues.clear(); |
| 148 | 148 |
anaResps.clear(); |
| 149 | 149 |
|
| 150 | 150 |
// initialize the new type to a empty value in case there is transformation rule |
| 151 |
anaValues.put(newType, ""); |
|
| 152 |
anaResps.put(newType, "#txm_recode"); |
|
| 151 |
anaValues.put(newType, ""); //$NON-NLS-1$
|
|
| 152 |
anaResps.put(newType, "#txm_recode"); //$NON-NLS-1$
|
|
| 153 | 153 |
} |
| 154 |
else if (localname.equals("ana")) {
|
|
| 154 |
else if (localname.equals("ana")) { //$NON-NLS-1$
|
|
| 155 | 155 |
inAna = true; |
| 156 |
typeName = parser.getAttributeValue(null, "type"); |
|
| 157 |
respName = parser.getAttributeValue(null, "resp"); |
|
| 156 |
typeName = parser.getAttributeValue(null, "type"); //$NON-NLS-1$
|
|
| 157 |
respName = parser.getAttributeValue(null, "resp"); //$NON-NLS-1$
|
|
| 158 | 158 |
anaResps.put(typeName, respName); |
| 159 | 159 |
// if (typeName != null) typeName = typeName.substring(1); // remove # |
| 160 |
typeValue = ""; |
|
| 160 |
typeValue = ""; //$NON-NLS-1$
|
|
| 161 | 161 |
} |
| 162 |
else if (localname.equals("form")) {
|
|
| 162 |
else if (localname.equals("form")) { //$NON-NLS-1$
|
|
| 163 | 163 |
inForm = true; |
| 164 |
formValue = ""; |
|
| 164 |
formValue = ""; //$NON-NLS-1$
|
|
| 165 | 165 |
} |
| 166 | 166 |
} |
| 167 | 167 |
} |
| ... | ... | |
| 186 | 186 |
} |
| 187 | 187 |
else {
|
| 188 | 188 |
|
| 189 |
if (localname.equals("w")) {
|
|
| 189 |
if (localname.equals("w")) { //$NON-NLS-1$
|
|
| 190 | 190 |
inW = false; |
| 191 | 191 |
|
| 192 | 192 |
// write W content |
| 193 | 193 |
try {
|
| 194 | 194 |
// get the value to test |
| 195 | 195 |
String value = null; |
| 196 |
if (oldType.equals("word")) {
|
|
| 196 |
if (oldType.equals("word")) { //$NON-NLS-1$
|
|
| 197 | 197 |
value = formValue; |
| 198 | 198 |
} |
| 199 | 199 |
else {
|
| 200 | 200 |
value = anaValues.get(oldType); |
| 201 | 201 |
} |
| 202 | 202 |
|
| 203 |
if (newType.equals("word")) { // update form property
|
|
| 203 |
if (newType.equals("word")) { // update form property //$NON-NLS-1$
|
|
| 204 | 204 |
updateFormValueIfMatch(value); |
| 205 | 205 |
} |
| 206 | 206 |
else { // update a ana property
|
| ... | ... | |
| 210 | 210 |
} |
| 211 | 211 |
|
| 212 | 212 |
// write the word element |
| 213 |
writer.writeStartElement("txm:form");
|
|
| 213 |
writer.writeStartElement("txm:form"); //$NON-NLS-1$
|
|
| 214 | 214 |
writer.writeCharacters(formValue); |
| 215 | 215 |
writer.writeEndElement(); |
| 216 | 216 |
|
| 217 | 217 |
for (String k : anaValues.keySet()) {
|
| 218 | 218 |
String resp = anaResps.get(k); |
| 219 |
if (resp == null) resp = "#txm_recode"; |
|
| 219 |
if (resp == null) resp = "#txm_recode"; //$NON-NLS-1$
|
|
| 220 | 220 |
|
| 221 |
writer.writeStartElement("txm:ana");
|
|
| 222 |
writer.writeAttribute("resp", resp);
|
|
| 223 |
writer.writeAttribute("type", k);
|
|
| 221 |
writer.writeStartElement("txm:ana"); //$NON-NLS-1$
|
|
| 222 |
writer.writeAttribute("resp", resp); //$NON-NLS-1$
|
|
| 223 |
writer.writeAttribute("type", k); //$NON-NLS-1$
|
|
| 224 | 224 |
writer.writeCharacters(anaValues.get(k)); |
| 225 | 225 |
writer.writeEndElement(); |
| 226 | 226 |
} |
| ... | ... | |
| 229 | 229 |
e.printStackTrace(); |
| 230 | 230 |
} |
| 231 | 231 |
} |
| 232 |
else if (localname.equals("ana")) {
|
|
| 232 |
else if (localname.equals("ana")) { //$NON-NLS-1$
|
|
| 233 | 233 |
anaValues.put(typeName, typeValue); |
| 234 | 234 |
inAna = false; |
| 235 | 235 |
} |
| 236 |
else if (localname.equals("form")) {
|
|
| 236 |
else if (localname.equals("form")) { //$NON-NLS-1$
|
|
| 237 | 237 |
inForm = false; |
| 238 | 238 |
} |
| 239 | 239 |
|
| ... | ... | |
| 279 | 279 |
String v = converter.getValue(parser, localname, null, value); |
| 280 | 280 |
if (v != null) {
|
| 281 | 281 |
anaValues.put(newType, v); // do a replace if newType == oldType :-) |
| 282 |
anaResps.put(newType, "#txm_recode"); |
|
| 282 |
anaResps.put(newType, "#txm_recode"); //$NON-NLS-1$
|
|
| 283 | 283 |
return; // ok stop |
| 284 | 284 |
} |
| 285 | 285 |
} |
| ... | ... | |
| 289 | 289 |
if (rule.matcher(value).matches()) {
|
| 290 | 290 |
value = rules.get(rule); |
| 291 | 291 |
anaValues.put(newType, value); // do a replace if newType == oldType :-) |
| 292 |
anaResps.put(newType, "#txm_recode"); |
|
| 292 |
anaResps.put(newType, "#txm_recode"); //$NON-NLS-1$
|
|
| 293 | 293 |
return; |
| 294 | 294 |
} |
| 295 | 295 |
} |
| 296 | 296 |
} |
| 297 |
if (DELETE.equals(mode)) anaValues.put(newType, ""); // do a replace if newType == oldType :-) |
|
| 298 |
else if (ABANDON.equals(mode)) anaValues.put(newType, "ERROR(" + value + ")"); // do a replace if newType == oldType :-)
|
|
| 297 |
if (DELETE.equals(mode)) anaValues.put(newType, ""); // do a replace if newType == oldType :-) //$NON-NLS-1$
|
|
| 298 |
else if (ABANDON.equals(mode)) anaValues.put(newType, "ERROR(" + value + ")"); // do a replace if newType == oldType :-) //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 299 | 299 |
|
| 300 | 300 |
noMatchValues.add(value); |
| 301 | 301 |
} |
| ... | ... | |
| 306 | 306 |
|
| 307 | 307 |
public static void main(String args[]) {
|
| 308 | 308 |
try {
|
| 309 |
File xmlFile = new File("/home/mdecorde/xml/recode/test.xml");
|
|
| 310 |
File tmpFile = new File("/home/mdecorde/xml/recode/test-o.xml");
|
|
| 311 |
String oldType = "myattr"; |
|
| 312 |
String newType = "myattr"; |
|
| 309 |
File xmlFile = new File("/home/mdecorde/xml/recode/test.xml"); //$NON-NLS-1$
|
|
| 310 |
File tmpFile = new File("/home/mdecorde/xml/recode/test-o.xml"); //$NON-NLS-1$
|
|
| 311 |
String oldType = "myattr"; //$NON-NLS-1$
|
|
| 312 |
String newType = "myattr"; //$NON-NLS-1$
|
|
| 313 | 313 |
LinkedHashMap<Pattern, String> rules = new LinkedHashMap<>(); |
| 314 |
rules.put(Pattern.compile(".+"), "NEWW");
|
|
| 315 |
rules.put(Pattern.compile("x.+"), "XWORD");
|
|
| 316 |
rules.put(Pattern.compile("y"), "YWORD");
|
|
| 317 |
rules.put(Pattern.compile("y.*"), "YMULTIWORD");
|
|
| 318 |
XMLTXMFileRuledConversion converter = new XMLTXMFileRuledConversion(xmlFile, rules, "w", oldType, newType, ABANDON); |
|
| 314 |
rules.put(Pattern.compile(".+"), "NEWW"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 315 |
rules.put(Pattern.compile("x.+"), "XWORD"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 316 |
rules.put(Pattern.compile("y"), "YWORD"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 317 |
rules.put(Pattern.compile("y.*"), "YMULTIWORD"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 318 |
XMLTXMFileRuledConversion converter = new XMLTXMFileRuledConversion(xmlFile, rules, "w", oldType, newType, ABANDON); //$NON-NLS-1$
|
|
| 319 | 319 |
converter.converter = new Converter() {
|
| 320 | 320 |
|
| 321 | 321 |
@Override |
| 322 | 322 |
public String getValue(XMLStreamReader parser, String localname, String attribute, String value) {
|
| 323 |
return "NEW" + value; |
|
| 323 |
return "NEW" + value; //$NON-NLS-1$
|
|
| 324 | 324 |
} |
| 325 | 325 |
}; |
| 326 | 326 |
System.out.println(converter.process(tmpFile)); |
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/AnnotationManager.java (revision 3883) | ||
|---|---|---|
| 68 | 68 |
|
| 69 | 69 |
List<Annotation> annots = tempManager.getAnnotations(); |
| 70 | 70 |
if (annots.isEmpty()) {
|
| 71 |
Log.info("No annotation to save. Aborting.");
|
|
| 71 |
Log.info(Messages.noAnnotationToSaveAborting);
|
|
| 72 | 72 |
dirty = false; |
| 73 | 73 |
return true; |
| 74 | 74 |
} |
| 75 | 75 |
|
| 76 | 76 |
if (monitor != null) {
|
| 77 | 77 |
monitor.beginTask(Messages.savingAnnotations, annots.size()); |
| 78 |
monitor.setTaskName("Writing annotations in XML-TXM files");
|
|
| 78 |
monitor.setTaskName(Messages.writingAnnotationsInXMLTXMFiles);
|
|
| 79 | 79 |
} |
| 80 | 80 |
|
| 81 | 81 |
AnnotationWriter writer = new AnnotationWriter(corpus); |
| ... | ... | |
| 116 | 116 |
tempManager.getEntityManager().getTransaction().begin(); |
| 117 | 117 |
for (Annotation a : temporaryAnnotations) {
|
| 118 | 118 |
if (job != null && job.isCanceled()) {
|
| 119 |
Log.info("Delete annotation canceled.");
|
|
| 119 |
Log.info(Messages.deleteAnnotationCanceled);
|
|
| 120 | 120 |
return false; |
| 121 | 121 |
} |
| 122 | 122 |
tempManager.deleteAnnotation(type, a.getStart(), a.getEnd()); |
| ... | ... | |
| 127 | 127 |
tempManager.getEntityManager().getTransaction().begin(); |
| 128 | 128 |
for (Annotation a : cqpAnnotations) {
|
| 129 | 129 |
if (job != null && job.isCanceled()) {
|
| 130 |
Log.info("Delete annotation canceled.");
|
|
| 130 |
Log.info(Messages.deleteAnnotationCanceled);
|
|
| 131 | 131 |
return false; |
| 132 | 132 |
} |
| 133 | 133 |
String value = cqpManager.getCQPAnnotationValue(a.getStart(), a.getEnd(), type); |
| 134 | 134 |
if (value != null) {
|
| 135 |
tempManager.createAnnotationNoCommit(type, new TypedValue("#del", "#del", type.getId()), a.getStart(), a.getEnd());
|
|
| 135 |
tempManager.createAnnotationNoCommit(type, new TypedValue("#del", "#del", type.getId()), a.getStart(), a.getEnd()); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 136 | 136 |
} |
| 137 | 137 |
else {
|
| 138 | 138 |
tempManager.deleteAnnotationNoCommit(type, a.getStart(), a.getEnd()); |
| ... | ... | |
| 166 | 166 |
} |
| 167 | 167 |
|
| 168 | 168 |
if (job != null && job.isCanceled()) {
|
| 169 |
Log.info("Delete annotation canceled.");
|
|
| 169 |
Log.info(Messages.deleteAnnotationCanceled);
|
|
| 170 | 170 |
return false; |
| 171 | 171 |
} |
| 172 | 172 |
|
| 173 | 173 |
String value = cqpManager.getCQPAnnotationValue(start, end, type); |
| 174 | 174 |
|
| 175 | 175 |
if (value != null) {
|
| 176 |
tempManager.createAnnotationNoCommit(type, new TypedValue("#del", "#del", type.getId()), start, end);
|
|
| 176 |
tempManager.createAnnotationNoCommit(type, new TypedValue("#del", "#del", type.getId()), start, end); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 177 | 177 |
} |
| 178 | 178 |
else {
|
| 179 | 179 |
tempManager.deleteAnnotationNoCommit(type, start, end); |
| ... | ... | |
| 206 | 206 |
// System.out.println("Temporary annotations: "+temporaryAnnotations);
|
| 207 | 207 |
// System.out.println("CQP annotations: "+cqpAnnotations);
|
| 208 | 208 |
if (cqpAnnotations.size() != matches.size() || temporaryAnnotations.size() != matches.size()) {
|
| 209 |
Log.info("ERROR in getAnnotationsForMatches methods! ");
|
|
| 209 |
Log.info("ERROR in getAnnotationsForMatches methods! "); //$NON-NLS-1$
|
|
| 210 | 210 |
return new ArrayList<>(matches.size()); |
| 211 | 211 |
} |
| 212 | 212 |
// merge the 2 results |
| ... | ... | |
| 237 | 237 |
tempManager.close(); |
| 238 | 238 |
} |
| 239 | 239 |
catch (Exception e) {
|
| 240 |
Log.warning(NLS.bind("Fail to clear AnnotationManager instance with error: {0}", e));
|
|
| 240 |
Log.warning(NLS.bind(Messages.failToClearTheAnnotationManagerInstanceP0, e));
|
|
| 241 | 241 |
Log.printStackTrace(e); |
| 242 | 242 |
} |
| 243 | 243 |
} |
| ... | ... | |
| 286 | 286 |
} |
| 287 | 287 |
|
| 288 | 288 |
if (job != null && job.isCanceled()) { // check if user canceled the job
|
| 289 |
Log.info("Affect annotation canceled.");
|
|
| 289 |
Log.info(Messages.affectAnnotationCanceled);
|
|
| 290 | 290 |
return null; |
| 291 | 291 |
} |
| 292 | 292 |
|
| ... | ... | |
| 345 | 345 |
|
| 346 | 346 |
} |
| 347 | 347 |
catch (Throwable e) {
|
| 348 |
Log.info(NLS.bind("Error during annotation creation: {0}.", e));
|
|
| 348 |
Log.info(NLS.bind(Messages.errorDuringAnnotationCreationP0, e));
|
|
| 349 | 349 |
Log.printStackTrace(e); |
| 350 | 350 |
|
| 351 | 351 |
if (tempManager.getEntityManager().getTransaction().isActive()) {
|
| ... | ... | |
| 390 | 390 |
} |
| 391 | 391 |
|
| 392 | 392 |
public void closeAll() {
|
| 393 |
Log.fine("Closing annotation manager of " + corpus);
|
|
| 393 |
Log.fine("Closing annotation manager of " + corpus); //$NON-NLS-1$
|
|
| 394 | 394 |
tempManager.close(); |
| 395 | 395 |
cqpManager.close(); |
| 396 | 396 |
} |
| ... | ... | |
| 423 | 423 |
|
| 424 | 424 |
List<String> krnames = KRAnnotationEngine.getKnowledgeRepositoryNames(corpus); |
| 425 | 425 |
if (krnames.size() == 0) {
|
| 426 |
Log.severe(NLS.bind("** Error: no knowledge repository found in {0} corpus.", corpus));
|
|
| 427 |
throw new IllegalArgumentException("No kr in " + corpus);
|
|
| 426 |
Log.severe(NLS.bind(Messages.errorNoKRFoundInP0Corpus, corpus));
|
|
| 427 |
throw new IllegalArgumentException("No kr in " + corpus); //$NON-NLS-1$
|
|
| 428 | 428 |
} |
| 429 | 429 |
String firtsKRName = krnames.get(0); |
| 430 | 430 |
KnowledgeRepository defaultKR = KRAnnotationEngine.getKnowledgeRepository(corpus, firtsKRName); |
| 431 | 431 |
if (defaultKR == null) {
|
| 432 |
Log.severe(NLS.bind("** Error: no knowledge repository {0} found in {0} corpus.", defaultKR, corpus));
|
|
| 433 |
throw new IllegalArgumentException("No kr " + defaultKR + " in " + corpus);
|
|
| 432 |
Log.severe(NLS.bind(Messages.errorNoKRP0FoundInP1Corpus, defaultKR, corpus));
|
|
| 433 |
throw new IllegalArgumentException("No kr " + defaultKR + " in " + corpus); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 434 | 434 |
} |
| 435 | 435 |
|
| 436 | 436 |
PrintWriter writer = IOUtils.getWriter(resultFile, encoding, false); |
| 437 |
String cols [] = {"text_id", "id", "end_id", "start", "end", "type", "value", "mode", "date", "annotator"};
|
|
| 437 |
String cols [] = {"text_id", "id", "end_id", "start", "end", "type", "value", "mode", "date", "annotator"}; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$ //$NON-NLS-5$ //$NON-NLS-6$ //$NON-NLS-7$ //$NON-NLS-8$ //$NON-NLS-9$ //$NON-NLS-10$
|
|
| 438 | 438 |
writer.println(StringUtils.join(cols, colseparator)); |
| 439 | 439 |
|
| 440 | 440 |
if (this.getTemporaryManager().getAnnotations().size() == 0) {
|
| 441 |
Log.warning("Warning: no annotations to export. Aborting.");
|
|
| 441 |
Log.warning(Messages.warningNoAnnotationsToExportAborting);
|
|
| 442 | 442 |
return false; |
| 443 | 443 |
} |
| 444 | 444 |
|
| ... | ... | |
| 456 | 456 |
for (int p : positions2Ids.keySet()) {
|
| 457 | 457 |
positions[i++] = p; |
| 458 | 458 |
} |
| 459 |
String values[] = corpus.getProperty("id").cpos2Str(positions);
|
|
| 459 |
String values[] = corpus.getProperty("id").cpos2Str(positions); //$NON-NLS-1$
|
|
| 460 | 460 |
for (i = 0 ; i < positions.length ; i++) {
|
| 461 | 461 |
positions2Ids.put(positions[i], values[i]); |
| 462 | 462 |
} |
| ... | ... | |
| 469 | 469 |
} |
| 470 | 470 |
}); |
| 471 | 471 |
|
| 472 |
Log.info(NLS.bind("Exporting {0} annotations.", annotations.size()));
|
|
| 472 |
Log.info(NLS.bind(Messages.exportingP0Annotations, annotations.size()));
|
|
| 473 | 473 |
ConsoleProgressBar cpb = new ConsoleProgressBar(annotations.size()); |
| 474 | 474 |
|
| 475 | 475 |
int currentText = 0; |
| ... | ... | |
| 515 | 515 |
|
| 516 | 516 |
List<String> krnames = KRAnnotationEngine.getKnowledgeRepositoryNames(corpus); |
| 517 | 517 |
if (krnames.size() == 0) {
|
| 518 |
Log.severe(NLS.bind("** Error: no knowledge repository found in {0} corpus.", corpus));
|
|
| 519 |
throw new IllegalArgumentException("No kr in " + corpus);
|
|
| 518 |
Log.severe(NLS.bind(Messages.errorNoKRFoundInP0Corpus, corpus));
|
|
| 519 |
throw new IllegalArgumentException("No kr in " + corpus); //$NON-NLS-1$
|
|
| 520 | 520 |
} |
| 521 | 521 |
String firtsKRName = krnames.get(0); |
| 522 | 522 |
KnowledgeRepository defaultKR = KRAnnotationEngine.getKnowledgeRepository(corpus, firtsKRName); |
| 523 | 523 |
if (defaultKR == null) {
|
| 524 |
Log.severe(NLS.bind("** Error: no knowledge repository {0} found in {0} corpus.", defaultKR, corpus));
|
|
| 525 |
throw new IllegalArgumentException("No kr " + defaultKR + " in " + corpus);
|
|
| 524 |
Log.severe(NLS.bind(Messages.errorNoKRP0FoundInP1Corpus, defaultKR, corpus));
|
|
| 525 |
throw new IllegalArgumentException("No kr " + defaultKR + " in " + corpus); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 526 | 526 |
} |
| 527 | 527 |
|
| 528 | 528 |
CsvReader reader = new CsvReader(annotationsFile.getAbsolutePath(), colseparator.charAt(0), Charset.forName(encoding)); |
| 529 | 529 |
if (txtseparator.length() > 0) reader.setTextQualifier(txtseparator.charAt(0)); |
| 530 | 530 |
reader.readHeaders(); |
| 531 | 531 |
String headers [] = reader.getHeaders(); |
| 532 |
String cols [] = {"text_id", "id", "end_id", "start", "end", "type", "value", "mode", "date", "annotator"};
|
|
| 532 |
String cols [] = {"text_id", "id", "end_id", "start", "end", "type", "value", "mode", "date", "annotator"}; //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$ //$NON-NLS-5$ //$NON-NLS-6$ //$NON-NLS-7$ //$NON-NLS-8$ //$NON-NLS-9$ //$NON-NLS-10$
|
|
| 533 | 533 |
if (headers.length != cols.length) {
|
| 534 |
Log.warning("Error: annotation file header format is not well-formatted: "+StringUtils.join(headers, ", ")+". Should be: "+StringUtils.join(cols, ", "));
|
|
| 534 |
Log.warning(Messages.errorAnnotationFileHeaderFormatIsNotWekkfomattedP0ItShouldBeP1+StringUtils.join(headers, ", ")+". Should be: "+StringUtils.join(cols, ", ")); //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
|
|
| 535 | 535 |
return false; |
| 536 | 536 |
} |
| 537 | 537 |
for (int i = 0 ; i < cols.length ; i++) {
|
| 538 | 538 |
if (!(headers[i].equals(cols[i]))) {
|
| 539 |
Log.warning("Error: annotation file header format is not well-formatted: "+StringUtils.join(headers, ", ")+". Should be: "+StringUtils.join(cols, ", "));
|
|
| 539 |
Log.warning(Messages.errorAnnotationFileHeaderFormatIsNotWekkfomattedP0ItShouldBeP1+StringUtils.join(headers, ", ")+". Should be: "+StringUtils.join(cols, ", ")); //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
|
|
| 540 | 540 |
return false; |
| 541 | 541 |
} |
| 542 | 542 |
} |
| ... | ... | |
| 548 | 548 |
reader.close(); |
| 549 | 549 |
|
| 550 | 550 |
if (nLines == 0) {
|
| 551 |
Log.warning("Error: no annotation line found");
|
|
| 551 |
Log.warning(Messages.errorNoAnnotationLineFound);
|
|
| 552 | 552 |
return false; |
| 553 | 553 |
} |
| 554 | 554 |
|
| 555 |
Log.info(NLS.bind("Importing {0} annotations...", nLines));
|
|
| 555 |
Log.info(NLS.bind(Messages.importingP0Annotations, nLines));
|
|
| 556 | 556 |
reader = new CsvReader(annotationsFile.getAbsolutePath(), colseparator.charAt(0), Charset.forName(encoding)); |
| 557 | 557 |
if (txtseparator.length() > 0) reader.setTextQualifier(txtseparator.charAt(0)); |
| 558 | 558 |
reader.readHeaders(); |
| ... | ... | |
| 574 | 574 |
if (startId.length() > 0 && endId.length() > 0) { // use the word id if present
|
| 575 | 575 |
String textId = record[0].trim(); |
| 576 | 576 |
if (startId.equals(endId)) {
|
| 577 |
CQLQuery cql = new CQLQuery(NLS.bind("[id=\"{0}\" & _.text_id=\"{1}\"]", startId, textId));
|
|
| 578 |
QueryResult qr = corpus.query(cql , "TMP", false); |
|
| 577 |
CQLQuery cql = new CQLQuery(NLS.bind("[id=\"{0}\" & _.text_id=\"{1}\"]", startId, textId)); //$NON-NLS-1$
|
|
| 578 |
QueryResult qr = corpus.query(cql , "TMP", false); //$NON-NLS-1$
|
|
| 579 | 579 |
if (qr.getNMatch() == 0) {
|
| 580 |
Log.warning("No match found with CQL="+cql);
|
|
| 580 |
Log.warning(Messages.noMatchFoundWithCQLEqualP0+cql);
|
|
| 581 | 581 |
continue; |
| 582 | 582 |
} |
| 583 | 583 |
startPos = qr.getMatch(0).getStart(); |
| 584 | 584 |
endPos = startPos; |
| 585 | 585 |
} else {
|
| 586 |
CQLQuery cql = new CQLQuery(NLS.bind("[id=\"{0}\" & _.text_id=\"{1}\"]", startId, textId));
|
|
| 587 |
QueryResult qr = corpus.query(cql , "TMP", false); |
|
| 586 |
CQLQuery cql = new CQLQuery(NLS.bind("[id=\"{0}\" & _.text_id=\"{1}\"]", startId, textId)); //$NON-NLS-1$
|
|
| 587 |
QueryResult qr = corpus.query(cql , "TMP", false); //$NON-NLS-1$
|
|
| 588 | 588 |
if (qr.getNMatch() == 0) {
|
| 589 |
Log.warning("No match found with CQL="+cql);
|
|
| 589 |
Log.warning(Messages.noMatchFoundWithCQLEqualP0+cql);
|
|
| 590 | 590 |
continue; |
| 591 | 591 |
} |
| 592 | 592 |
startPos = qr.getMatch(0).getStart(); |
| 593 | 593 |
qr.drop(); |
| 594 |
cql = new CQLQuery(NLS.bind("[id=\"{0}\" & _.text_id=\"{1}\"]", endId, textId));
|
|
| 595 |
qr = corpus.query(cql , "TMP", false); |
|
| 594 |
cql = new CQLQuery(NLS.bind("[id=\"{0}\" & _.text_id=\"{1}\"]", endId, textId)); //$NON-NLS-1$
|
|
| 595 |
qr = corpus.query(cql , "TMP", false); //$NON-NLS-1$
|
|
| 596 | 596 |
if (qr.getNMatch() == 0) {
|
| 597 |
Log.warning("No match found with CQL="+cql);
|
|
| 597 |
Log.warning(Messages.noMatchFoundWithCQLEqualP0+cql);
|
|
| 598 | 598 |
continue; |
| 599 | 599 |
} |
| 600 | 600 |
endPos = qr.getMatch(0).getStart(); |
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/AnnotationWriter.java (revision 3883) | ||
|---|---|---|
| 45 | 45 |
this.corpus = corpus; |
| 46 | 46 |
List<String> krnames = KRAnnotationEngine.getKnowledgeRepositoryNames(corpus); |
| 47 | 47 |
if (krnames.size() == 0) {
|
| 48 |
Log.severe(NLS.bind("** Error: no knowledge repository found in {0} corpus.", corpus));
|
|
| 49 |
throw new IllegalArgumentException("No kr in " + corpus);
|
|
| 48 |
Log.severe(NLS.bind(Messages.ErrorNoKnowledgeRepositoryFoundInP0Corpus, corpus));
|
|
| 49 |
throw new IllegalArgumentException("No kr in " + corpus); //$NON-NLS-1$
|
|
| 50 | 50 |
} |
| 51 | 51 |
String firtsKRName = krnames.get(0); |
| 52 | 52 |
defaultKR = KRAnnotationEngine.getKnowledgeRepository(corpus, firtsKRName); |
| 53 | 53 |
if (defaultKR == null) {
|
| 54 |
Log.severe(NLS.bind("** Error: no knowledge repository {0} found in {0} corpus.", defaultKR, corpus));
|
|
| 55 |
throw new IllegalArgumentException("No kr " + defaultKR + " in " + corpus);
|
|
| 54 |
Log.severe(NLS.bind(Messages.ErrorNoKnowledgeRepositoryP0FoundInP1Corpus, defaultKR, corpus));
|
|
| 55 |
throw new IllegalArgumentException("No kr " + defaultKR + " in " + corpus); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 56 | 56 |
} |
| 57 | 57 |
|
| 58 | 58 |
types = new ArrayList<AnnotationType>(); |
| ... | ... | |
| 78 | 78 |
* @throws XMLStreamException |
| 79 | 79 |
*/ |
| 80 | 80 |
protected boolean writeTextAnnotationToSyMoGIH(String textid, File currentXMLFile, File currentXMLStandoffFile, File xmlStandOffDirectory) throws IOException, CqiServerError, CqiClientException, InvalidCqpIdException, XMLStreamException {
|
| 81 |
Log.info(" text=" + textid);
|
|
| 81 |
Log.info(Messages.TestP0 + textid);
|
|
| 82 | 82 |
boolean show_debug = Log.getLevel().intValue() < Level.INFO.intValue(); |
| 83 | 83 |
|
| 84 | 84 |
AnnotationSyMoGIHWriter annotationstdoff = new AnnotationSyMoGIHWriter(textid, currentXMLFile, xmlStandOffDirectory, types, show_debug); |
| ... | ... | |
| 89 | 89 |
return true; |
| 90 | 90 |
} |
| 91 | 91 |
else {
|
| 92 |
Log.warning(NLS.bind("Error: the {0} result file is malformed.", currentXMLStandoffFile));
|
|
| 92 |
Log.warning(NLS.bind(Messages.ErrorTheP0ResultFileIsMalformed, currentXMLStandoffFile));
|
|
| 93 | 93 |
} |
| 94 | 94 |
} |
| 95 | 95 |
else {
|
| 96 |
Log.warning(NLS.bind("Error: while processing the {0} file in standoff dir", currentXMLStandoffFile));
|
|
| 96 |
Log.warning(NLS.bind(Messages.ErrorWhileProcessingTheP0FileInStandOffDir, currentXMLStandoffFile));
|
|
| 97 | 97 |
} |
| 98 | 98 |
return false; |
| 99 | 99 |
|
| ... | ... | |
| 113 | 113 |
public boolean writeAnnotationsInStandoff(File resultZipFile) throws IOException, CqiServerError, CqiClientException, InvalidCqpIdException, XMLStreamException {
|
| 114 | 114 |
|
| 115 | 115 |
List<String> textsIds = Arrays.asList(corpus.getCorpusTextIdsList()); |
| 116 |
Log.info(NLS.bind("Exporting annotations of the {0} text(s).", StringUtils.join(textsIds, ", ")));
|
|
| 116 |
Log.info(NLS.bind(Messages.ExportingAnnotationsOfTheP0Texts, StringUtils.join(textsIds, ", "))); //$NON-NLS-2$
|
|
| 117 | 117 |
|
| 118 |
File resultDirectory = new File(Toolbox.getTxmHomePath(), "results/" + corpus.getID() + "_annotations"); |
|
| 118 |
File resultDirectory = new File(Toolbox.getTxmHomePath(), "results/" + corpus.getID() + "_annotations"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 119 | 119 |
DeleteDir.deleteDirectory(resultDirectory); |
| 120 | 120 |
resultDirectory.mkdirs(); |
| 121 | 121 |
if (!(resultDirectory.exists() && resultDirectory.canWrite())) {
|
| 122 |
Log.warning(NLS.bind("ERROR: could not create/write temporary directory: {0}.", resultDirectory));
|
|
| 122 |
Log.warning(NLS.bind(Messages.ErrorCouldNotCreateWriteTemporaryDirectoryP0, resultDirectory));
|
|
| 123 | 123 |
return false; |
| 124 | 124 |
} |
| 125 | 125 |
|
| 126 | 126 |
File inputDirectory = corpus.getProjectDirectory(); |
| 127 |
File corpusTxmDirectory = new File(inputDirectory, "txm/" + corpus.getID()); |
|
| 127 |
File corpusTxmDirectory = new File(inputDirectory, "txm/" + corpus.getID()); //$NON-NLS-1$
|
|
| 128 | 128 |
|
| 129 |
Log.info(NLS.bind("Writing annotations XML files in {0}.", resultDirectory));
|
|
| 129 |
Log.info(NLS.bind(Messages.WritingAnnotationsXMLFilesInP0, resultDirectory));
|
|
| 130 | 130 |
for (String textid : textsIds) {
|
| 131 |
File currentXMLFile = new File(corpusTxmDirectory, textid + ".xml"); |
|
| 131 |
File currentXMLFile = new File(corpusTxmDirectory, textid + ".xml"); //$NON-NLS-1$
|
|
| 132 | 132 |
if (!currentXMLFile.exists()) {
|
| 133 |
Log.warning(NLS.bind("Error: cannot found ab XML-TXM file for text with the ''{0}'' id.", textid));
|
|
| 133 |
Log.warning(NLS.bind(Messages.ErrorCannotFoundAXMLTXMFileForTextWithTheP0Id, textid));
|
|
| 134 | 134 |
return false; |
| 135 | 135 |
} |
| 136 |
File currentXMLStandoffFile = new File(resultDirectory, textid + ".xml"); // To Be Changed ? |
|
| 136 |
File currentXMLStandoffFile = new File(resultDirectory, textid + ".xml"); // To Be Changed ? //$NON-NLS-1$
|
|
| 137 | 137 |
if (!writeTextAnnotationToSyMoGIH(textid, currentXMLFile, currentXMLStandoffFile, resultDirectory)) {
|
| 138 |
Log.warning(NLS.bind("Error while writing annotations of text in {0}.", currentXMLStandoffFile));
|
|
| 138 |
Log.warning(NLS.bind(Messages.ErrorWhileWritingAnnotationsOfTextInP0, currentXMLStandoffFile));
|
|
| 139 | 139 |
return false; |
| 140 | 140 |
} |
| 141 | 141 |
} |
| ... | ... | |
| 143 | 143 |
Zip.compress(resultDirectory, resultZipFile, new ConsoleProgressBar(1)); |
| 144 | 144 |
DeleteDir.deleteDirectory(resultDirectory); |
| 145 | 145 |
|
| 146 |
Log.info(NLS.bind("Annotations saved in ''{0}''.", resultZipFile.getAbsolutePath()));
|
|
| 146 |
Log.info(NLS.bind(Messages.AnnotationsSavedInP0, resultZipFile.getAbsolutePath()));
|
|
| 147 | 147 |
return resultZipFile.exists(); |
| 148 | 148 |
} |
| 149 | 149 |
|
| ... | ... | |
| 169 | 169 |
List<String> textsIds = Arrays.asList(corpus.getCorpusTextIdsList()); |
| 170 | 170 |
|
| 171 | 171 |
File inputDirectory = corpus.getProjectDirectory(); |
| 172 |
File txmDirectory = new File(inputDirectory, "txm/" + corpus.getID()); |
|
| 172 |
File txmDirectory = new File(inputDirectory, "txm/" + corpus.getID()); //$NON-NLS-1$
|
|
| 173 | 173 |
|
| 174 | 174 |
ArrayList<Annotation> textAnnotations = new ArrayList<Annotation>(); |
| 175 | 175 |
HashMap<String, ArrayList<Annotation>> annotationsPerTexts = new HashMap<String, ArrayList<Annotation>>(); |
| 176 | 176 |
|
| 177 | 177 |
int currentText = 0; |
| 178 |
File currentXMLFile = new File(txmDirectory, textsIds.get(currentText) + ".xml"); |
|
| 178 |
File currentXMLFile = new File(txmDirectory, textsIds.get(currentText) + ".xml"); //$NON-NLS-1$
|
|
| 179 | 179 |
|
| 180 | 180 |
textAnnotations = new ArrayList<Annotation>(); |
| 181 | 181 |
annotationsPerTexts.put(textsIds.get(currentText), textAnnotations); |
| ... | ... | |
| 206 | 206 |
textAnnotations.add(currentAnnot); |
| 207 | 207 |
} |
| 208 | 208 |
|
| 209 |
File tmpXMLTXMDirectory = new File(txmDirectory.getAbsolutePath() + "_tmp"); |
|
| 209 |
File tmpXMLTXMDirectory = new File(txmDirectory.getAbsolutePath() + "_tmp"); //$NON-NLS-1$
|
|
| 210 | 210 |
DeleteDir.deleteDirectory(tmpXMLTXMDirectory); |
| 211 | 211 |
tmpXMLTXMDirectory.mkdirs(); |
| 212 | 212 |
if (!(tmpXMLTXMDirectory.exists() && tmpXMLTXMDirectory.canWrite())) {
|
| 213 |
Log.warning(NLS.bind("ERROR: could not create directory: {0}.", tmpXMLTXMDirectory));
|
|
| 213 |
Log.warning(NLS.bind(Messages.ErrorCouldNotCreateDirectoryP0, tmpXMLTXMDirectory));
|
|
| 214 | 214 |
return false; |
| 215 | 215 |
} |
| 216 | 216 |
|
| 217 |
File previousXMLTXMDirectory = new File(txmDirectory.getAbsolutePath() + "_previous"); |
|
| 217 |
File previousXMLTXMDirectory = new File(txmDirectory.getAbsolutePath() + "_previous"); //$NON-NLS-1$
|
|
| 218 | 218 |
// DeleteDir.deleteDirectory(tmpXMLTXMDirectory); |
| 219 | 219 |
previousXMLTXMDirectory.mkdirs(); |
| 220 | 220 |
if (!(previousXMLTXMDirectory.exists() && previousXMLTXMDirectory.canWrite())) {
|
| 221 |
Log.warning(NLS.bind("ERROR: could not create directory: {0}.", previousXMLTXMDirectory));
|
|
| 221 |
Log.warning(NLS.bind(Messages.ErrorCouldNotCreateDirectoryP0, previousXMLTXMDirectory));
|
|
| 222 | 222 |
return false; |
| 223 | 223 |
} |
| 224 | 224 |
|
| 225 |
Log.fine("Annotations grouped per text for " + annotationsPerTexts.size() + " text" + (annotationsPerTexts.size() > 0 ? "s" : ""));
|
|
| 226 |
Log.fine(" - Writing temporary XML files in: " + tmpXMLTXMDirectory);
|
|
| 227 |
Log.fine(" - Copying previous version of XML files in: " + previousXMLTXMDirectory);
|
|
| 225 |
Log.fine("Annotations grouped per text for " + annotationsPerTexts.size() + " text" + (annotationsPerTexts.size() > 0 ? "s" : "")); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$ //$NON-NLS-4$
|
|
| 226 |
Log.fine(" - Writing temporary XML files in: " + tmpXMLTXMDirectory); //$NON-NLS-1$
|
|
| 227 |
Log.fine(" - Copying previous version of XML files in: " + previousXMLTXMDirectory); //$NON-NLS-1$
|
|
| 228 | 228 |
// for all annotation PER TEXT, update the XML-TXM files |
| 229 | 229 |
currentText = 0; |
| 230 | 230 |
ConsoleProgressBar cpb = new ConsoleProgressBar(annotationsPerTexts.size()); |
| 231 | 231 |
while (currentText < end_limits.length) { // end limits : 10, 30, 45, 55, 103
|
| 232 |
currentXMLFile = new File(txmDirectory, textsIds.get(currentText) + ".xml"); |
|
| 232 |
currentXMLFile = new File(txmDirectory, textsIds.get(currentText) + ".xml"); //$NON-NLS-1$
|
|
| 233 | 233 |
ArrayList<Annotation> allAnnotations = annotationsPerTexts.get(textsIds.get(currentText)); |
| 234 | 234 |
if (allAnnotations != null && allAnnotations.size() > 0) {
|
| 235 | 235 |
ArrayList<Annotation> allSegmentAnnotations = new ArrayList<Annotation>(); |
| ... | ... | |
| 246 | 246 |
allTokenAnnotations.add(a); |
| 247 | 247 |
} |
| 248 | 248 |
else {
|
| 249 |
Log.fine("Annotation " + a + " with type=" + a.getType() + " not found in default KR=" + defaultKR);
|
|
| 249 |
Log.fine("Annotation " + a + " with type=" + a.getType() + " not found in default KR=" + defaultKR); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
|
| 250 | 250 |
} |
| 251 | 251 |
} |
| 252 | 252 |
else {
|
| 253 |
Log.warning("Warning: unknowed type: " + a.getType());
|
|
| 253 |
Log.warning(Messages.WarningUnknownTypeP0 + a.getType());
|
|
| 254 | 254 |
} |
| 255 | 255 |
} |
| 256 | 256 |
|
| 257 | 257 |
if (!writeAnnotationsInFile(currentXMLFile, start_limits[currentText], |
| 258 | 258 |
allSegmentAnnotations, allTokenAnnotations, |
| 259 | 259 |
tmpXMLTXMDirectory, previousXMLTXMDirectory)) {
|
| 260 |
Log.severe("Error while writing annotations of text " + currentXMLFile);
|
|
| 260 |
Log.severe(Messages.ErrorWhileWritingAnnotationsOfTextP0 + currentXMLFile);
|
|
| 261 | 261 |
return false; |
| 262 | 262 |
} |
| 263 | 263 |
else {
|
| ... | ... | |
| 279 | 279 |
protected boolean writeAnnotationsInFile(File xmlFile, int text_start_position, |
| 280 | 280 |
ArrayList<Annotation> segmentAnnotations, ArrayList<Annotation> tokenAnnotations, File tmpXMLTXMDirectory, File previousXMLTXMDirectory) throws CqiClientException, IOException, |
| 281 | 281 |
CqiServerError, InvalidCqpIdException, XMLStreamException {
|
| 282 |
Log.fine("Writing annotations for text " + xmlFile + " segment annotations=" + segmentAnnotations.size() + " token annotations=" + tokenAnnotations.size());
|
|
| 282 |
Log.fine("Writing annotations for text " + xmlFile + " segment annotations=" + segmentAnnotations.size() + " token annotations=" + tokenAnnotations.size()); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
|
| 283 | 283 |
// System.out.println(segmentAnnotations); |
| 284 | 284 |
// System.out.println(tokenAnnotations); |
| 285 | 285 |
|
| ... | ... | |
| 297 | 297 |
return true; |
| 298 | 298 |
} |
| 299 | 299 |
else {
|
| 300 |
Log.severe("Error: could not replace XML-TXM file: " + xmlFile + " with " + tmpfile);
|
|
| 300 |
Log.severe(Messages.ErrorCouldNotReplaceXMLTXMFileP0WithP1 + xmlFile + " with " + tmpfile); //$NON-NLS-2$
|
|
| 301 | 301 |
} |
| 302 | 302 |
} |
| 303 | 303 |
else {
|
| 304 |
Log.severe("Error: result file " + tmpfile + " is malformed.");
|
|
| 304 |
Log.severe(Messages.ErrorResultFileP0IsMalformed + tmpfile + " is malformed."); //$NON-NLS-2$
|
|
| 305 | 305 |
} |
| 306 | 306 |
} |
| 307 | 307 |
else {
|
| 308 |
Log.severe("Error: while processing " + xmlFile + " in temp dir");
|
|
| 308 |
Log.severe(Messages.ErrorWhileProcessingP0InTempDir + xmlFile + " in temp dir"); //$NON-NLS-2$
|
|
| 309 | 309 |
} |
| 310 | 310 |
return false; |
| 311 | 311 |
} |
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/AnnotationSyMoGIHWriter.java (revision 3883) | ||
|---|---|---|
| 15 | 15 |
import javax.xml.stream.XMLStreamWriter; |
| 16 | 16 |
|
| 17 | 17 |
import org.eclipse.osgi.util.NLS; |
| 18 |
import org.txm.annotation.kr.core.messages.Messages; |
|
| 18 | 19 |
import org.txm.annotation.kr.core.repository.AnnotationType; |
| 19 | 20 |
import org.txm.importer.StaxIdentityParser; |
| 20 | 21 |
import org.txm.utils.logger.Log; |
| ... | ... | |
| 142 | 143 |
*/ |
| 143 | 144 |
boolean foundAnnot = false; |
| 144 | 145 |
|
| 145 |
if (types.contains(localname) && parser.getPrefix().equals("txm")) { // don't write txm annotation elements
|
|
| 146 |
if (types.contains(localname) && parser.getPrefix().equals("txm")) { // don't write txm annotation elements //$NON-NLS-1$
|
|
| 146 | 147 |
// System.out.println(" START "+ localname);
|
| 147 | 148 |
foundAnnot = true; |
| 148 | 149 |
currentType = localname; |
| 149 | 150 |
// <txm:actr author="gazelledess" ref="PhileasFogg" date="2016-09-05" start="56" end="57"> |
| 150 |
currentAuthor = parser.getAttributeValue(null, "author"); |
|
| 151 |
currentRef = parser.getAttributeValue(null, "ref"); |
|
| 152 |
currentStartPos = parser.getAttributeValue(null, "start"); |
|
| 153 |
currentEndPos = parser.getAttributeValue(null, "end"); |
|
| 154 |
currentDate = parser.getAttributeValue(null, "date"); |
|
| 151 |
currentAuthor = parser.getAttributeValue(null, "author"); //$NON-NLS-1$
|
|
| 152 |
currentRef = parser.getAttributeValue(null, "ref"); //$NON-NLS-1$
|
|
| 153 |
currentStartPos = parser.getAttributeValue(null, "start"); //$NON-NLS-1$
|
|
| 154 |
currentEndPos = parser.getAttributeValue(null, "end"); //$NON-NLS-1$
|
|
| 155 |
currentDate = parser.getAttributeValue(null, "date"); //$NON-NLS-1$
|
|
| 155 | 156 |
// annotation is here |
| 156 | 157 |
startAnnotation = true; |
| 157 | 158 |
positions = new ArrayList<String>(); |
| ... | ... | |
| 176 | 177 |
super.processStartElement(); /// continue writing in file all elements, except the tags that are now in stand-off files |
| 177 | 178 |
|
| 178 | 179 |
// get words ids of the current annotations |
| 179 |
if (localname.equals("w") && startAnnotation) {
|
|
| 180 |
String id = parser.getAttributeValue(null, "id"); |
|
| 180 |
if (localname.equals("w") && startAnnotation) { //$NON-NLS-1$
|
|
| 181 |
String id = parser.getAttributeValue(null, "id"); //$NON-NLS-1$
|
|
| 181 | 182 |
for (String typeIn : annotationsPositions.keySet()) {
|
| 182 | 183 |
positions = annotationsPositions.get(typeIn); |
| 183 | 184 |
positions.add(id); |
| ... | ... | |
| 217 | 218 |
private XMLStreamWriter writeStartStandOffFile(File file) {
|
| 218 | 219 |
// System.out.println("writeStartStandOffFile ...");
|
| 219 | 220 |
|
| 220 |
String ns = "http://www.tei-c.org/ns/1.0"; |
|
| 221 |
String ns = "http://www.tei-c.org/ns/1.0"; //$NON-NLS-1$
|
|
| 221 | 222 |
XMLOutputFactory output = XMLOutputFactory.newInstance(); |
| 222 | 223 |
XMLStreamWriter writer = null; |
| 223 | 224 |
try {
|
| 224 | 225 |
writer = output.createXMLStreamWriter(new FileWriter(file)); |
| 225 | 226 |
writer.writeStartDocument(); |
| 226 |
writer.setPrefix("tei", ns);
|
|
| 227 |
writer.setPrefix("tei", ns); //$NON-NLS-1$
|
|
| 227 | 228 |
writer.setDefaultNamespace(ns); |
| 228 | 229 |
|
| 229 |
writer.writeStartElement("TEI");
|
|
| 230 |
writer.writeStartElement("TEI"); //$NON-NLS-1$
|
|
| 230 | 231 |
|
| 231 |
writer.writeStartElement("teiHeader");
|
|
| 232 |
writer.writeStartElement("fileDesc");
|
|
| 232 |
writer.writeStartElement("teiHeader"); //$NON-NLS-1$
|
|
| 233 |
writer.writeStartElement("fileDesc"); //$NON-NLS-1$
|
|
| 233 | 234 |
|
| 234 |
writer.writeStartElement("titleStmt");
|
|
| 235 |
writer.writeStartElement("title");
|
|
| 235 |
writer.writeStartElement("titleStmt"); //$NON-NLS-1$
|
|
| 236 |
writer.writeStartElement("title"); //$NON-NLS-1$
|
|
| 236 | 237 |
writer.writeCharacters(textid); |
| 237 | 238 |
writer.writeEndElement(); // title |
| 238 | 239 |
writer.writeEndElement(); // titleStmt |
| 239 | 240 |
|
| 240 |
writer.writeStartElement("publicationStmt");
|
|
| 241 |
writer.writeStartElement("p");
|
|
| 242 |
writer.writeCharacters("PUBLICATION INFO à renseigner");
|
|
| 241 |
writer.writeStartElement("publicationStmt"); //$NON-NLS-1$
|
|
| 242 |
writer.writeStartElement("p"); //$NON-NLS-1$
|
|
| 243 |
writer.writeCharacters("PUBLICATION INFO à renseigner"); //$NON-NLS-1$
|
|
| 243 | 244 |
writer.writeEndElement(); // p |
| 244 | 245 |
writer.writeEndElement(); // publicationStmt |
| 245 | 246 |
|
| 246 |
writer.writeStartElement("sourceDesc");
|
|
| 247 |
writer.writeStartElement("p");
|
|
| 248 |
writer.writeCharacters("Ce document permet l'annotation sémantique de tous les textes, par auteur");
|
|
| 247 |
writer.writeStartElement("sourceDesc"); //$NON-NLS-1$
|
|
| 248 |
writer.writeStartElement("p"); //$NON-NLS-1$
|
|
| 249 |
writer.writeCharacters("Ce document permet l'annotation sémantique de tous les textes, par auteur"); //$NON-NLS-1$
|
|
| 249 | 250 |
writer.writeEndElement(); // p |
| 250 | 251 |
writer.writeEndElement(); // sourceDesc |
| 251 | 252 |
|
| 252 | 253 |
writer.writeEndElement(); // </fileDesc> |
| 253 |
writer.writeStartElement("encodingDesc");
|
|
| 254 |
writer.writeStartElement("projectDesc");
|
|
| 255 |
writer.writeStartElement("p");
|
|
| 256 |
writer.writeCharacters("Annotations created by " + currentAuthor + ", for the use in Symogih XML platform");
|
|
| 254 |
writer.writeStartElement("encodingDesc"); //$NON-NLS-1$
|
|
| 255 |
writer.writeStartElement("projectDesc"); //$NON-NLS-1$
|
|
| 256 |
writer.writeStartElement("p"); //$NON-NLS-1$
|
|
| 257 |
writer.writeCharacters("Annotations created by " + currentAuthor + ", for the use in Symogih XML platform"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 257 | 258 |
writer.writeEndElement(); // p |
| 258 | 259 |
writer.writeEndElement(); // </projectDesc> |
| 259 | 260 |
writer.writeEndElement(); // </encodingDesc> |
| 260 | 261 |
writer.writeEndElement(); // </teiHeader> |
| 261 | 262 |
|
| 262 | 263 |
|
| 263 |
writer.writeStartElement("text");
|
|
| 264 |
writer.writeStartElement("body");
|
|
| 265 |
writer.writeCharacters("\n");
|
|
| 266 |
writer.writeStartElement("div");
|
|
| 267 |
writer.writeCharacters("\n");
|
|
| 264 |
writer.writeStartElement("text"); //$NON-NLS-1$
|
|
| 265 |
writer.writeStartElement("body"); //$NON-NLS-1$
|
|
| 266 |
writer.writeCharacters("\n"); //$NON-NLS-1$
|
|
| 267 |
writer.writeStartElement("div"); //$NON-NLS-1$
|
|
| 268 |
writer.writeCharacters("\n"); //$NON-NLS-1$
|
|
| 268 | 269 |
} |
| 269 | 270 |
catch (XMLStreamException e) {
|
| 270 | 271 |
// TODO Auto-generated catch block |
| ... | ... | |
| 291 | 292 |
private void writeStartAnnotationToStandoffFile(String[] data, ArrayList<String> positions) {
|
| 292 | 293 |
// System.out.println("writeStartAnnotationToStandoffFile ...");
|
| 293 | 294 |
try {
|
| 294 |
currentWriter.writeStartElement("span");
|
|
| 295 |
currentWriter.writeAttribute("type", data[2]);
|
|
| 296 |
currentWriter.writeAttribute("ana", data[1]);
|
|
| 295 |
currentWriter.writeStartElement("span"); //$NON-NLS-1$
|
|
| 296 |
currentWriter.writeAttribute("type", data[2]); //$NON-NLS-1$
|
|
| 297 |
currentWriter.writeAttribute("ana", data[1]); //$NON-NLS-1$
|
|
| 297 | 298 |
StringBuffer listWids = new StringBuffer(); |
| 298 | 299 |
for (String posW : positions) {
|
| 299 |
listWids.append("#" + posW + " ");
|
|
| 300 |
listWids.append("#" + posW + " "); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 300 | 301 |
} |
| 301 |
currentWriter.writeAttribute("target", listWids.toString().trim());
|
|
| 302 |
currentWriter.writeAttribute("target", listWids.toString().trim()); //$NON-NLS-1$
|
|
| 302 | 303 |
currentWriter.writeEndElement(); // span |
| 303 |
currentWriter.writeCharacters("\n");
|
|
| 304 |
currentWriter.writeCharacters("\n"); //$NON-NLS-1$
|
|
| 304 | 305 |
} |
| 305 | 306 |
catch (XMLStreamException e) {
|
| 306 | 307 |
// TODO Auto-generated catch block |
| ... | ... | |
| 310 | 311 |
|
| 311 | 312 |
protected void processEndElement() throws XMLStreamException {
|
| 312 | 313 |
boolean foundAnnot = false; |
| 313 |
if (types.contains(localname) && parser.getPrefix().equals("txm")) { // skip annotation end element
|
|
| 314 |
if (types.contains(localname) && parser.getPrefix().equals("txm")) { // skip annotation end element //$NON-NLS-1$
|
|
| 314 | 315 |
// System.out.println(" END "+ localname);
|
| 315 | 316 |
foundAnnot = true; |
| 316 | 317 |
// annotation ends here |
| ... | ... | |
| 324 | 325 |
} |
| 325 | 326 |
|
| 326 | 327 |
if (!foundAnnot) {
|
| 327 |
if (localname.equals("TEI")) {
|
|
| 328 |
if (localname.equals("TEI")) { //$NON-NLS-1$
|
|
| 328 | 329 |
try {
|
| 329 | 330 |
for (String author : allannotations.keySet()) {
|
| 330 |
Log.info(" author=" + author);
|
|
| 331 |
File currentXmlFile = new File(xmlStandOffDirectory, filename.substring(0, filename.length() - 4) + "_" + currentAuthor + "_annotations.xml"); |
|
| 331 |
Log.info(" author=" + author); //$NON-NLS-1$
|
|
| 332 |
File currentXmlFile = new File(xmlStandOffDirectory, filename.substring(0, filename.length() - 4) + "_" + currentAuthor + "_annotations.xml"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 332 | 333 |
|
| 333 | 334 |
// write start of the stand-off file |
| 334 | 335 |
currentWriter = writeStartStandOffFile(currentXmlFile); |
| ... | ... | |
| 340 | 341 |
ArrayList<String[]> datas = dates.get(date); |
| 341 | 342 |
ArrayList<ArrayList<String>> positions = datesPositions.get(date); |
| 342 | 343 |
|
| 343 |
currentWriter.writeStartElement("div");
|
|
| 344 |
currentWriter.writeCharacters("\n");
|
|
| 345 |
currentWriter.writeStartElement("head");
|
|
| 346 |
currentWriter.writeStartElement("date");
|
|
| 347 |
currentWriter.writeAttribute("type", "annotation_date");
|
|
| 348 |
currentWriter.writeAttribute("when", date);
|
|
| 344 |
currentWriter.writeStartElement("div"); //$NON-NLS-1$
|
|
| 345 |
currentWriter.writeCharacters("\n"); //$NON-NLS-1$
|
|
| 346 |
currentWriter.writeStartElement("head"); //$NON-NLS-1$
|
|
| 347 |
currentWriter.writeStartElement("date"); //$NON-NLS-1$
|
|
| 348 |
currentWriter.writeAttribute("type", "annotation_date"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 349 |
currentWriter.writeAttribute("when", date); //$NON-NLS-1$
|
|
| 349 | 350 |
currentWriter.writeEndElement(); // date |
| 350 | 351 |
currentWriter.writeEndElement(); // head |
| 351 |
currentWriter.writeCharacters("\n");
|
|
| 352 |
currentWriter.writeCharacters("\n"); //$NON-NLS-1$
|
|
| 352 | 353 |
|
| 353 | 354 |
for (int i = 0; i < datas.size(); i++) {
|
| 354 | 355 |
writeStartAnnotationToStandoffFile(datas.get(i), positions.get(i)); |
| 355 | 356 |
} |
| 356 | 357 |
|
| 357 | 358 |
currentWriter.writeEndElement(); // div |
| 358 |
currentWriter.writeCharacters("\n");
|
|
| 359 |
currentWriter.writeCharacters("\n"); //$NON-NLS-1$
|
|
| 359 | 360 |
} |
| 360 | 361 |
|
| 361 | 362 |
// write the end of the stand-off file |
| ... | ... | |
| 363 | 364 |
} |
| 364 | 365 |
} |
| 365 | 366 |
catch (Exception e) {
|
| 366 |
Log.warning(NLS.bind("Error with current writer: {0}.", currentWriter));
|
|
| 367 |
Log.warning(NLS.bind(Messages.ErrorWithCurrentWriterP0, currentWriter));
|
|
| 367 | 368 |
Log.printStackTrace(e); |
| 368 | 369 |
} |
| 369 | 370 |
} |
| ... | ... | |
| 373 | 374 |
|
| 374 | 375 |
public static void main(String strs[]) {
|
| 375 | 376 |
try {
|
| 376 |
File xmlFile = new File("/home/mdecorde/TXM/corpora/TDM80J/txm/TDM80J", "tdm80j.xml");
|
|
| 377 |
File outfile = new File(xmlFile.getParentFile(), "tdm80j-tei.xml"); |
|
| 377 |
File xmlFile = new File("/home/mdecorde/TXM/corpora/TDM80J/txm/TDM80J", "tdm80j.xml"); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 378 |
File outfile = new File(xmlFile.getParentFile(), "tdm80j-tei.xml"); //$NON-NLS-1$
|
|
| 378 | 379 |
|
| 379 | 380 |
File xmlStandOffDirectory = xmlFile.getParentFile(); |
| 380 | 381 |
List<AnnotationType> types = new ArrayList<AnnotationType>(); |
| 381 |
types.add(new AnnotationType("local", "actr_id", "actr"));
|
|
| 382 |
types.add(new AnnotationType("local", "actr_id", "actr")); //$NON-NLS-1$ //$NON-NLS-2$ //$NON-NLS-3$
|
|
| 382 | 383 |
AnnotationSyMoGIHWriter annotStdff; |
| 383 | 384 |
try {
|
| 384 |
annotStdff = new AnnotationSyMoGIHWriter("TDBM80J", xmlFile, xmlStandOffDirectory, types, true);
|
|
| 385 |
annotStdff = new AnnotationSyMoGIHWriter("TDBM80J", xmlFile, xmlStandOffDirectory, types, true); //$NON-NLS-1$
|
|
| 385 | 386 |
annotStdff.process(outfile); |
| 386 | 387 |
} |
| 387 | 388 |
catch (MalformedURLException e) {
|
| TXM/trunk/bundles/org.txm.annotation.kr.core/src/org/txm/annotation/kr/core/AnnotationInjector.java (revision 3883) | ||
|---|---|---|
| 15 | 15 |
|
| 16 | 16 |
import org.eclipse.osgi.util.NLS; |
| 17 | 17 |
import org.txm.Toolbox; |
| 18 |
import org.txm.annotation.kr.core.messages.Messages; |
|
| 18 | 19 |
import org.txm.importer.StaxIdentityParser; |
| 19 | 20 |
import org.txm.importer.ValidateXml; |
| 20 | 21 |
import org.txm.utils.logger.Log; |
| ... | ... | |
| 86 | 87 |
XMLStreamException {
|
| 87 | 88 |
super(xmlFile.toURI().toURL()); // init reader and writer |
| 88 | 89 |
|
| 89 |
this.user = System.getProperty("user.name");
|
|
| 90 |
this.user = System.getProperty("user.name"); //$NON-NLS-1$
|
|
| 90 | 91 |
|
| 91 | 92 |
this.debug = debug; |
| 92 | 93 |
this.xmlFile = xmlFile; |
| ... | ... | |
| 149 | 150 |
}); |
| 150 | 151 |
|
| 151 | 152 |
if (debug) {
|
| 152 |
Log.finer("annotations for " + xmlFile);
|
|
| 153 |
Log.finer(" segment grouped and ordered by start position: " + annotationsToAddByStartPos);
|
|
| 154 |
Log.finer(" segment grouped and ordered by end position: " + annotationsToAddByEndPos);
|
|
| 155 |
Log.finer(" token ordered by start position: " + tokenAnnotations);
|
|
| 153 |
Log.finer("annotations for " + xmlFile); //$NON-NLS-1$
|
|
| 154 |
Log.finer(" segment grouped and ordered by start position: " + annotationsToAddByStartPos); //$NON-NLS-1$
|
|
| 155 |
Log.finer(" segment grouped and ordered by end position: " + annotationsToAddByEndPos); //$NON-NLS-1$
|
|
| 156 |
Log.finer(" token ordered by start position: " + tokenAnnotations); //$NON-NLS-1$
|
|
| 156 | 157 |
} |
| 157 | 158 |
} |
| 158 | 159 |
|
| ... | ... | |
| 172 | 173 |
// System.out.println("A Starts: "+currentStartAnnotations+" "+currentEndAnnotations);
|
| 173 | 174 |
// if (debug) |
| 174 | 175 |
// System.out.println("----- "+localname+"@"+position_counter+" START ANNOT = "+currentStartAnnotations);
|
| 175 |
if ("teiHeader".equals(localname)) {
|
|
| 176 |
if ("teiHeader".equals(localname)) { //$NON-NLS-1$
|
|
| 176 | 177 |
super.processStartElement(); |
| 177 |
goToEnd("teiHeader"); // skip teiHeader // will be closed in
|
|
| 178 |
goToEnd("teiHeader"); // skip teiHeader // will be closed in //$NON-NLS-1$
|
|
| 178 | 179 |
// processEndElement() |
| 179 | 180 |
} |
| 180 |
else if ("w".equals(localname)) {
|
|
| 181 |
else if ("w".equals(localname)) { //$NON-NLS-1$
|
|
| 181 | 182 |
// System.out.println("W START pos="+position_counter+" annots="+currentStartAnnotations);
|
| 182 | 183 |
if (currentStartAnnotations != null) {// there are still annotations
|
| 183 | 184 |
// to write, possible |
| ... | ... | |
| 185 | 186 |
// if (debug) |
| 186 | 187 |
// System.out.println("WRITING START at "+position_counter+": "+currentStartAnnotations);
|
| 187 | 188 |
for (Annotation a : currentStartAnnotations) {
|
| 188 |
if (!"#del".equals(a.getValue())) {
|
|
| 189 |
if (!"#del".equals(a.getValue())) { //$NON-NLS-1$
|
|
| 189 | 190 |
if (debug) {
|
| 190 |
Log.finer(" force write start annotation " + a);
|
|
| 191 |
Log.finer(" force write start annotation " + a); //$NON-NLS-1$
|
|
| 191 | 192 |
} |
| 192 | 193 |
writeStartAnnotation(a); |
| 193 | 194 |
} |
| 194 | 195 |
else {
|
| 195 | 196 |
if (debug) |
| 196 |
Log.finer(" no need to write start annotation " + a);
|
|
| 197 |
Log.finer(" no need to write start annotation " + a); //$NON-NLS-1$
|
|
| 197 | 198 |
currentEndAnnotations = annotationsToAddByEndPos.get(a.getEnd()); |
| 198 | 199 |
// System.out.println(" shall we need to change annotation attribute : "+a.getValue()+
|
| 199 | 200 |
// " | currentEndAnnotations : "+currentEndAnnotations); |
| 200 | 201 |
if (currentEndAnnotations == null) {
|
| 201 |
Log.finer("WARNING ERROR null pointer for end position of annotation " + a);
|
|
| 202 |
Log.finer("WARNING ERROR null pointer for end position of annotation " + a); //$NON-NLS-1$
|
|
| 202 | 203 |
} |
| 203 | 204 |
else {
|
| 204 | 205 |
currentEndAnnotations.remove(a); |
| ... | ... | |
| 230 | 231 |
super.processStartElement(); // write the tag |
| 231 | 232 |
|
| 232 | 233 |
} |
| 233 |
else if ("form".equals(localname) && inW) {
|
|
| 234 |
if (currentTokenAnnotations.containsKey("word")) {
|
|
| 235 |
Annotation a = currentTokenAnnotations.get("word");
|
|
| 234 |
else if ("form".equals(localname) && inW) { //$NON-NLS-1$
|
|
| 235 |
if (currentTokenAnnotations.containsKey("word")) { //$NON-NLS-1$
|
|
| 236 |
Annotation a = currentTokenAnnotations.get("word"); //$NON-NLS-1$
|
|
| 236 | 237 |
// System.out.println("Updating token annotation with: " + a);
|
| 237 | 238 |
|
| 238 | 239 |
String value = a.getValue(); |
| 239 |
if ("#del".equals(value)) {
|
|
| 240 |
value = ""; |
|
| 240 |
if ("#del".equals(value)) { //$NON-NLS-1$
|
|
| 241 |
value = ""; //$NON-NLS-1$
|
|
| 241 | 242 |
} |
| 242 | 243 |
if (value == null) {
|
| 243 |
value = "ERROR"; |
|
| 244 |
value = "ERROR"; //$NON-NLS-1$
|
|
| 244 | 245 |
} |
| 245 | 246 |
try {
|
| 246 |
writer.writeStartElement(TXMNS, "form"); |
|
| 247 |
writer.writeStartElement(TXMNS, "form"); //$NON-NLS-1$
|
|
| 247 | 248 |
writer.writeCharacters(value); |
| 248 | 249 |
mustChangeWordValue = true; |
| 249 | 250 |
} |
| ... | ... | |
| 252 | 253 |
e.printStackTrace(); |
| 253 | 254 |
} |
| 254 | 255 |
|
| 255 |
currentTokenAnnotations.remove("word"); // annotation updated
|
|
| 256 |
currentTokenAnnotations.remove("word"); // annotation updated //$NON-NLS-1$
|
|
| 256 | 257 |
} |
| 257 | 258 |
else {
|
| 258 | 259 |
super.processStartElement(); |
| 259 | 260 |
} |
| 260 | 261 |
} |
| 261 |
else if ("ana".equals(localname) && inW) {
|
|
| 262 |
else if ("ana".equals(localname) && inW) { //$NON-NLS-1$
|
|
| 262 | 263 |
inAna = true; |
| 263 |
String type = parser.getAttributeValue(null, "type").substring(1); |
|
| 264 |
String type = parser.getAttributeValue(null, "type").substring(1); //$NON-NLS-1$
|
|
| 264 | 265 |
if (currentTokenAnnotations.containsKey(type)) {
|
| 265 | 266 |
Annotation a = currentTokenAnnotations.get(type); |
| 266 | 267 |
// System.out.println("Updating token annotation with: " + a);
|
| 267 | 268 |
|
| 268 | 269 |
String value = a.getValue(); |
| 269 |
if ("#del".equals(value)) {
|
|
| 270 |
value = ""; |
|
| 270 |
if ("#del".equals(value)) { //$NON-NLS-1$
|
|
| 271 |
value = ""; //$NON-NLS-1$
|
|
| 271 | 272 |
} |
| 272 | 273 |
if (value == null) {
|
| 273 |
value = "ERROR"; |
|
| 274 |
value = "ERROR"; //$NON-NLS-1$
|
|
| 274 | 275 |
} |
| 275 | 276 |
|
| 276 | 277 |
try {
|
| 277 |
writer.writeStartElement("ana");
|
|
| 278 |
writer.writeAttribute("type", "#" + type);
|
|
| 279 |
writer.writeAttribute("resp", "#" + a.getAnnotator()); // change
|
|
| 278 |
writer.writeStartElement("ana"); //$NON-NLS-1$
|
|
| 279 |
writer.writeAttribute("type", "#" + type); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 280 |
writer.writeAttribute("resp", "#" + a.getAnnotator()); // change //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 280 | 281 |
// resp |
| 281 | 282 |
writer.writeCharacters(value); |
| 282 | 283 |
mustChangeAnaValue = true; |
| ... | ... | |
| 295 | 296 |
else if (currentStartAnnotations != null) { // is it an annotations ?
|
| 296 | 297 |
// System.out.println("A START="+localname);
|
| 297 | 298 |
|
| 298 |
String s_end = parser.getAttributeValue(null, "end"); |
|
| 299 |
String s_start = parser.getAttributeValue(null, "start"); |
|
| 300 |
String s_author = parser.getAttributeValue(null, "author"); |
|
| 301 |
String s_ref = parser.getAttributeValue(null, "ref"); |
|
| 299 |
String s_end = parser.getAttributeValue(null, "end"); //$NON-NLS-1$
|
|
| 300 |
String s_start = parser.getAttributeValue(null, "start"); //$NON-NLS-1$
|
|
| 301 |
String s_author = parser.getAttributeValue(null, "author"); //$NON-NLS-1$
|
|
| 302 |
String s_ref = parser.getAttributeValue(null, "ref"); //$NON-NLS-1$
|
|
| 302 | 303 |
|
| 303 | 304 |
boolean toDelete = false; |
| 304 | 305 |
|
| ... | ... | |
| 328 | 329 |
// updated = true; |
| 329 | 330 |
// if (debug) |
| 330 | 331 |
// System.out.println(" found existing annotation "+a);
|
| 331 |
if ("#del".equals(a.getValue())) {
|
|
| 332 |
if ("#del".equals(a.getValue())) { //$NON-NLS-1$
|
|
| 332 | 333 |
toDelete = true; |
| 333 | 334 |
// currentEndAnnotations.remove(a); // MAYBE NOT TO |
| 334 | 335 |
// DO THIS HERE ? |
| 335 | 336 |
if (debug) {
|
| 336 |
Log.finer(" force delete start annotation " + a);
|
|
| 337 |
Log.finer(" force delete start annotation " + a); //$NON-NLS-1$
|
|
| 337 | 338 |
} |
| 338 | 339 |
} |
| 339 | 340 |
else { // update existing annotation, no need to store
|
| ... | ... | |
| 341 | 342 |
writeStartAnnotation(a); |
| 342 | 343 |
toDelete = true; |
| 343 | 344 |
if (debug) {
|
| 344 |
Log.finer(" update annotation " + a);
|
|
| 345 |
Log.finer(" update annotation " + a); //$NON-NLS-1$
|
|
| 345 | 346 |
} |
| 346 | 347 |
currentEndAnnotations = annotationsToAddByEndPos.get(a.getEnd());// EL |
| 347 | 348 |
// NO |
| ... | ... | |
| 351 | 352 |
// System.out.println(" shall we need to change annotation attribute : "+a.getValue()+
|
| 352 | 353 |
// " | currentEndAnnotations : "+currentEndAnnotations); |
| 353 | 354 |
if (currentEndAnnotations == null) {
|
| 354 |
Log.warning(NLS.bind("Error: null pointer for end position of annotation {0}.", a));
|
|
| 355 |
Log.warning(NLS.bind(Messages.ErrorNPEForEndPositionOfAnnotationP0, a));
|
|
| 355 | 356 |
} |
| 356 | 357 |
else {
|
| 357 | 358 |
currentEndAnnotations.remove(a); |
| ... | ... | |
| 369 | 370 |
|
| 370 | 371 |
if (s_start != null && s_end != null && s_author != null && s_ref != null) {
|
| 371 | 372 |
if (a.getEnd() >= end) { // must write a
|
| 372 |
if (!"#del".equals(a.getValue())) {
|
|
| 373 |
if (!"#del".equals(a.getValue())) { //$NON-NLS-1$
|
|
| 373 | 374 |
if (debug) {
|
| 374 |
Log.finer(NLS.bind(" writing of start annotation {0}.", a));
|
|
| 375 |
Log.finer(NLS.bind(" writing of start annotation {0}.", a)); //$NON-NLS-1$
|
|
| 375 | 376 |
} |
| 376 | 377 |
writeStartAnnotation(a); |
| 377 | 378 |
|
| ... | ... | |
| 379 | 380 |
else {
|
| 380 | 381 |
toDelete = true; |
| 381 | 382 |
if (debug) {
|
| 382 |
Log.finer(NLS.bind(" no writing of start annotation {0}.", a));
|
|
| 383 |
Log.finer(NLS.bind(" no writing of start annotation {0}.", a)); //$NON-NLS-1$
|
|
| 383 | 384 |
} |
| 384 | 385 |
currentEndAnnotations = annotationsToAddByEndPos.get(a.getEnd()); |
| 385 | 386 |
// System.out.println(" shall we need to change annotation attribute : "+a.getValue()+
|
| 386 | 387 |
// " | currentEndAnnotations : "+currentEndAnnotations); |
| 387 | 388 |
if (currentEndAnnotations == null) {
|
| 388 |
Log.warning(NLS.bind("Error: null pointer for end position of annotation {0}.", a));
|
|
| 389 |
Log.warning(NLS.bind(Messages.ErrorNPEForEndPositionOfAnnotationP0, a));
|
|
| 389 | 390 |
} |
| 390 | 391 |
else {
|
| 391 | 392 |
currentEndAnnotations.remove(a); |
| ... | ... | |
| 441 | 442 |
// if (debug) |
| 442 | 443 |
// System.out.println("----- "+localname+"@"+position_counter+" END ANNOT = "+currentEndAnnotations);
|
| 443 | 444 |
|
| 444 |
if ("w".equals(localname)) {
|
|
| 445 |
if ("w".equals(localname)) { //$NON-NLS-1$
|
|
| 445 | 446 |
// System.out.println("W END");
|
| 446 | 447 |
|
| 447 | 448 |
for (String type : currentTokenAnnotations.keySet()) {
|
| 448 | 449 |
try {
|
| 449 |
writer.writeStartElement("ana");
|
|
| 450 |
writer.writeAttribute("type", "#" + type);
|
|
| 451 |
writer.writeAttribute("resp", "#" + currentTokenAnnotations.get(type).getAnnotator()); // change
|
|
| 450 |
writer.writeStartElement("ana"); //$NON-NLS-1$
|
|
| 451 |
writer.writeAttribute("type", "#" + type); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 452 |
writer.writeAttribute("resp", "#" + currentTokenAnnotations.get(type).getAnnotator()); // change //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 452 | 453 |
// resp |
| 453 | 454 |
writer.writeCharacters(currentTokenAnnotations.get(type).getValue()); |
| 454 | 455 |
writer.writeEndElement(); |
| ... | ... | |
| 469 | 470 |
for (Annotation a : currentEndAnnotations) {
|
| 470 | 471 |
// if (debug) |
| 471 | 472 |
// System.out.println("=============== End annot : "+a);
|
| 472 |
if (!"#del".equals(a.getValue())) {
|
|
| 473 |
if (!"#del".equals(a.getValue())) { //$NON-NLS-1$
|
|
| 473 | 474 |
if (debug) {
|
| 474 |
Log.finer(" force write end annotation " + a);
|
|
| 475 |
Log.finer(" force write end annotation " + a); //$NON-NLS-1$
|
|
| 475 | 476 |
} |
| 476 | 477 |
writeEndAnnotation(a); |
| 477 | 478 |
} /* |
| ... | ... | |
| 485 | 486 |
|
| 486 | 487 |
position_counter++; |
| 487 | 488 |
} |
| 488 |
else if ("form".equals(localname) && inW) {
|
|
| 489 |
else if ("form".equals(localname) && inW) { //$NON-NLS-1$
|
|
| 489 | 490 |
// if (!mustChangeAnaValue) |
| 490 | 491 |
super.processEndElement(); |
| 491 | 492 |
mustChangeWordValue = false; |
| 492 | 493 |
inForm = false; |
| 493 | 494 |
} |
| 494 |
else if ("ana".equals(localname) && inW) {
|
|
| 495 |
else if ("ana".equals(localname) && inW) { //$NON-NLS-1$
|
|
| 495 | 496 |
// if (!mustChangeAnaValue) |
| 496 | 497 |
super.processEndElement(); |
| 497 | 498 |
inAna = false; |
| ... | ... | |
| 508 | 509 |
// if (debug) |
| 509 | 510 |
// System.out.println(" found existing end annotation "+a);
|
| 510 | 511 |
|
| 511 |
if ("#del".equals(a.getValue())) { // if
|
|
| 512 |
if ("#del".equals(a.getValue())) { // if //$NON-NLS-1$
|
|
| 512 | 513 |
// (!"#del".equals(a.getValue())) |
| 513 | 514 |
// {
|
| 514 | 515 |
toDelete = true; |
| 515 | 516 |
if (debug) {
|
| 516 |
Log.finer(" force delete end annotation " + a);
|
|
| 517 |
Log.finer(" force delete end annotation " + a); //$NON-NLS-1$
|
|
| 517 | 518 |
} |
| 518 | 519 |
} |
| 519 | 520 |
|
| ... | ... | |
| 538 | 539 |
|
| 539 | 540 |
private void writeStartAnnotation(Annotation a) {
|
| 540 | 541 |
try {
|
| 541 |
writer.writeStartElement("txm:" + a.getType());
|
|
| 542 |
writer.writeAttribute("author", "" + a.getAnnotator());
|
|
| 543 |
writer.writeAttribute("ref", a.getValue());
|
|
| 544 |
writer.writeAttribute("date", Toolbox.dateformat.format(new Date()));
|
|
| 545 |
writer.writeAttribute("start", Integer.toString(a.getStart()));
|
|
| 546 |
writer.writeAttribute("end", Integer.toString(a.getEnd()));
|
|
| 542 |
writer.writeStartElement("txm:" + a.getType()); //$NON-NLS-1$
|
|
| 543 |
writer.writeAttribute("author", "" + a.getAnnotator()); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 544 |
writer.writeAttribute("ref", a.getValue()); //$NON-NLS-1$
|
|
| 545 |
writer.writeAttribute("date", Toolbox.dateformat.format(new Date())); //$NON-NLS-1$
|
|
| 546 |
writer.writeAttribute("start", Integer.toString(a.getStart())); //$NON-NLS-1$
|
|
| 547 |
writer.writeAttribute("end", Integer.toString(a.getEnd())); //$NON-NLS-1$
|
|
| 547 | 548 |
} |
| 548 | 549 |
catch (XMLStreamException e) {
|
| 549 | 550 |
// TODO Auto-generated catch block |
| ... | ... | |
| 565 | 566 |
|
| 566 | 567 |
public static ArrayList<Annotation> testAdding() {
|
| 567 | 568 |
|
| 568 |
Annotation a1 = new Annotation("Actr", "Actr100", 15, 16);
|
|
| 569 |
Annotation a1 = new Annotation("Actr", "Actr100", 15, 16); //$NON-NLS-1$ //$NON-NLS-2$
|
|
| 569 | 570 |
ArrayList<Annotation> annotations = new ArrayList<>(); |
Formats disponibles : Unified diff