org.apache.uima.jcas.cas.StringList Java Examples

/**
 * Annotate regex patterns (URLs, IPs, email addresses and Phone numbers)
 *
 * @param jcas
 *            the jcas
 * @param pattern
 *            the pattern
 * @param type
 *            the type
 * @return the array list
 */
public ArrayList<DictTerm> annotateRegex(JCas jcas, Pattern pattern, String type) {
	String docText = jcas.getDocumentText();
	ArrayList<DictTerm> regexMatches = new ArrayList<DictTerm>();
	Matcher matcher = pattern.matcher(docText);
	// Check all occurrences
	while (matcher.find()) {
		DictTerm dictTerm = new DictTerm(jcas);
		dictTerm.setBegin(matcher.start());
		dictTerm.setEnd(matcher.end());
		StringList typeList = new StringList(jcas);
		StringList baseFormList = new StringList(jcas);
		typeList = typeList.push(type);
		baseFormList = baseFormList.push(matcher.group());
		dictTerm.setDictType(typeList);
		dictTerm.setDictTerm(baseFormList);
		dictTerm.addToIndexes();
		regexMatches.add(dictTerm);
	}
	return regexMatches;
}
 

/**
 * Annotate dict types.
 *
 * @param jcas
 *            the jcas
 * @param token
 *            the token to annotate
 */
public void annotateDictTypes(JCas jcas, Token token) {

	String tokenStem = dictTermExtractor.stem(token.getCoveredText()).toLowerCase();
	String tokenValue = token.getCoveredText().toLowerCase();

	boolean dictTermFound = false;
	StringList typeList = new StringList(jcas);
	StringList baseFormList = new StringList(jcas);

	for (String dictType : unigramDictionaries.keySet()) {
		HashMap<String, String> dict = unigramDictionaries.get(dictType);
		if (dict.containsKey(tokenStem)) {
			String baseForm = dict.get(tokenStem);
			if (tokenValue.startsWith(baseForm)) {
				typeList = typeList.push(dictType);
				baseFormList = baseFormList.push(baseForm);
				dictTermFound = true;
			}
		}
	}

	// add to cas index
	if (dictTermFound) {
		DictTerm dictTerm = new DictTerm(jcas);
		dictTerm.setBegin(token.getBegin());
		dictTerm.setEnd(token.getEnd());
		dictTerm.setDictType(typeList);
		dictTerm.setDictTerm(baseFormList);
		dictTerm.addToIndexes();
	}

}
 

public static List<String> create(StringList aList) {
  List<String> data = new ArrayList<String>();
  StringList i = aList;
  while (i instanceof NonEmptyStringList) {
    NonEmptyStringList l = (NonEmptyStringList) i;
    data.add(l.getHead());
    i = l.getTail();
  }

  return asList(data.toArray(new String[data.size()]));
}
 

public void testStringListAPI() {
  StringList sl = new EmptyStringList(jcas);
  sl = sl.push("2");
  sl = sl.push("1");
  
  String[] sa = new String[2];
  int i = 0;
  for (String s : sl) {
    sa[i++] = s;
  }
  
  String[] expected = {"1",  "2"};
  assert(Arrays.equals(expected, sa));
}
 

/** getter for dictType - gets Type of dictionary added
 * @generated
 * @return value of the feature 
 */
public StringList getDictType() {
  if (DictTerm_Type.featOkTst && ((DictTerm_Type)jcasType).casFeat_dictType == null)
    jcasType.jcas.throwFeatMissing("dictType", "uhh_lt.newsleak.types.DictTerm");
  return (StringList)(jcasType.ll_cas.ll_getFSForRef(jcasType.ll_cas.ll_getRefValue(addr, ((DictTerm_Type)jcasType).casFeatCode_dictType)));}
 

/** setter for dictType - sets Type of dictionary added 
 * @generated
 * @param v value to set into the feature 
 */
public void setDictType(StringList v) {
  if (DictTerm_Type.featOkTst && ((DictTerm_Type)jcasType).casFeat_dictType == null)
    jcasType.jcas.throwFeatMissing("dictType", "uhh_lt.newsleak.types.DictTerm");
  jcasType.ll_cas.ll_setRefValue(addr, ((DictTerm_Type)jcasType).casFeatCode_dictType, jcasType.ll_cas.ll_getFSRef(v));}
 

/** getter for dictTerm - gets Base word types from dictionary list
 * @generated
 * @return value of the feature 
 */
public StringList getDictTerm() {
  if (DictTerm_Type.featOkTst && ((DictTerm_Type)jcasType).casFeat_dictTerm == null)
    jcasType.jcas.throwFeatMissing("dictTerm", "uhh_lt.newsleak.types.DictTerm");
  return (StringList)(jcasType.ll_cas.ll_getFSForRef(jcasType.ll_cas.ll_getRefValue(addr, ((DictTerm_Type)jcasType).casFeatCode_dictTerm)));}
 

/** setter for dictTerm - sets Base word types from dictionary list 
 * @generated
 * @param v value to set into the feature 
 */
public void setDictTerm(StringList v) {
  if (DictTerm_Type.featOkTst && ((DictTerm_Type)jcasType).casFeat_dictTerm == null)
    jcasType.jcas.throwFeatMissing("dictTerm", "uhh_lt.newsleak.types.DictTerm");
  jcasType.ll_cas.ll_setRefValue(addr, ((DictTerm_Type)jcasType).casFeatCode_dictTerm, jcasType.ll_cas.ll_getFSRef(v));}
 

public static StringList createStringList(JCas aJCas, String... aValues) {
  return createStringList(aJCas.getCas(), aValues);
}
 

public static StringList createStringList(JCas aJCas, Collection<String> aCollection) {
  return createStringList(aJCas.getCas(), aCollection);
}
 

/** getter for aListString - gets 
 * @generated
 * @return value of the feature 
 */
public StringList getAListString() { return (StringList)(_getFeatureValueNc(wrapGetIntCatchException(_FH_aListString)));}
 

/** setter for aListString - sets  
 * @generated
 * @param v value to set into the feature 
 */
public void setAListString(StringList v) {
  _setFeatureValueNcWj(wrapGetIntCatchException(_FH_aListString), v);
}
 

/** getter for aListMrString - gets 
 * @generated
 * @return value of the feature 
 */
public StringList getAListMrString() { return (StringList)(_getFeatureValueNc(wrapGetIntCatchException(_FH_aListMrString)));}
 

/** setter for aListMrString - sets  
 * @generated
 * @param v value to set into the feature 
 */
public void setAListMrString(StringList v) {
  _setFeatureValueNcWj(wrapGetIntCatchException(_FH_aListMrString), v);
}