uima-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ofer Bronstein (JIRA)" <...@uima.apache.org>
Subject [jira] [Created] (UIMA-3724) Adding (J)CasUtil.selectCoveredSingle()
Date Tue, 08 Apr 2014 12:36:14 GMT
Ofer Bronstein created UIMA-3724:
------------------------------------

             Summary: Adding (J)CasUtil.selectCoveredSingle()
                 Key: UIMA-3724
                 URL: https://issues.apache.org/jira/browse/UIMA-3724
             Project: UIMA
          Issue Type: Improvement
          Components: uimaFIT
            Reporter: Ofer Bronstein
            Priority: Minor


JCasUtil (and its equivalent CASUtil) have two very useful methods:

- {{selectSingle(JCas jCas, Class<T> type)}}, retrieving all T type annotations from
the CAS, making sure only one exists, and returning it
- {{selectCovered(JCas jCas, final Class<T> type, AnnotationFS coveringAnnotation)}},
retrieving all T type annotations from the CAS, that exist in the span defined by the covering
annotation.

My request is to add a {{selectCoveredSingle}} method (I already implemented and tested, written
below). It would retrieve all T type annotation in the span of the covering annotation, make
sure there is only one, and return it.
The uses for such a method would be really common, as this assists in performing something
very basic in UIMA - once you have some annotation, you want to retrieve other information
on it, and if it is a small unit, then you know in advance that many types of information
should only appear once for that annotation. Otherwise, it's an error.
For instance, say I have linguistically annotated text, and I hold some intersting annotation
of a Token. Then I want to find out what is its Part Of Speech. Then I would simply call:
{{selectCoveredSingle(jcas, PartOfSpeech.class, token)}}, and get exactly one PartOfSpeech
annotation as I expected. If there is none or more then one, an exception is thrown.

One could also use a method like {{selectCoveredSingleOptional}} (not implemented yet), that
does exactly the same, but returns null when no such annotation exists, instead of throwing
an exception. This makes sense when a token can have some optional attribute. In linguistics,
it could be when a Token may be a NamedEntity, but not necessarily.

the implementation is made out of two methods - the first should be in JCasUtil, the second
should be in CASUtil.
===
Implementation:
===

	/**
	 * Get the annotation of the given annotation type constrained by a 'covering' annotation.
	 * Iterates over all annotations of the given type to find the covered annotations.
	 * Does not use subiterators.
	 *
	 * @param <T>
	 *            the JCas type.
	 * @param jCas
	 *            a JCas containing the annotation.
	 * @param type
	 *            a UIMA type.
	 * @param coveringAnnotation
	 *            the covering annotation.
	 * @return the single instance of the given type.
	 * @throws IllegalArgumentException if not exactly one instance if the given type is present
	 * 		   under the covering annotation.
	 * @see Subiterator
	 * 
	 * @author Ofer Bronstein
	 * @since April 2014
	 */
	@SuppressWarnings("unchecked")
	public static <T extends Annotation> T selectCoveredSingle(JCas jCas, final Class<T>
type,
			AnnotationFS coveringAnnotation) {
		return (T) selectCoveredSingle(jCas.getCas(), JCasUtil.getType(jCas, type),
				coveringAnnotation);
	}

	/**
	 * Get the annotation of the given annotation type constrained by a 'covering' annotation.
	 * Iterates over all annotations of the given type to find the covered annotations.
	 * Does not use subiterators.
	 *
	 * @param cas
	 *            a cas containing the annotation.
	 * @param type
	 *            a UIMA type.
	 * @param coveringAnnotation
	 *            the covering annotation.
	 * @return the single instance of the given type.
	 * @throws IllegalArgumentException if not exactly one instance if the given type is present
	 * 		   under the covering annotation.
	 * @see Subiterator
	 * 
	 * @author Ofer Bronstein
	 * @since April 2014
	 */
	public static AnnotationFS selectCoveredSingle(CAS cas, Type type,
			AnnotationFS coveringAnnotation) {
		List<AnnotationFS> annotations = CasUtil.selectCovered(cas, type, coveringAnnotation);
		
		if (annotations.isEmpty()) {
			throw new IllegalArgumentException("CAS does not contain any [" + type.getName() + "]");
		}
		if (annotations.size() > 1)  {
			throw new IllegalArgumentException("CAS contains more than one [" + type.getName()
					+ "]");
		}
		
		return annotations.get(0);
	}





--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message