Author: rwesten
Date: Tue May 15 09:10:33 2012
New Revision: 1338608
URL: http://svn.apache.org/viewvc?rev=1338608&view=rev
Log:
STANBOL-583: CELI Language Identifiaction Engine
* added assertion that "fr" is detected as language for the test text.
Modified:
incubator/stanbol/branches/celi-enhancement-engines/engines/celi/src/test/java/org/apache/stanbol/enhancer/engines/celi/langid/impl/CeliLanguageIdentifierEnhancementEngineTest.java
Modified:
incubator/stanbol/branches/celi-enhancement-engines/engines/celi/src/test/java/org/apache/stanbol/enhancer/engines/celi/langid/impl/CeliLanguageIdentifierEnhancementEngineTest.java
URL:
http://svn.apache.org/viewvc/incubator/stanbol/branches/celi-enhancement-engines/engines/celi/src/test/java/org/apache/stanbol/enhancer/engines/celi/langid/impl/CeliLanguageIdentifierEnhancementEngineTest.java?rev=1338608&r1=1338607&r2=1338608&view=diff
==============================================================================
---
incubator/stanbol/branches/celi-enhancement-engines/engines/celi/src/test/java/org/apache/stanbol/enhancer/engines/celi/langid/impl/CeliLanguageIdentifierEnhancementEngineTest.java
(original)
+++
incubator/stanbol/branches/celi-enhancement-engines/engines/celi/src/test/java/org/apache/stanbol/enhancer/engines/celi/langid/impl/CeliLanguageIdentifierEnhancementEngineTest.java
Tue May 15 09:10:33 2012
@@ -85,7 +85,13 @@ public class CeliLanguageIdentifierEnhan
expectedValues.put(Properties.DC_CREATOR,
LiteralFactory.getInstance().createTypedLiteral(
langIdentifier.getClass().getName()));
- validateLanguageAnnotation(ci.getMetadata(),
TEXT,expectedValues);
+ PlainLiteral detectedLnaguage =
validateLanguageAnnotation(ci.getMetadata(), TEXT,expectedValues);
+ //even through this tests do not validate service quality but
rather
+ //the correct integration of the CELI service as
EnhancementEngine
+ //we expect the "fr" is detected for the parsed text
+ assertEquals("The detected language for text '"+TEXT+"' MUST BE
'fr'",
+ "fr",detectedLnaguage.getLexicalForm());
+
int entityAnnoNum =
validateAllEntityAnnotations(ci.getMetadata(), expectedValues);
assertEquals("No EntityAnnotations are expected",0,
entityAnnoNum);
log.info(entityAnnoNum + " EntityAnnotations found ...");
@@ -99,7 +105,7 @@ public class CeliLanguageIdentifierEnhan
}
- private void validateLanguageAnnotation(MGraph g, String
content,HashMap<UriRef,Resource> expectedValues) {
+ private PlainLiteral validateLanguageAnnotation(MGraph g, String
content,HashMap<UriRef,Resource> expectedValues) {
Iterator<Triple> textAnnotationIterator = g.filter(null,
RDF_TYPE, ENHANCER_TEXTANNOTATION);
// test if a textAnnotation is present
assertTrue("The Language Annotation is
missing!",textAnnotationIterator.hasNext());
@@ -116,6 +122,7 @@ public class CeliLanguageIdentifierEnhan
assertTrue("The dc:langauge value MUST BE a plain
literal",languageResource instanceof PlainLiteral);
assertTrue("The dc:language value MIST BE at least two chars long",
((PlainLiteral)languageResource).getLexicalForm().length()>=2);
+ return (PlainLiteral)languageResource;
}
// removed: other tests now add a simple triple with
<{ciUri},dc:langauge,{lang}>
/** public static void addEnanchements(ContentItem ci) throws IOException,
ConfigurationException, EngineException {