Author: rwesten
Date: Wed May 16 09:44:41 2012
New Revision: 1339092
URL: http://svn.apache.org/viewvc?rev=1339092&view=rev
Log:
STANBOL-583: Lemmatizer engine now create valid enhancements. We should create
an own issue discussion how to represent Lemmas and Morphological features
within the Enhancement Structure.
Details:
LemmatizerEngine
* added extensive validation of the created enhancements
* corrected generation of the enhancements to confirm to the rules
* improved the HTTP client to stream the request data (similar to the HTTP
client of the NER engine)
Added a Utils class
* calculating the fise:selection-context (as it is used for NER and the
Lemmatizer Engine (might be even a candidate to move that to the
EnhancementEngineHelper utility
* creating a POST request (used by all **ClientHTTP classes)
EnhancementStructureTest:
* improved messages provided by assertions (to make debuging easier)
* removed assertion that at least a single TextAnnotation MUST BE present when
calling validateAllTextAnnotations. This allows to use this method to ensure
that no TextAnnotations are present (makes it consistent with
validateAllEntityAnnotations)
Modified:
incubator/stanbol/branches/celi-enhancement-engines/generic/test/src/main/java/org/apache/stanbol/enhancer/test/helper/EnhancementStructureHelper.java
Modified:
incubator/stanbol/branches/celi-enhancement-engines/generic/test/src/main/java/org/apache/stanbol/enhancer/test/helper/EnhancementStructureHelper.java
URL:
http://svn.apache.org/viewvc/incubator/stanbol/branches/celi-enhancement-engines/generic/test/src/main/java/org/apache/stanbol/enhancer/test/helper/EnhancementStructureHelper.java?rev=1339092&r1=1339091&r2=1339092&view=diff
==============================================================================
---
incubator/stanbol/branches/celi-enhancement-engines/generic/test/src/main/java/org/apache/stanbol/enhancer/test/helper/EnhancementStructureHelper.java
(original)
+++
incubator/stanbol/branches/celi-enhancement-engines/generic/test/src/main/java/org/apache/stanbol/enhancer/test/helper/EnhancementStructureHelper.java
Wed May 16 09:44:41 2012
@@ -58,7 +58,8 @@ public class EnhancementStructureHelper
Iterator<Triple> textAnnotationIterator = enhancements.filter(null,
RDF_TYPE, ENHANCER_TEXTANNOTATION);
// test if a textAnnotation is present
- assertTrue(textAnnotationIterator.hasNext());
+ //assertTrue(textAnnotationIterator.hasNext());
+ // -> this might be used to test that there are no TextAnnotations
int textAnnotationCount = 0;
while (textAnnotationIterator.hasNext()) {
UriRef textAnnotation = (UriRef)
textAnnotationIterator.next().getSubject();
@@ -85,16 +86,18 @@ public class EnhancementStructureHelper
Iterator<Triple> selectedTextIterator =
enhancements.filter(textAnnotation,
ENHANCER_SELECTED_TEXT, null);
// check if the selected text is added
- assertTrue("TextAnnotations MUST have a fise:selected-text
value",selectedTextIterator.hasNext());
+ assertTrue("TextAnnotations MUST have a fise:selected-text value (uri:
"+textAnnotation+")",
+ selectedTextIterator.hasNext());
// test if the selected text is part of the TEXT_TO_TEST
Resource selectedTextResource =
selectedTextIterator.next().getObject();
- assertTrue("fise:selected-text MUST BE of type
PlainLiteral",selectedTextResource instanceof PlainLiteral);
+ assertTrue("fise:selected-text MUST BE of type PlainLiteral (uri:
"+textAnnotation+")",
+ selectedTextResource instanceof PlainLiteral);
Literal selectedText = (Literal)selectedTextResource;
assertTrue("The parsed content MUST contain the fise:selected-text
value '"
-
+selectedText.getLexicalForm()+"'!",content.contains(selectedText.getLexicalForm()));
+ +selectedText.getLexicalForm()+"' (uri:
"+textAnnotation+")!",content.contains(selectedText.getLexicalForm()));
Resource expectedSelectedText =
expectedValues.get(ENHANCER_SELECTED_TEXT);
if(expectedSelectedText != null){
- assertEquals("The fise:selected-text is not the expected value
"+expectedSelectedText+"!",
+ assertEquals("The fise:selected-text is not the expected value
"+expectedSelectedText+" (uri: "+textAnnotation+")!",
expectedSelectedText, selectedText);
}
Resource selectionContextResource;
@@ -104,7 +107,8 @@ public class EnhancementStructureHelper
if(selectionContextIterator.hasNext()) { //context is optional
// test if the selected text is part of the TEXT_TO_TEST
selectionContextResource =
selectionContextIterator.next().getObject();
- assertTrue("The fise:selection-context MUST BE of type
PlainLiteral",selectionContextResource instanceof PlainLiteral);
+ assertTrue("The fise:selection-context MUST BE of type
PlainLiteral (uri: "+textAnnotation+")",
+ selectionContextResource instanceof PlainLiteral);
//check that the content contains the context
assertTrue("The fise:selection-context MUST BE contained in the
Content | context= "+ selectionContextResource,
content.contains(((Literal)selectionContextResource).getLexicalForm()));
@@ -129,26 +133,26 @@ public class EnhancementStructureHelper
TypedLiteral startPosLiteral;
TypedLiteral endPosLiteral;
if(startPosIterator.hasNext()){
- assertNotNull("If fise:start is present the fise:selection-context
MUST also be present!",
+ assertNotNull("If fise:start is present the fise:selection-context
MUST also be present (uri: "+textAnnotation+")!",
selectionContextResource);
Resource resource = startPosIterator.next().getObject();
//only a single start position is supported
- assertFalse("fise:start MUST HAVE only a single
value!",startPosIterator.hasNext());
- assertTrue("fise:start MUST be a typed Literal!",resource
instanceof TypedLiteral);
+ assertFalse("fise:start MUST HAVE only a single value (uri:
"+textAnnotation+")!",startPosIterator.hasNext());
+ assertTrue("fise:start MUST be a typed Literal (uri:
"+textAnnotation+")!",resource instanceof TypedLiteral);
startPosLiteral = (TypedLiteral) resource;
- assertEquals("fise:start MUST use xsd:int as data type",XSD.int_,
startPosLiteral.getDataType());
+ assertEquals("fise:start MUST use xsd:int as data type (uri:
"+textAnnotation+")",XSD.int_, startPosLiteral.getDataType());
resource = null;
Integer start =
LiteralFactory.getInstance().createObject(Integer.class, startPosLiteral);
assertNotNull("Unable to parse Integer from TypedLiteral
"+startPosLiteral,start);
//now get the end
//end must be defined if start is present
- assertTrue("If fise:start is present also fise:end MUST BE
defined!",endPosIterator.hasNext());
+ assertTrue("If fise:start is present also fise:end MUST BE defined
(uri: "+textAnnotation+")!",endPosIterator.hasNext());
resource = endPosIterator.next().getObject();
//only a single end position is supported
- assertFalse("fise:end MUST HAVE only a single
value!",endPosIterator.hasNext());
- assertTrue("fise:end values MUST BE TypedLiterals",resource
instanceof TypedLiteral);
+ assertFalse("fise:end MUST HAVE only a single value (uri:
"+textAnnotation+")!",endPosIterator.hasNext());
+ assertTrue("fise:end values MUST BE TypedLiterals (uri:
"+textAnnotation+")",resource instanceof TypedLiteral);
endPosLiteral = (TypedLiteral) resource;
- assertEquals("fise:end MUST use xsd:int as data type",XSD.int_,
endPosLiteral.getDataType());
+ assertEquals("fise:end MUST use xsd:int as data type (uri:
"+textAnnotation+")",XSD.int_, endPosLiteral.getDataType());
resource = null;
Integer end =
LiteralFactory.getInstance().createObject(Integer.class, endPosLiteral);
assertNotNull("Unable to parse Integer from TypedLiteral
"+endPosLiteral,end);
