Author: mostarda
Date: Fri Jun 1 15:18:16 2012
New Revision: 1345230
URL: http://svn.apache.org/viewvc?rev=1345230&view=rev
Log:
Added IgnoreAccidentalRDFa and IgnoreTitlesOfEmptyDocuments filter in
WebResponder, added logic to prevent text/css link serialization in RDFa11.
This commit is related to issue #ANY23-95.
Modified:
incubator/any23/trunk/core/src/main/java/org/apache/any23/extractor/rdfa/RDFa11Parser.java
incubator/any23/trunk/core/src/test/java/org/apache/any23/Any23Test.java
incubator/any23/trunk/service/src/main/java/org/apache/any23/servlet/WebResponder.java
Modified:
incubator/any23/trunk/core/src/main/java/org/apache/any23/extractor/rdfa/RDFa11Parser.java
URL:
http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/main/java/org/apache/any23/extractor/rdfa/RDFa11Parser.java?rev=1345230&r1=1345229&r2=1345230&view=diff
==============================================================================
---
incubator/any23/trunk/core/src/main/java/org/apache/any23/extractor/rdfa/RDFa11Parser.java
(original)
+++
incubator/any23/trunk/core/src/main/java/org/apache/any23/extractor/rdfa/RDFa11Parser.java
Fri Jun 1 15:18:16 2012
@@ -73,6 +73,9 @@ public class RDFa11Parser {
public static final String SRC_ATTRIBUTE = "src";
public static final String HREF_ATTRIBUTE = "href";
+ public static final String TYPE_ATTRIBUTE = "type";
+ public static final String ATTRIBUTE_CSS = "text/css";
+
public static final String[] SUBJECT_ATTRIBUTES = {
ABOUT_ATTRIBUTE,
SRC_ATTRIBUTE,
@@ -165,6 +168,7 @@ public class RDFa11Parser {
}
protected static boolean isRelativeNode(Node node) {
+ if( ATTRIBUTE_CSS.equals( DomUtils.readAttribute(node, TYPE_ATTRIBUTE)
) ) return false;
return DomUtils.hasAttribute(node, REL_ATTRIBUTE) ||
DomUtils.hasAttribute(node, REV_ATTRIBUTE);
}
Modified:
incubator/any23/trunk/core/src/test/java/org/apache/any23/Any23Test.java
URL:
http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/Any23Test.java?rev=1345230&r1=1345229&r2=1345230&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/Any23Test.java
(original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/Any23Test.java
Fri Jun 1 15:18:16 2012
@@ -287,7 +287,7 @@ public class Any23Test extends Any23Onli
final String bufferContent = byteArrayOutputStream.toString();
logger.debug(bufferContent);
- Assert.assertSame("Unexpected number of triples.", 65,
StringUtils.countNL(bufferContent));
+ Assert.assertSame("Unexpected number of triples.", 60,
StringUtils.countNL(bufferContent));
}
Modified:
incubator/any23/trunk/service/src/main/java/org/apache/any23/servlet/WebResponder.java
URL:
http://svn.apache.org/viewvc/incubator/any23/trunk/service/src/main/java/org/apache/any23/servlet/WebResponder.java?rev=1345230&r1=1345229&r2=1345230&view=diff
==============================================================================
---
incubator/any23/trunk/service/src/main/java/org/apache/any23/servlet/WebResponder.java
(original)
+++
incubator/any23/trunk/service/src/main/java/org/apache/any23/servlet/WebResponder.java
Fri Jun 1 15:18:16 2012
@@ -24,6 +24,7 @@ import org.apache.any23.extractor.Extrac
import org.apache.any23.extractor.Extractor;
import org.apache.any23.extractor.IssueReport;
import org.apache.any23.filter.IgnoreAccidentalRDFa;
+import org.apache.any23.filter.IgnoreTitlesOfEmptyDocuments;
import org.apache.any23.source.DocumentSource;
import org.apache.any23.validator.SerializationException;
import org.apache.any23.validator.XMLValidationReportSerializer;
@@ -321,7 +322,10 @@ class WebResponder {
tripleHandlers.add(new CountingTripleHandler());
rdfWriter = new CompositeTripleHandler(tripleHandlers);
reporter = new ReportingTripleHandler(rdfWriter);
- rdfWriter = reporter;
+ rdfWriter = new IgnoreAccidentalRDFa(
+ new IgnoreTitlesOfEmptyDocuments(reporter),
+ true // suppress stylesheet triples.
+ );
return true;
}