You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@any23.apache.org by mo...@apache.org on 2012/02/05 14:01:04 UTC
svn commit: r1240713 - in /incubator/any23/trunk/core/src/test:
java/org/apache/any23/extractor/html/ java/org/apache/any23/extractor/rdfa/
resources/html/rdfa/
Author: mostarda
Date: Sun Feb 5 13:01:03 2012
New Revision: 1240713
URL: http://svn.apache.org/viewvc?rev=1240713&view=rev
Log:
Changed behavior of assertExtracts to verify also issues by default.
Added flag to disable verification. Removed custom verification logic
from RDFa11ExtractorTest. Refactored tests to be compliant with issue
verification.
Modified:
incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/AbstractExtractorTestCase.java
incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/LicenseExtractorTest.java
incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/RDFMergerTest.java
incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/RDFa11ExtractorTest.java
incubator/any23/trunk/core/src/test/resources/html/rdfa/basic.html
incubator/any23/trunk/core/src/test/resources/html/rdfa/incomplete-triples.html
incubator/any23/trunk/core/src/test/resources/html/rdfa/rdfa-11-curies.html
Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/AbstractExtractorTestCase.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/AbstractExtractorTestCase.java?rev=1240713&r1=1240712&r2=1240713&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/AbstractExtractorTestCase.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/AbstractExtractorTestCase.java Sun Feb 5 13:01:03 2012
@@ -144,6 +144,15 @@ public abstract class AbstractExtractorT
}
/**
+ * Returns the list of errors raised by the extractor under testing.
+ *
+ * @return collection of errors.
+ */
+ protected Collection<ErrorReporter.Error> getErrors() {
+ return getErrors( getExtractorFactory().getExtractorName() );
+ }
+
+ /**
* Applies the extractor provided by the {@link #getExtractorFactory()} to the specified resource.
*
* @param resource resource name.
@@ -163,13 +172,15 @@ public abstract class AbstractExtractorT
/**
* Performs data extraction over the content of a resource
- * and assert that the extraction was correct.
+ * and assert that the extraction was fine.
*
* @param resource resource name.
+ * @param assertNoIssues if <code>true</code>invokes {@link #assertNoIssues()} after the extraction.
*/
- protected void assertExtracts(String resource) {
+ protected void assertExtracts(String resource, boolean assertNoIssues) {
try {
extract(resource);
+ if(assertNoIssues) assertNoIssues();
} catch (ExtractionException ex) {
throw new RuntimeException(ex);
} catch (IOException ex) {
@@ -178,6 +189,16 @@ public abstract class AbstractExtractorT
}
/**
+ * Performs data extraction over the content of a resource
+ * and assert that the extraction was fine and raised no issues.
+ *
+ * @param resource
+ */
+ protected void assertExtracts(String resource) {
+ assertExtracts(resource, true);
+ }
+
+ /**
* Asserts that the extracted triples contain the pattern <code>(_ p o)</code>.
*
* @param p predicate
Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/LicenseExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/LicenseExtractorTest.java?rev=1240713&r1=1240712&r2=1240713&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/LicenseExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/LicenseExtractorTest.java Sun Feb 5 13:01:03 2012
@@ -17,14 +17,18 @@
package org.apache.any23.extractor.html;
+import org.apache.any23.extractor.ErrorReporter;
import org.apache.any23.extractor.ExtractorFactory;
import org.apache.any23.rdf.RDFUtils;
import org.apache.any23.vocab.SINDICE;
import org.apache.any23.vocab.XHTML;
+import org.junit.Assert;
import org.junit.Test;
import org.openrdf.model.URI;
import org.openrdf.repository.RepositoryException;
+import java.util.Collection;
+
/**
*
* Reference Test class for the {@link LicenseExtractor} extractor.
@@ -66,9 +70,15 @@ public class LicenseExtractorTest extend
@Test
public void testMultipleEmptyHref() throws RepositoryException {
- assertExtracts("microformats/license/multiple-empty-href.html");
+ assertExtracts("microformats/license/multiple-empty-href.html", false);
assertNotContains(baseURI, vXHTML.license, "");
assertContains(baseURI, vXHTML.license, apache);
+
+ final Collection<ErrorReporter.Error> errors = getErrors();
+ Assert.assertEquals(1, errors.size());
+ ErrorReporter.Error error = errors.iterator().next();
+ Assert.assertTrue(error.getMessage().contains("Invalid license link detected"));
+ Assert.assertEquals(ErrorReporter.ErrorLevel.WARN, error.getLevel());
}
@Test
Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/RDFMergerTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/RDFMergerTest.java?rev=1240713&r1=1240712&r2=1240713&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/RDFMergerTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/RDFMergerTest.java Sun Feb 5 13:01:03 2012
@@ -74,7 +74,7 @@ public class RDFMergerTest extends Abstr
@Test
public void test01XFNFoaf() throws RepositoryException {
- assertExtracts("mixed/01-xfn-foaf.html");
+ assertExtracts("mixed/01-xfn-foaf.html", false);
assertModelNotEmpty();
assertStatementsSize(RDF.TYPE, vVCARD.VCard, 1);
Resource vcard = findExactlyOneBlankSubject(RDF.TYPE, vVCARD.VCard);
Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/RDFa11ExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/RDFa11ExtractorTest.java?rev=1240713&r1=1240712&r2=1240713&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/RDFa11ExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/RDFa11ExtractorTest.java Sun Feb 5 13:01:03 2012
@@ -48,7 +48,7 @@ public class RDFa11ExtractorTest extends
*/
@Test
public void testObjectResourceConversion() throws RepositoryException {
- extractsAndCheckNoIssues("html/rdfa/object-resource-test.html");
+ assertExtracts("html/rdfa/object-resource-test.html");
logger.debug(dumpModelToTurtle());
assertContains(
null,
@@ -67,7 +67,7 @@ public class RDFa11ExtractorTest extends
*/
@Test
public void testExplicitDatatypeDeclaration() throws RepositoryException {
- extractsAndCheckNoIssues("html/rdfa/xmlliteral-datatype-test.html");
+ assertExtracts("html/rdfa/xmlliteral-datatype-test.html");
logger.debug(dumpModelToTurtle());
Literal literal = RDFUtils.literal(
@@ -88,7 +88,7 @@ public class RDFa11ExtractorTest extends
*/
@Test
public void testRelWithHref() throws RepositoryException {
- extractsAndCheckNoIssues("html/rdfa/rel-href.html");
+ assertExtracts("html/rdfa/rel-href.html");
logger.debug(dumpModelToTurtle());
assertContains(
@@ -110,7 +110,7 @@ public class RDFa11ExtractorTest extends
*/
@Test
public void testRelRevSupport() throws RepositoryException {
- extractsAndCheckNoIssues("html/rdfa/rel-rev.html");
+ assertExtracts("html/rdfa/rel-rev.html");
logger.debug(dumpModelToTurtle());
assertContains(
@@ -132,7 +132,7 @@ public class RDFa11ExtractorTest extends
*/
@Test
public void testVocabSupport() throws RepositoryException {
- extractsAndCheckNoIssues("html/rdfa/vocab.html");
+ assertExtracts("html/rdfa/vocab.html");
logger.debug(dumpModelToTurtle());
assertContains(
@@ -152,7 +152,7 @@ public class RDFa11ExtractorTest extends
*/
@Test
public void testTolerantParsing() {
- assertExtracts("html/rdfa/oreilly-invalid-datatype.html");
+ assertExtracts("html/rdfa/oreilly-invalid-datatype.html", false);
assertError(ErrorReporter.ErrorLevel.WARN, ".*Cannot map prefix \'mailto\'.*");
}
@@ -170,7 +170,7 @@ public class RDFa11ExtractorTest extends
throws RepositoryException, RDFHandlerException, IOException, RDFParseException {
final int EXPECTED_STATEMENTS = 33;
- extractsAndCheckNoIssues("html/rdfa/goodrelations-rdfa10.html");
+ assertExtracts("html/rdfa/goodrelations-rdfa10.html");
logger.debug(dumpModelToNQuads());
Assert.assertEquals(EXPECTED_STATEMENTS, dumpAsListOfStatements().size());
@@ -191,7 +191,7 @@ public class RDFa11ExtractorTest extends
throws RepositoryException, RDFHandlerException, IOException, RDFParseException {
final int EXPECTED_STATEMENTS = 33;
- extractsAndCheckNoIssues("html/rdfa/goodrelations-rdfa11.html");
+ assertExtracts("html/rdfa/goodrelations-rdfa11.html");
logger.debug(dumpHumanReadableTriples());
Assert.assertEquals(EXPECTED_STATEMENTS, dumpAsListOfStatements().size());
@@ -208,7 +208,7 @@ public class RDFa11ExtractorTest extends
*/
@Test
public void testOpenGraphStructuredProperties() throws IOException, ExtractionException, RepositoryException {
- extractsAndCheckNoIssues("html/rdfa/opengraph-structured-properties.html");
+ assertExtracts("html/rdfa/opengraph-structured-properties.html");
logger.info( dumpHumanReadableTriples() );
Assert.assertEquals(8, getStatementsSize(null, null, null) );
@@ -233,10 +233,5 @@ public class RDFa11ExtractorTest extends
protected ExtractorFactory<?> getExtractorFactory() {
return RDFa11Extractor.factory;
}
-
- private void extractsAndCheckNoIssues(String resource) {
- assertExtracts(resource);
- assertNoIssues();
- }
}
Modified: incubator/any23/trunk/core/src/test/resources/html/rdfa/basic.html
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/resources/html/rdfa/basic.html?rev=1240713&r1=1240712&r2=1240713&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/resources/html/rdfa/basic.html (original)
+++ incubator/any23/trunk/core/src/test/resources/html/rdfa/basic.html Sun Feb 5 13:01:03 2012
@@ -1,4 +1,4 @@
-<html xml:lang="en">
+<html xml:lang="en" xmlns="http://www.w3.org/1999/xhtml">
<head></head>
<body>
<div xmlns:dc="http://purl.org/dc/terms/" xmlns:fake="http://fake.org/">
Modified: incubator/any23/trunk/core/src/test/resources/html/rdfa/incomplete-triples.html
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/resources/html/rdfa/incomplete-triples.html?rev=1240713&r1=1240712&r2=1240713&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/resources/html/rdfa/incomplete-triples.html (original)
+++ incubator/any23/trunk/core/src/test/resources/html/rdfa/incomplete-triples.html Sun Feb 5 13:01:03 2012
@@ -1,4 +1,4 @@
-<html>
+<html xmlns="http://www.w3.org/1999/xhtml">
<head></head>
<body xmlns:dbp="http://dbpedia.org/property/" xmlns:foaf="http://xmlns.com/foaf/0.1/" xmlns:xsd="ttp://www.w3.org/2001/XMLSchema#">
<div about="http://dbpedia.org/resource/Albert_Einstein">
Modified: incubator/any23/trunk/core/src/test/resources/html/rdfa/rdfa-11-curies.html
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/resources/html/rdfa/rdfa-11-curies.html?rev=1240713&r1=1240712&r2=1240713&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/resources/html/rdfa/rdfa-11-curies.html (original)
+++ incubator/any23/trunk/core/src/test/resources/html/rdfa/rdfa-11-curies.html Sun Feb 5 13:01:03 2012
@@ -1,4 +1,4 @@
-<html>
+<html xmlns="http://www.w3.org/1999/xhtml">
<head>
</head>
<body>