incubator-any23-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From ans...@apache.org
Subject svn commit: r1371537 [2/30] - in /incubator/any23/trunk: ./ core/ core/src/test/java/org/apache/any23/ core/src/test/java/org/apache/any23/cli/ core/src/test/java/org/apache/any23/extractor/ core/src/test/java/org/apache/any23/extractor/csv/ core/src/t...
Date Fri, 10 Aug 2012 00:40:22 GMT
Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HCardExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HCardExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HCardExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HCardExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -49,7 +49,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testEMailNotUriReal() throws RepositoryException {
-        assertExtract("microformats/hcard/17-email-not-uri.html");
+        assertExtract("/microformats/hcard/17-email-not-uri.html");
         assertDefaultVCard();
         assertJohn();
         assertContains(vVCARD.email, RDFUtils.uri("mailto:john@example.com"));
@@ -57,7 +57,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testTel() throws RepositoryException {
-        assertExtract("microformats/hcard/21-tel.html");
+        assertExtract("/microformats/hcard/21-tel.html");
         assertDefaultVCard();
         String[] tels = {
                 "+1.415.555.1231", "+1.415.555.1235",
@@ -77,7 +77,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testAbbrTitleEverything() throws RepositoryException {
-        assertExtract("microformats/hcard/23-abbr-title-everything.html");
+        assertExtract("/microformats/hcard/23-abbr-title-everything.html");
         assertDefaultVCard();
 
         assertContains(vVCARD.fn, "John Doe");
@@ -122,7 +122,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testGeoAbbr() throws RepositoryException {
-        assertExtract("microformats/hcard/25-geo-abbr.html");
+        assertExtract("/microformats/hcard/25-geo-abbr.html");
         assertModelNotEmpty();
         assertContains(vVCARD.fn, "Paradise");
          assertContains(RDF.TYPE, vVCARD.Organization);
@@ -136,7 +136,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testAncestors() throws RepositoryException {
-        assertExtract("microformats/hcard/26-ancestors.html");
+        assertExtract("/microformats/hcard/26-ancestors.html");
         assertModelNotEmpty();
 
         assertContains(vVCARD.fn, "John Doe");
@@ -183,7 +183,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testfnOrg() throws RepositoryException {
-        assertExtract("microformats/hcard/30-fn-org.html");
+        assertExtract("/microformats/hcard/30-fn-org.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 4);
         RepositoryResult<Statement> repositoryResult = getStatements(null, RDF.TYPE, vVCARD.VCard);
@@ -209,7 +209,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testInclude() throws RepositoryException {
-        assertExtract("microformats/hcard/31-include.html");
+        assertExtract("/microformats/hcard/31-include.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 3);
         assertStatementsSize(vVCARD.email, (Value) null, 3);
@@ -248,7 +248,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testHeader() throws RepositoryException {
-        assertExtract("microformats/hcard/32-header.html");
+        assertExtract("/microformats/hcard/32-header.html");
         assertModelNotEmpty();
         // check fn, name, family, nick.
         assertJohn();
@@ -291,7 +291,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testAreaFull() throws RepositoryException {
-        assertExtract("microformats/hcard/33-area.html");
+        assertExtract("/microformats/hcard/33-area.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 5);
 
@@ -315,7 +315,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testCategories() throws RepositoryException {
-        assertExtract("microformats/hcard/36-categories.html");
+        assertExtract("/microformats/hcard/36-categories.html");
         assertModelNotEmpty();
         assertContains(vVCARD.given_name, "Joe");
         assertContains(vVCARD.given_name, "john");
@@ -340,7 +340,7 @@ public class HCardExtractorTest extends 
     @Test
     public void testSingleton() throws RepositoryException {
         // this tests probably tests that e just get the first fn and so on
-        assertExtract("microformats/hcard/37-singleton.html");
+        assertExtract("/microformats/hcard/37-singleton.html");
         assertModelNotEmpty();
         assertStatementsSize(vVCARD.fn, (Value) null, 1);
         assertContains(vVCARD.fn, "john doe 1");
@@ -375,7 +375,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testUidFull() throws RepositoryException {
-        assertExtract("microformats/hcard/38-uid.html");
+        assertExtract("/microformats/hcard/38-uid.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 4);
         RepositoryResult<Statement> statements = getStatements(null, RDF.TYPE, vVCARD.VCard);
@@ -411,7 +411,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testRomanianWikipedia() throws RepositoryException {
-        assertExtract("microformats/hcard/40-fn-inside-adr.html");
+        assertExtract("/microformats/hcard/40-fn-inside-adr.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 1);
         RepositoryResult<Statement> statements = getStatements(null, RDF.TYPE, vVCARD.VCard);
@@ -439,13 +439,13 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testNoMicroformats() throws RepositoryException, IOException, ExtractionException {
-        extract("html/html-without-uf.html");
+        extract("/html/html-without-uf.html");
         assertModelEmpty();
     }
 
     @Test
     public void testBasic() throws RepositoryException {
-        assertExtract("microformats/hcard/01-tantek-basic.html");
+        assertExtract("/microformats/hcard/01-tantek-basic.html");
         assertModelNotEmpty();
         assertContains(RDF.TYPE, vVCARD.VCard);
         // assertContains(RDF.TYPE, vVCARD.Organization);
@@ -472,7 +472,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testMultipleclassNamesOnVCard() throws RepositoryException {
-        assertExtract("microformats/hcard/02-multiple-class-names-on-vcard.html");
+        assertExtract("/microformats/hcard/02-multiple-class-names-on-vcard.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 4);
         Resource name;
@@ -531,7 +531,7 @@ public class HCardExtractorTest extends 
                 "Thomas"
         };
         List<String> NAMES = Arrays.asList(ns);
-        assertExtract("microformats/hcard/03-implied-n.html");
+        assertExtract("/microformats/hcard/03-implied-n.html");
         assertModelNotEmpty();
 
         RepositoryResult<Statement> statements = getStatements(null, vVCARD.fn, null);
@@ -564,7 +564,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testIgnoreUnknowns() throws RepositoryException {
-        assertExtract("microformats/hcard/04-ignore-unknowns.html");
+        assertExtract("/microformats/hcard/04-ignore-unknowns.html");
         assertDefaultVCard();
         assertContains(vVCARD.fn, "Ryan King");
         assertContains(vVCARD.n, (Resource) null);
@@ -575,7 +575,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testMailto1() throws RepositoryException {
-        assertExtract("microformats/hcard/05-mailto-1.html");
+        assertExtract("/microformats/hcard/05-mailto-1.html");
         assertDefaultVCard();
         assertContains(vVCARD.fn, "Ryan King");
         assertContains(RDF.TYPE, vVCARD.Name);
@@ -591,7 +591,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testMailto2() throws RepositoryException {
-        assertExtract("microformats/hcard/06-mailto-2.html");
+        assertExtract("/microformats/hcard/06-mailto-2.html");
         assertDefaultVCard();
         assertContains(vVCARD.fn, "Brian Suda");
 
@@ -605,7 +605,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testRelativeUrl() throws RepositoryException {
-        assertExtract("microformats/hcard/07-relative-url.html");
+        assertExtract("/microformats/hcard/07-relative-url.html");
         assertDefaultVCard();
         assertJohn();
         assertContains( vVCARD.url, RDFUtils.uri(baseURI + "home/blah") );
@@ -613,7 +613,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testRelativeUrlBase() throws RepositoryException {
-        assertExtract("microformats/hcard/08-relative-url-base.html");
+        assertExtract("/microformats/hcard/08-relative-url-base.html");
         assertDefaultVCard();
         assertContains(vVCARD.url, RDFUtils.uri(baseURI + "home/blah"));
         assertJohn();
@@ -621,7 +621,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testRelativeUrlXmlBase1() throws RepositoryException {
-        assertExtract("microformats/hcard/09-relative-url-xmlbase-1.html");
+        assertExtract("/microformats/hcard/09-relative-url-xmlbase-1.html");
         assertDefaultVCard();
         assertContains(vVCARD.url, RDFUtils.uri((baseURI + "home/blah")));
         assertJohn();
@@ -629,7 +629,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testRelativeUrlXmlBase2() throws RepositoryException {
-        assertExtract("microformats/hcard/10-relative-url-xmlbase-2.html");
+        assertExtract("/microformats/hcard/10-relative-url-xmlbase-2.html");
         assertDefaultVCard();
         assertContains(vVCARD.url, RDFUtils.uri((baseURI + "home/blah")));
         assertJohn();
@@ -637,7 +637,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testMultipleUrls() throws RepositoryException {
-        assertExtract("microformats/hcard/11-multiple-urls.html");
+        assertExtract("/microformats/hcard/11-multiple-urls.html");
         assertDefaultVCard();
         assertContains(vVCARD.url, RDFUtils.uri(("http://example.com/foo")));
         assertContains(vVCARD.url, RDFUtils.uri(("http://example.com/bar")));
@@ -647,14 +647,14 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testImageSrc() throws RepositoryException {
-        assertExtract("microformats/hcard/12-img-src-url.html");
+        assertExtract("/microformats/hcard/12-img-src-url.html");
         assertDefaultVCard();
         assertJohn();
     }
 
     @Test
     public void testPhotoLogo() throws RepositoryException {
-        assertExtract("microformats/hcard/13-photo-logo.html");
+        assertExtract("/microformats/hcard/13-photo-logo.html");
         assertDefaultVCard();
         assertContains(vVCARD.photo, RDFUtils.uri(("http://example.org/picture1.png")));
         assertContains(vVCARD.photo, RDFUtils.uri(("http://example.org/picture2.png")));
@@ -665,7 +665,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testImgSrcDataUrl() throws RepositoryException {
-        assertExtract("microformats/hcard/14-img-src-data-url.html");
+        assertExtract("/microformats/hcard/14-img-src-data-url.html");
         assertDefaultVCard();
         Resource data = RDFUtils.uri(
                           "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAADIAAAAyCAMAAAAp4XiDAAAABGdBTUEAAK/"
@@ -693,7 +693,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testHonorificAdditionalSingle() throws RepositoryException {
-        assertExtract("microformats/hcard/15-honorific-additional-single.html");
+        assertExtract("/microformats/hcard/15-honorific-additional-single.html");
         assertDefaultVCard();
         assertContains(vVCARD.fn, "Mr. John Maurice Doe, Ph.D.");
 
@@ -707,7 +707,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testHonorificAdditionalMultiple() throws RepositoryException {
-        assertExtract("microformats/hcard/16-honorific-additional-multiple.html");
+        assertExtract("/microformats/hcard/16-honorific-additional-multiple.html");
         assertDefaultVCard();
         assertContains(vVCARD.honorific_prefix, "Mr.");
         assertContains(vVCARD.honorific_prefix, "Dr.");
@@ -728,7 +728,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testEMailNotUri() throws RepositoryException {
-        assertExtract("microformats/hcard/17-email-not-uri.html");
+        assertExtract("/microformats/hcard/17-email-not-uri.html");
         assertDefaultVCard();
         assertJohn();
         assertContains( vVCARD.email, RDFUtils.uri("mailto:john@example.com") );
@@ -736,14 +736,14 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testObjectDataHttpUri() throws RepositoryException {
-        assertExtract("microformats/hcard/18-object-data-http-uri.html");
+        assertExtract("/microformats/hcard/18-object-data-http-uri.html");
         assertDefaultVCard();
         assertJohn();
     }
 
     @Test
     public void testObjectDataDataUri() throws RepositoryException {
-        assertExtract("microformats/hcard/19-object-data-data-uri.html");
+        assertExtract("/microformats/hcard/19-object-data-data-uri.html");
         assertDefaultVCard();
         assertJohn();
 
@@ -753,7 +753,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testImgAlt() throws RepositoryException {
-        assertExtract("microformats/hcard/20-image-alt.html");
+        assertExtract("/microformats/hcard/20-image-alt.html");
         assertDefaultVCard();
         Resource uri = RDFUtils.uri("http://example.com/foo.png");
         assertContains(vVCARD.photo, uri);
@@ -763,7 +763,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testAdr() throws RepositoryException {
-        assertExtract("microformats/hcard/22-adr.html");
+        assertExtract("/microformats/hcard/22-adr.html");
         assertDefaultVCard();
         assertJohn();
         assertStatementsSize(RDF.TYPE, vVCARD.Address, 0);
@@ -771,7 +771,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testBirthDayDate() throws RepositoryException {
-        assertExtract("microformats/hcard/27-bday-date.html");
+        assertExtract("/microformats/hcard/27-bday-date.html");
         assertModelNotEmpty();
         assertContains(vVCARD.fn         , "john doe");
         assertContains(vVCARD.given_name , "john");
@@ -781,7 +781,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testBirthDayDateTime() throws RepositoryException {
-        assertExtract("microformats/hcard/28-bday-datetime.html");
+        assertExtract("/microformats/hcard/28-bday-datetime.html");
         assertModelNotEmpty();
         assertContains(vVCARD.fn         , "john doe");
         assertContains(vVCARD.given_name , "john");
@@ -791,7 +791,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testBirthDayDateTimeTimeZone() throws RepositoryException {
-        assertExtract("microformats/hcard/29-bday-datetime-timezone.html");
+        assertExtract("/microformats/hcard/29-bday-datetime-timezone.html");
         assertModelNotEmpty();
         assertContains(vVCARD.fn, "john doe");
         assertContains(vVCARD.given_name, "john");
@@ -801,7 +801,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testArea() throws RepositoryException {
-        assertExtract("microformats/hcard/33-area.html");
+        assertExtract("/microformats/hcard/33-area.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 5);
         RepositoryResult<Statement> statements = getStatements(null, RDF.TYPE, vVCARD.VCard);
@@ -841,7 +841,7 @@ public class HCardExtractorTest extends 
     public void testNotes() throws RepositoryException {
         final String[] NOTES = {"Note 1", "Note 3", "Note 4 with a ; and a , to be escaped"};
 
-        assertExtract("microformats/hcard/34-notes.html");
+        assertExtract("/microformats/hcard/34-notes.html");
         assertModelNotEmpty();
         RepositoryResult<Statement> statements = getStatements(null, RDF.TYPE, vVCARD.VCard);
         try {
@@ -862,7 +862,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testIncludePattern() throws RepositoryException {
-        assertExtract("microformats/hcard/35-include-pattern.html");
+        assertExtract("/microformats/hcard/35-include-pattern.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 3);
 
@@ -907,7 +907,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testUid() throws RepositoryException {
-        assertExtract("microformats/hcard/38-uid.html");
+        assertExtract("/microformats/hcard/38-uid.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 4);
         RepositoryResult<Statement> iter = getStatements(null, RDF.TYPE, vVCARD.VCard);
@@ -928,7 +928,7 @@ public class HCardExtractorTest extends 
 
     @Test
     public void testIgnoreChildren() throws RepositoryException {
-        assertExtract("microformats/hcard/41-ignore-children.html");
+        assertExtract("/microformats/hcard/41-ignore-children.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 1);
         assertContains(vVCARD.fn, "Melanie Kl\u00f6\u00df");
@@ -948,7 +948,7 @@ public class HCardExtractorTest extends 
      */
     @Test
     public void testCumulativeHNames() throws RepositoryException {
-        assertExtract("microformats/hcard/linkedin-michelemostarda.html");
+        assertExtract("/microformats/hcard/linkedin-michelemostarda.html");
         assertModelNotEmpty();
         assertStatementsSize(vVCARD.given_name, "Michele"  , 7);
         assertStatementsSize(vVCARD.family_name, "Mostarda", 7);
@@ -963,7 +963,7 @@ public class HCardExtractorTest extends 
      */
     @Test
     public void testInfiniteLoop() throws IOException, ExtractionException {
-        assertExtract("microformats/hcard/infinite-loop.html", false);
+        assertExtract("/microformats/hcard/infinite-loop.html", false);
         assertIssue(IssueReport.IssueLevel.Warning, ".*Current node tries to include an ancestor node.*");
     }
 
@@ -974,7 +974,7 @@ public class HCardExtractorTest extends 
     @Ignore
     @Test(timeout = 30 * 1000)
     public void testExtractionPerformance() {
-        assertExtract("microformats/hcard/performance.html");
+        assertExtract("/microformats/hcard/performance.html");
     }
 
     private void assertDefaultVCard() throws RepositoryException {

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HListingExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HListingExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HListingExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HListingExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -51,27 +51,27 @@ public class HListingExtractorTest exten
 
     @Test
     public void testNoMicroformats() throws RepositoryException {
-        assertExtract("html/html-without-uf.html");
+        assertExtract("/html/html-without-uf.html");
         assertModelEmpty();
     }
 
     @Test
     public void testListingWithouthContent() throws RepositoryException {
-        assertExtract("microformats/hlisting/empty.html");
+        assertExtract("/microformats/hlisting/empty.html");
         assertModelNotEmpty();
         assertStatementsSize(null, null, null, 3);
     }
 
     @Test
     public void testSingleAction() throws RepositoryException {
-        assertExtract("microformats/hlisting/single-action.html");
+        assertExtract("/microformats/hlisting/single-action.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.action, vHLISTING.offer);
     }
 
     @Test
     public void testMultipleActions() throws RepositoryException {
-        assertExtract("microformats/hlisting/multiple-actions.html");
+        assertExtract("/microformats/hlisting/multiple-actions.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.action, vHLISTING.offer);
         assertContains(vHLISTING.action, vHLISTING.sell);
@@ -79,7 +79,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testMultipleActionsNested() throws RepositoryException {
-        assertExtract("microformats/hlisting/multiple-actions-nested.html");
+        assertExtract("/microformats/hlisting/multiple-actions-nested.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.action, vHLISTING.offer);
         assertContains(vHLISTING.action, vHLISTING.sell);
@@ -88,14 +88,14 @@ public class HListingExtractorTest exten
 
     @Test
     public void testActionsOutside() throws RepositoryException {
-        assertExtract("microformats/hlisting/single-action-outside.html");
+        assertExtract("/microformats/hlisting/single-action-outside.html");
         assertModelNotEmpty();
         assertNotContains(vHLISTING.action, vHLISTING.offer);
     }
 
     @Test
     public void testListerFn() throws RepositoryException {
-        assertExtract("microformats/hlisting/actions-lister-fn.html");
+        assertExtract("/microformats/hlisting/actions-lister-fn.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.action,     vHLISTING.offer);
         assertContains(RDF.TYPE,             vHLISTING.Lister);
@@ -104,7 +104,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testListerFnTel() throws RepositoryException {
-        assertExtract("microformats/hlisting/actions-lister-fn-tel.html");
+        assertExtract("/microformats/hlisting/actions-lister-fn-tel.html");
         assertModelNotEmpty();
 
         assertContains(vHLISTING.action    , vHLISTING.offer);
@@ -115,7 +115,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testItemFn() throws RepositoryException {
-        assertExtract("microformats/hlisting/item-fn.html");
+        assertExtract("/microformats/hlisting/item-fn.html");
         assertModelNotEmpty();
         assertContains(RDF.TYPE,           vHLISTING.Item);
         assertContains(vHLISTING.itemName, "Parking space");
@@ -123,7 +123,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testItemFnUrl() throws RepositoryException {
-        assertExtract("microformats/hlisting/item-fn-url.html");
+        assertExtract("/microformats/hlisting/item-fn-url.html");
         assertModelNotEmpty();
         assertContains(RDF.TYPE,             vHLISTING.Item);
         assertContains(vHLISTING.itemUrl,    RDFUtils.uri("http://item.com/"));
@@ -132,7 +132,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testItemPhotoImg() throws RepositoryException {
-        assertExtract("microformats/hlisting/item-fn-url-photo-img.html");
+        assertExtract("/microformats/hlisting/item-fn-url-photo-img.html");
         assertModelNotEmpty();
         assertContains(RDF.TYPE,             vHLISTING.Item);
         assertContains(vHLISTING.itemUrl,    RDFUtils.uri("http://item.com/"));
@@ -142,7 +142,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testItemPhotoHref() throws RepositoryException {
-        assertExtract("microformats/hlisting/item-fn-photo-href.html");
+        assertExtract("/microformats/hlisting/item-fn-photo-href.html");
         assertModelNotEmpty();
         assertContains(RDF.TYPE,             vHLISTING.Item);
         assertContains(vHLISTING.itemName,   "Parking space");
@@ -151,7 +151,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testKelkoo() throws RepositoryException {
-        assertExtract("microformats/hlisting/kelkoo.html");
+        assertExtract("/microformats/hlisting/kelkoo.html");
         assertModelNotEmpty();
 
         assertContains(RDF.TYPE,             vHLISTING.Listing);
@@ -180,7 +180,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testKelkooFull() throws RepositoryException {
-        assertExtract("microformats/hlisting/kelkoo-full.html");
+        assertExtract("/microformats/hlisting/kelkoo-full.html");
         assertModelNotEmpty();
         assertContains(RDF.TYPE,            vHLISTING.Listing);
         assertContains(RDF.TYPE,            vHLISTING.Item);
@@ -228,7 +228,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testListerURL() throws RepositoryException {
-        assertExtract("microformats/hlisting/actions-lister-url.html");
+        assertExtract("/microformats/hlisting/actions-lister-url.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.action,     vHLISTING.offer);
         assertContains(vHLISTING.listerName, "John Broker");
@@ -238,7 +238,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testListerEmail() throws RepositoryException {
-        assertExtract("microformats/hlisting/actions-lister-email.html");
+        assertExtract("/microformats/hlisting/actions-lister-email.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.action,     vHLISTING.offer);
         assertContains(vHLISTING.listerName, "John Broker");
@@ -248,7 +248,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testListerEmailHref() throws RepositoryException {
-        assertExtract("microformats/hlisting/actions-lister-email-href.html");
+        assertExtract("/microformats/hlisting/actions-lister-email-href.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.action,     vHLISTING.offer);
         assertContains(RDF.TYPE,             vHLISTING.Lister);
@@ -258,7 +258,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testDtListed() throws RepositoryException {
-        assertExtract("microformats/hlisting/dtlisted-dtexpired.html");
+        assertExtract("/microformats/hlisting/dtlisted-dtexpired.html");
         assertModelNotEmpty();
         assertNotContains(vHLISTING.action, vHLISTING.offer);
         assertContains(vHLISTING.dtlisted,  "2006-02-02");
@@ -266,7 +266,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testDtExpired() throws RepositoryException {
-        assertExtract("microformats/hlisting/dtlisted-dtexpired.html");
+        assertExtract("/microformats/hlisting/dtlisted-dtexpired.html");
         assertModelNotEmpty();
         assertNotContains(vHLISTING.action,  vHLISTING.offer);
         assertContains(vHLISTING.dtexpired, "2006-04-01");
@@ -274,14 +274,14 @@ public class HListingExtractorTest exten
 
     @Test
     public void testSummary() throws RepositoryException {
-        assertExtract("microformats/hlisting/summary.html");
+        assertExtract("/microformats/hlisting/summary.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.summary, "summary stuff");
     }
 
     @Test
     public void testDtListedAndExpired() throws RepositoryException {
-        assertExtract("microformats/hlisting/dtlisted-dtexpired.html");
+        assertExtract("/microformats/hlisting/dtlisted-dtexpired.html");
         assertModelNotEmpty();
         assertNotContains(vHLISTING.action,   vHLISTING.offer);
         assertContains(vHLISTING.dtlisted,   "2006-02-02");
@@ -290,14 +290,14 @@ public class HListingExtractorTest exten
 
     @Test
     public void testPrice() throws RepositoryException {
-        assertExtract("microformats/hlisting/price.html");
+        assertExtract("/microformats/hlisting/price.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.price,      "$215/qtr");
     }
 
     @Test
     public void testPriceAndDt() throws RepositoryException {
-        assertExtract("microformats/hlisting/dtlisted-dtexpired.html");
+        assertExtract("/microformats/hlisting/dtlisted-dtexpired.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.price,      "$215/qtr");
         assertContains(vHLISTING.dtlisted,   "2006-02-02");
@@ -306,7 +306,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testPermalink() throws RepositoryException {
-        assertExtract("microformats/hlisting/summary-bookmark.html");
+        assertExtract("/microformats/hlisting/summary-bookmark.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.permalink,  "http://livre.com/book");
         assertContains(vHLISTING.listerUrl,  RDFUtils.uri("http://livre.com/author"));
@@ -314,7 +314,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testComplexDescription() throws RepositoryException {
-        assertExtract("microformats/hlisting/description-complex.html");
+        assertExtract("/microformats/hlisting/description-complex.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.description,
                 "BenQ today introduced two new additions to its renowned bus... + Show details");
@@ -322,7 +322,7 @@ public class HListingExtractorTest exten
 
     @Test
     public void testDescription() throws RepositoryException {
-        assertExtract("microformats/hlisting/description.html");
+        assertExtract("/microformats/hlisting/description.html");
         assertModelNotEmpty();
         assertContains(vHLISTING.description,    "bla bla bla");
     }

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HRecipeExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HRecipeExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HRecipeExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HRecipeExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -41,13 +41,13 @@ public class HRecipeExtractorTest extend
 
     @Test
     public void testNoMicroformats() throws RepositoryException {
-        assertExtract("html/html-without-uf.html");
+        assertExtract("/html/html-without-uf.html");
         assertModelEmpty();
     }
 
     @Test
     public void testExtraction() throws RepositoryException {
-        assertExtract("microformats/hrecipe/01-spec.html");
+        assertExtract("/microformats/hrecipe/01-spec.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vHRECIPE.Recipe    , 1);
         assertStatementsSize(RDF.TYPE, vHRECIPE.Ingredient, 3);

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HResumeExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HResumeExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HResumeExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HResumeExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -56,13 +56,13 @@ public class HResumeExtractorTest extend
 
     @Test
     public void testNoMicroformats() throws RepositoryException {
-        assertExtract("html/html-without-uf.html");
+        assertExtract("/html/html-without-uf.html");
         assertModelEmpty();
     }
 
     @Test
     public void testLinkedIn() throws RepositoryException {
-        assertExtract("microformats/hresume/steveganz.html");
+        assertExtract("/microformats/hresume/steveganz.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vFOAF.Person, 1);
 
@@ -92,7 +92,7 @@ public class HResumeExtractorTest extend
     @Test
     public void testLinkedInComplete() throws RepositoryException {
 
-        assertExtract("microformats/hresume/steveganz.html");
+        assertExtract("/microformats/hresume/steveganz.html");
         assertModelNotEmpty();
 
         assertStatementsSize(RDF.TYPE, vFOAF.Person, 1);
@@ -139,7 +139,7 @@ public class HResumeExtractorTest extend
 
     @Test
     public void testAnt() throws RepositoryException {
-        assertExtract("microformats/hresume/ant.html");
+        assertExtract("/microformats/hresume/ant.html");
         assertModelNotEmpty();
 
         assertStatementsSize(RDF.TYPE, vFOAF.Person, 1);

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HReviewExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HReviewExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HReviewExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HReviewExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -53,13 +53,13 @@ public class HReviewExtractorTest extend
 
     @Test
     public void testNoMicroformats() throws RepositoryException {
-        assertExtract("html/html-without-uf.html");
+        assertExtract("/html/html-without-uf.html");
         assertModelEmpty();
     }
 
     @Test
     public void test01Basic() throws RepositoryException {
-        assertExtract("microformats/hreview/01-spec.html");
+        assertExtract("/microformats/hreview/01-spec.html");
         assertModelNotEmpty();
 
         assertStatementsSize(RDF.TYPE, vREVIEW.Review, 1);
@@ -110,7 +110,7 @@ public class HReviewExtractorTest extend
     @Test
     public void test02RatedTags() throws RepositoryException {
         
-        assertExtract("microformats/hreview/02-spec-2.html");
+        assertExtract("/microformats/hreview/02-spec-2.html");
         assertModelNotEmpty();
 
         assertStatementsSize(RDF.TYPE, vREVIEW.Review, 1);
@@ -166,7 +166,7 @@ public class HReviewExtractorTest extend
     @Test
     public void test03NoHcardForItem() throws RepositoryException {
 
-        assertExtract("microformats/hreview/03-spec-3.html");
+        assertExtract("/microformats/hreview/03-spec-3.html");
         assertModelNotEmpty();
 
         assertStatementsSize(RDF.TYPE, vREVIEW.Review, 1);
@@ -219,7 +219,7 @@ public class HReviewExtractorTest extend
     @Test
     public void test04NoHcardForItem() throws RepositoryException {
         
-        assertExtract("microformats/hreview/04-spec-4.html");
+        assertExtract("/microformats/hreview/04-spec-4.html");
         assertModelNotEmpty();
 
         assertStatementsSize(RDF.TYPE, vREVIEW.Review, 1);
@@ -278,7 +278,7 @@ public class HReviewExtractorTest extend
      */
     @Test
     public void testCaseSensitiveness() throws RepositoryException {
-        assertExtract("microformats/hreview/05-spec.html");
+        assertExtract("/microformats/hreview/05-spec.html");
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vREVIEW.Review, 1);
         // reviewer, no item

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HTMLFixture.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HTMLFixture.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HTMLFixture.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HTMLFixture.java Fri Aug 10 00:40:13 2012
@@ -19,6 +19,7 @@ package org.apache.any23.extractor.html;
 
 import org.apache.any23.source.DocumentSource;
 import org.apache.any23.source.FileDocumentSource;
+import org.junit.Assert;
 import org.w3c.dom.Node;
 
 import java.io.File;
@@ -30,17 +31,15 @@ import java.io.IOException;
  */
 public class HTMLFixture {
 
-    private final String filename;
+    private final File file;
 
-    public HTMLFixture(String filename) {
-        this.filename = filename;
+    public HTMLFixture(File file) {
+        Assert.assertNotNull("Test resource file was null", file);
+        Assert.assertTrue("Test resource file does not exist", file.exists());
+        this.file = file;
     }
 
     private File getFile() {
-        File file = new File(
-                System.getProperty("test.data", "src/test/resources/") + filename);
-        if (!file.exists())
-            throw new AssertionError("the file " + file.getPath() + " does not exist");
         return file;
     }
 

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HTMLMetaExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HTMLMetaExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HTMLMetaExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HTMLMetaExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -38,7 +38,7 @@ public class HTMLMetaExtractorTest exten
 
     @Test
     public void testExtractPageMeta() throws RepositoryException {
-        assertExtract("html/html-head-meta-extractor.html");
+        assertExtract("/html/html-head-meta-extractor.html");
         assertModelNotEmpty();
         assertStatementsSize(null, null, null, 7);
         assertContains(
@@ -87,7 +87,7 @@ public class HTMLMetaExtractorTest exten
 
     @Test
     public void testNoMeta() throws RepositoryException {
-        assertExtract("html/html-head-link-extractor.html");
+        assertExtract("/html/html-head-link-extractor.html");
         assertModelEmpty();
     }
     

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HeadLinkExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HeadLinkExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HeadLinkExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/HeadLinkExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -38,7 +38,7 @@ public class HeadLinkExtractorTest exten
 
     @Test
     public void testLinkExtraction() throws RepositoryException {
-        assertExtract("html/html-head-link-extractor.html");
+        assertExtract("/html/html-head-link-extractor.html");
         assertModelNotEmpty();
         final ValueFactory valueFactory = new ValueFactoryImpl();
         final URI externalLinkURI = valueFactory.createURI("http://www.myexperiment.org/workflows/16.rdf");

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=1371537&r1=1371536&r2=1371537&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 Fri Aug 10 00:40:13 2012
@@ -49,28 +49,28 @@ public class LicenseExtractorTest extend
 
     @Test
     public void testOnlyCc() throws RepositoryException {
-        assertExtract("microformats/license/ccBy.html");
+        assertExtract("/microformats/license/ccBy.html");
         assertContains(baseURI, vXHTML.license, ccBy);
         assertNotContains(baseURI, vXHTML.license, apache);
     }
 
     @Test
     public void testOnlyApache() throws RepositoryException {
-        assertExtract("microformats/license/apache.html");
+        assertExtract("/microformats/license/apache.html");
         assertNotContains(baseURI, vXHTML.license, ccBy);
         assertContains(baseURI, vXHTML.license, apache);
     }
 
     @Test
     public void testMultipleLicenses() throws RepositoryException {
-        assertExtract("microformats/license/multiple.html");
+        assertExtract("/microformats/license/multiple.html");
         assertContains(baseURI, vXHTML.license, ccBy);
         assertContains(baseURI, vXHTML.license, apache);
     }
 
     @Test
     public void testMultipleEmptyHref() throws RepositoryException {
-        assertExtract("microformats/license/multiple-empty-href.html", false);
+        assertExtract("/microformats/license/multiple-empty-href.html", false);
         assertNotContains(baseURI, vXHTML.license, "");
         assertContains(baseURI, vXHTML.license, apache);
         
@@ -83,13 +83,13 @@ public class LicenseExtractorTest extend
 
     @Test
     public void testEmpty() throws RepositoryException {
-        assertExtract("microformats/license/empty.html");
+        assertExtract("/microformats/license/empty.html");
         assertModelEmpty();
     }
 
     @Test
     public void testMixedCaseTitleTag() throws RepositoryException {
-        assertExtract("microformats/license/multiple-mixed-case.html");
+        assertExtract("/microformats/license/multiple-mixed-case.html");
         assertContains(baseURI, vXHTML.license, ccBy);
         assertContains(baseURI, vXHTML.license, apache);
     }

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=1371537&r1=1371536&r2=1371537&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 Fri Aug 10 00:40:13 2012
@@ -38,9 +38,11 @@ import org.openrdf.repository.Repository
 import org.openrdf.repository.RepositoryResult;
 import org.w3c.dom.Document;
 
+import java.io.BufferedInputStream;
 import java.io.File;
 import java.io.FileInputStream;
 import java.io.IOException;
+import java.io.InputStream;
 import java.util.HashMap;
 import java.util.Map;
 
@@ -68,13 +70,13 @@ public class RDFMergerTest extends Abstr
 
     @Test
     public void testNoMicroformats() throws RepositoryException, ExtractionException, IOException {
-        extract("html-without-uf.html");
+        extract("/html/html-without-uf.html");
         assertModelEmpty();
     }
 
     @Test
     public void test01XFNFoaf() throws RepositoryException {
-        assertExtract("mixed/01-xfn-foaf.html", false);
+        assertExtract("/html/mixed/01-xfn-foaf.html", false);
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vVCARD.VCard, 1);
         Resource vcard = findExactlyOneBlankSubject(RDF.TYPE, vVCARD.VCard);
@@ -96,7 +98,7 @@ public class RDFMergerTest extends Abstr
 
     @Test
     public void testAbbrTitleEverything() throws ExtractionException, IOException, RepositoryException {
-        extractHCardAndRelated("microformats/hcard/23-abbr-title-everything.html");
+        extractHCardAndRelated("/microformats/hcard/23-abbr-title-everything.html");
 
         assertContains(vVCARD.fn, "John Doe");
         assertContains(vVCARD.nickname, "JJ");
@@ -131,7 +133,7 @@ public class RDFMergerTest extends Abstr
 
     @Test
     public void testAdr() throws ExtractionException, IOException, RepositoryException {
-        extractHRevAndRelated("microformats/hcard/22-adr.html");
+        extractHRevAndRelated("/microformats/hcard/22-adr.html");
 
         assertStatementsSize(RDF.TYPE, vVCARD.Address, 4);
 
@@ -204,7 +206,7 @@ public class RDFMergerTest extends Abstr
 
     @Test
     public void testGeoAbbr() throws ExtractionException, IOException, RepositoryException {
-        extractHCardAndRelated("microformats/hcard/25-geo-abbr.html");
+        extractHCardAndRelated("/microformats/hcard/25-geo-abbr.html");
         assertModelNotEmpty();
         assertContains(vVCARD.fn, "Paradise");
         assertContains(RDF.TYPE, vVCARD.Organization);
@@ -217,7 +219,7 @@ public class RDFMergerTest extends Abstr
 
     @Test
     public void testAncestors() throws ExtractionException, IOException, RepositoryException {
-        extractHCardAndRelated("microformats/hcard/26-ancestors.html");
+        extractHCardAndRelated("/microformats/hcard/26-ancestors.html");
         assertModelNotEmpty();
         
         assertContains(vVCARD.fn, "John Doe");
@@ -260,7 +262,7 @@ public class RDFMergerTest extends Abstr
 
     @Test
     public void testSingleton() throws ExtractionException, IOException, RepositoryException {
-        extractHCardAndRelated("microformats/hcard/37-singleton.html");
+        extractHCardAndRelated("/microformats/hcard/37-singleton.html");
         assertModelNotEmpty();
         assertStatementsSize(vVCARD.fn, (Value) null, 1);
         assertContains(vVCARD.fn, "john doe 1");
@@ -291,7 +293,7 @@ public class RDFMergerTest extends Abstr
 
     @Test
     public void test01Basic() throws ExtractionException, IOException, RepositoryException {
-        extractHRevAndRelated("microformats/hreview/01-spec.html");
+        extractHRevAndRelated("/microformats/hreview/01-spec.html");
         assertModelNotEmpty();
 
         assertStatementsSize(RDF.TYPE, vREVIEW.Review, 1);
@@ -336,7 +338,7 @@ public class RDFMergerTest extends Abstr
     
     @Test
     public void test02RatedTags() throws ExtractionException, IOException, RepositoryException {
-        extractHRevAndRelated("microformats/hreview/02-spec-2.html");
+        extractHRevAndRelated("/microformats/hreview/02-spec-2.html");
 
         assertStatementsSize(vREVIEW.reviewer, (Value) null, 1);
         assertStatementsSize(vREVIEW.hasReview, (Value) null, 1);
@@ -372,7 +374,7 @@ public class RDFMergerTest extends Abstr
 
     @Test
     public void test03NoHcardForItem() throws ExtractionException, IOException, RepositoryException {
-        extractHRevAndRelated("microformats/hreview/03-spec-3.html");
+        extractHRevAndRelated("/microformats/hreview/03-spec-3.html");
 
         assertModelNotEmpty();
         assertStatementsSize(RDF.TYPE, vREVIEW.Review, 1);
@@ -431,11 +433,9 @@ public class RDFMergerTest extends Abstr
     @Override
     protected void extract(String filename) throws ExtractionException, IOException {
 
-        File file = new File(
-                System.getProperty("test.data", "src/test/resources") +
-                        "/html/" + filename);
-
-        Document document = new TagSoupParser(new FileInputStream(file), baseURI.stringValue()).getDOM();
+        InputStream input = new BufferedInputStream(this.getClass().getResourceAsStream(filename));
+        
+        Document document = new TagSoupParser(input, baseURI.stringValue()).getDOM();
         HCardExtractor hCardExtractor = HCardExtractor.factory.createExtractor();
         ExtractionContext hcExtractionContext = new ExtractionContext(
                 hCardExtractor.getDescription().getExtractorName(),
@@ -469,10 +469,10 @@ public class RDFMergerTest extends Abstr
     }
 
     private void extractHCardAndRelated(String filename) throws IOException, ExtractionException {
-        File file = new File(
-                System.getProperty("test.data", "src/test/resources/") + filename);
+        
+        InputStream input = new BufferedInputStream(this.getClass().getResourceAsStream(filename));
 
-        Document document = new TagSoupParser(new FileInputStream(file), baseURI.stringValue()).getDOM();
+        Document document = new TagSoupParser(input, baseURI.stringValue()).getDOM();
         HCardExtractor hCardExtractor = HCardExtractor.factory.createExtractor();
         ExtractionContext hCardExtractionContext = new ExtractionContext(
                 hCardExtractor.getDescription().getExtractorName(), baseURI
@@ -521,9 +521,8 @@ public class RDFMergerTest extends Abstr
 
     private void extractHRevAndRelated(String filename) throws ExtractionException, IOException {
         extractHCardAndRelated(filename);
-        File file = new File(
-                System.getProperty("test.data", "src/test/resources/") + filename);
-        Document document = new TagSoupParser(new FileInputStream(file), baseURI.stringValue()).getDOM();
+        InputStream input = new BufferedInputStream(this.getClass().getResourceAsStream(filename));
+        Document document = new TagSoupParser(input, baseURI.stringValue()).getDOM();
         HReviewExtractor hReviewExtractor = HReviewExtractor.factory.createExtractor();
         ExtractionContext hreviewExtractionContext = new ExtractionContext(
                 hReviewExtractor.getDescription().getExtractorName(), baseURI

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/SpeciesExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/SpeciesExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/SpeciesExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/SpeciesExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -47,7 +47,7 @@ public class SpeciesExtractorTest extend
      */
     @Test
     public void testSpeciesMicroformatExtractOverTaxoBox() throws RepositoryException {
-        assertExtract("microformats/species/species-example-2.html");
+        assertExtract("/microformats/species/species-example-2.html");
         assertModelNotEmpty();
         logger.debug(dumpModelToRDFXML());
 

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TagSoupParserTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TagSoupParserTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TagSoupParserTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TagSoupParserTest.java Fri Aug 10 00:40:13 2012
@@ -27,6 +27,7 @@ import org.w3c.dom.NamedNodeMap;
 import org.w3c.dom.Node;
 import org.w3c.dom.NodeList;
 
+import java.io.BufferedInputStream;
 import java.io.ByteArrayInputStream;
 import java.io.ByteArrayOutputStream;
 import java.io.File;
@@ -67,9 +68,7 @@ public class TagSoupParserTest {
     public void testExplicitEncodingBehavior()
     throws IOException, ExtractionException, RepositoryException {
         this.tagSoupParser = new TagSoupParser(
-                new FileInputStream(
-                    new File("src/test/resources/html/encoding-test.html")
-                ),
+                new BufferedInputStream(this.getClass().getResourceAsStream("/html/encoding-test.html")),
                 page,
                 "UTF-8"
         );
@@ -92,9 +91,7 @@ public class TagSoupParserTest {
     @Test
     public void testImplicitEncodingBehavior() throws IOException, ExtractionException, RepositoryException {
         this.tagSoupParser = new TagSoupParser(
-                new FileInputStream(
-                    new File("src/test/resources/html/encoding-test.html")
-                ),
+                new BufferedInputStream(this.getClass().getResourceAsStream("/html/encoding-test.html")),
                 page
         );
         Assert.assertNotSame(
@@ -113,12 +110,12 @@ public class TagSoupParserTest {
     @Test
     public void testEmptySpanElements() throws IOException {
         final String page = "http://example.com/test-page";
-        InputStream brokenEmptySpanHtml = new FileInputStream(
-                new File("src/test/resources/html/empty-span-broken.html")
-        );
-        InputStream worksEmptySpanHtml = new FileInputStream(
-                new File("src/test/resources/html/empty-span-works.html")
-        );
+        InputStream brokenEmptySpanHtml = 
+                new BufferedInputStream(this.getClass().getResourceAsStream("/html/empty-span-broken.html"))
+        ;
+        InputStream worksEmptySpanHtml = 
+                new BufferedInputStream(this.getClass().getResourceAsStream("/html/empty-span-works.html"))
+        ;
         this.tagSoupParser = new TagSoupParser(brokenEmptySpanHtml, page);
         Document brokenElementDom = this.tagSoupParser.getDOM();
         this.tagSoupParser = null; // useless but force GC

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TitleExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TitleExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TitleExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TitleExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -42,25 +42,25 @@ public class TitleExtractorTest extends 
 
     @Test
     public void testExtractPageTitle() throws RepositoryException {
-        assertExtract("microformats/xfn/simple-me.html");
+        assertExtract("/microformats/xfn/simple-me.html");
         assertContains(baseURI, vDCTERMS.title, helloLiteral);
     }
 
     @Test
     public void testStripSpacesFromTitle() throws RepositoryException {
-        assertExtract("microformats/xfn/strip-spaces.html");
+        assertExtract("/microformats/xfn/strip-spaces.html");
         assertContains(baseURI, vDCTERMS.title, helloLiteral);
     }
 
     @Test
     public void testNoPageTitle() throws RepositoryException {
-        assertExtract("microformats/xfn/tagsoup.html");
+        assertExtract("/microformats/xfn/tagsoup.html");
         assertModelEmpty();
     }
 
     @Test
     public void testMixedCaseTitleTag() throws RepositoryException {
-        assertExtract("microformats/xfn/mixed-case.html");
+        assertExtract("/microformats/xfn/mixed-case.html");
         assertContains(baseURI, vDCTERMS.title, helloLiteral);
     }
 
@@ -71,7 +71,7 @@ public class TitleExtractorTest extends 
      */
     @Test
     public void testTitleWithDefaultLanguage() throws RepositoryException {
-        assertExtract("html/default-language.html");
+        assertExtract("/html/default-language.html");
         assertContains   (baseURI, vDCTERMS.title, RDFUtils.literal("Welcome to mydomain.net", "en"));
         assertNotContains(baseURI, vDCTERMS.title, RDFUtils.literal("Welcome to mydomain.net",(String) null));
     }

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TurtleHTMLExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TurtleHTMLExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TurtleHTMLExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/TurtleHTMLExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -50,7 +50,7 @@ public class TurtleHTMLExtractorTest ext
      */
     @Test
     public void testExtraction() throws IOException, ExtractionException, RepositoryException {
-        assertExtract("html/html-turtle.html");
+        assertExtract("/html/html-turtle.html");
         logger.debug( dumpModelToRDFXML() );
         assertStatementsSize( null, (Value) null, 10);
     }

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/XFNExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/XFNExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/XFNExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/html/XFNExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -52,30 +52,30 @@ public class XFNExtractorTest extends Ab
 
     @Test
     public void testNoMicroformats() throws RepositoryException {
-        assertExtract("html/html-without-uf.html");
+        assertExtract("/html/html-without-uf.html");
         assertModelEmpty();
     }
 
     @Test
     public void testLinkWithoutRel() throws RepositoryException {
-        assertExtract("microformats/xfn/no-rel.html");
+        assertExtract("/microformats/xfn/no-rel.html");
         assertModelEmpty();
     }
 
     @Test
     public void testNoXFNRel() throws RepositoryException {
-        assertExtract("microformats/xfn/no-valid-rel.html");
+        assertExtract("/microformats/xfn/no-valid-rel.html");
         assertModelEmpty();
     }
 
     @Test
     public void testDetectPresenceOfXFN() throws RepositoryException {
-        assertExtract("microformats/xfn/simple-me.html");
+        assertExtract("/microformats/xfn/simple-me.html");
     }
 
     @Test
     public void testSimpleMeLink() throws RepositoryException {
-        assertExtract("microformats/xfn/simple-me.html");
+        assertExtract("/microformats/xfn/simple-me.html");
         Resource person = findExactlyOneBlankSubject(RDF.TYPE, vFOAF.Person);
         assertContains(person, vXFN.mePage, baseURI);
         assertContains(person, vXFN.mePage, bobsHomepage);
@@ -83,20 +83,20 @@ public class XFNExtractorTest extends Ab
 
     @Test
     public void testRelativeURIisResolvedAgainstBase() throws RepositoryException {
-        assertExtract("microformats/xfn/with-relative-uri.html");
+        assertExtract("/microformats/xfn/with-relative-uri.html");
         assertContains(null, vXFN.mePage, RDFUtils.uri("http://bob.example.com/foo"));
     }
 
     @Test
     public void testParseTagSoup() throws RepositoryException {
-        assertExtract("microformats/xfn/tagsoup.html");
+        assertExtract("/microformats/xfn/tagsoup.html");
         Resource person = findExactlyOneBlankSubject(RDF.TYPE, vFOAF.Person);
         assertContains(person, vXFN.mePage, baseURI);
     }
 
     @Test
     public void testSimpleFriend() throws RepositoryException {
-        assertExtract("microformats/xfn/simple-friend.html");
+        assertExtract("/microformats/xfn/simple-friend.html");
         Resource bob = findExactlyOneBlankSubject(vXFN.mePage, baseURI);
         Resource alice = findExactlyOneBlankSubject(vXFN.mePage, alicesHomepage);
         assertContains(bob, RDF.TYPE, vFOAF.Person);
@@ -108,7 +108,7 @@ public class XFNExtractorTest extends Ab
 
     @Test
     public void testFriendAndSweetheart() throws RepositoryException {
-        assertExtract("microformats/xfn/multiple-rel.html");
+        assertExtract("/microformats/xfn/multiple-rel.html");
         Resource bob = findExactlyOneBlankSubject(vXFN.mePage, baseURI);
         Resource alice = findExactlyOneBlankSubject(vXFN.mePage, alicesHomepage);
         assertContains(bob, vXFN.friend, alice);
@@ -120,7 +120,7 @@ public class XFNExtractorTest extends Ab
 
     @Test
     public void testMultipleFriends() throws RepositoryException {
-        assertExtract("microformats/xfn/multiple-friends.html");
+        assertExtract("/microformats/xfn/multiple-friends.html");
         Resource bob = findExactlyOneBlankSubject(vXFN.mePage, baseURI);
         Resource alice = findExactlyOneBlankSubject(vXFN.mePage, alicesHomepage);
         Resource charlie = findExactlyOneBlankSubject(vXFN.mePage, charliesHomepage);
@@ -133,31 +133,31 @@ public class XFNExtractorTest extends Ab
 
     @Test
     public void testSomeLinksWithoutRel() throws RepositoryException {
-        assertExtract("microformats/xfn/some-links-without-rel.html");
+        assertExtract("/microformats/xfn/some-links-without-rel.html");
         assertNotContains(null, null, alicesHomepage);
         assertContains   (null, null, charliesHomepage);
     }
 
     @Test
     public void testForSomeReasonICantBeMyOwnSweetheart() throws RepositoryException {
-        assertExtract("microformats/xfn/me-and-sweetheart.html");
+        assertExtract("/microformats/xfn/me-and-sweetheart.html");
         assertModelEmpty();
     }
 
     @Test
     public void testIgnoreExtraSpacesInRel() throws RepositoryException {
-        assertExtract("microformats/xfn/strip-spaces.html");
+        assertExtract("/microformats/xfn/strip-spaces.html");
         assertContains(null, vXFN.mePage, baseURI);
     }
 
     @Test
     public void testMixedCaseATag() throws RepositoryException {
-        assertExtract("microformats/xfn/mixed-case.html");
+        assertExtract("/microformats/xfn/mixed-case.html");
         assertContains(null, vXFN.mePage, baseURI);
     }
     @Test
     public void testUpcaseHREF() throws RepositoryException {
-        assertExtract("microformats/xfn/upcase-href.html");
+        assertExtract("/microformats/xfn/upcase-href.html");
         assertContains(null, vXFN.mePage, baseURI);
     }
 }

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/microdata/MicrodataExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/microdata/MicrodataExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/microdata/MicrodataExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/microdata/MicrodataExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -163,10 +163,10 @@ public class MicrodataExtractorTest exte
 
     private void extractAndVerifyAgainstNQuads(String actual, String expected)
     throws RepositoryException, RDFHandlerException, IOException, RDFParseException {
-        assertExtract("microdata/" + actual);
+        assertExtract("/microdata/" + actual);
         assertModelNotEmpty();
         logger.debug( dumpModelToNQuads() );
-        List<Statement> expectedStatements = loadResultStatement("microdata/" + expected);
+        List<Statement> expectedStatements = loadResultStatement("/microdata/" + expected);
         int actualStmtSize = getStatementsSize(null, null, null);
         Assert.assertEquals( expectedStatements.size(), actualStmtSize);
         for (Statement statement : expectedStatements) {
@@ -185,8 +185,7 @@ public class MicrodataExtractorTest exte
         NQuadsParser nQuadsParser = new NQuadsParser();
         TestRDFHandler rdfHandler = new TestRDFHandler();
         nQuadsParser.setRDFHandler(rdfHandler);
-        File file = new File(
-                System.getProperty("test.data", "src/test/resources/") + resultFilePath);
+        File file = copyResourceToTempFile(resultFilePath);
         nQuadsParser.parse(
                 new FileReader(file),
                 baseURI.toString()

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/AbstractRDFaExtractorTestCase.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/AbstractRDFaExtractorTestCase.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/AbstractRDFaExtractorTestCase.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/AbstractRDFaExtractorTestCase.java Fri Aug 10 00:40:13 2012
@@ -45,7 +45,7 @@ public abstract class AbstractRDFaExtrac
      */
     @Test
     public void testBasic() throws RepositoryException {
-        assertExtract("html/rdfa/basic.html");
+        assertExtract("/html/rdfa/basic.html");
         System.out.println( dumpModelToNQuads() );
         assertContains(null, vDCTERMS.creator, RDFUtils.literal("Alice", "en") );
         assertContains(null, vDCTERMS.title  , RDFUtils.literal("The trouble with Bob", "en") );
@@ -62,7 +62,7 @@ public abstract class AbstractRDFaExtrac
      */
     @Test
     public void testRDFa11CURIEs() throws RepositoryException {
-        assertExtract("html/rdfa/rdfa-11-curies.html");
+        assertExtract("/html/rdfa/rdfa-11-curies.html");
         assertModelNotEmpty();
         assertContains(
                 RDFUtils.uri("http://dbpedia.org/resource/Albert_Einstein"),
@@ -109,7 +109,7 @@ public abstract class AbstractRDFaExtrac
      */
     @Test
     public void testEmptyDatatypeDeclarationWithInnerXMLTags() throws RepositoryException {
-        assertExtract("html/rdfa/null-datatype-test.html");
+        assertExtract("/html/rdfa/null-datatype-test.html");
         logger.debug(dumpModelToRDFXML());
 
         assertContains(
@@ -129,7 +129,7 @@ public abstract class AbstractRDFaExtrac
      */
     @Test
     public void testDrupalTestPage() throws RepositoryException {
-        assertExtract("html/rdfa/drupal-test-frontpage.html");
+        assertExtract("/html/rdfa/drupal-test-frontpage.html");
         logger.debug(dumpModelToTurtle());
         assertContains(
                 RDFUtils.uri("http://bob.example.com/node/3"),
@@ -145,7 +145,7 @@ public abstract class AbstractRDFaExtrac
      */
     @Test
     public void testIncompleteTripleManagement() throws RepositoryException {
-        assertExtract("html/rdfa/incomplete-triples.html");
+        assertExtract("/html/rdfa/incomplete-triples.html");
         logger.debug(dumpModelToTurtle());
 
         assertContains(

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=1371537&r1=1371536&r2=1371537&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 Fri Aug 10 00:40:13 2012
@@ -48,7 +48,7 @@ public class RDFa11ExtractorTest extends
      */
     @Test
     public void testObjectResourceConversion() throws RepositoryException {
-        assertExtract("html/rdfa/object-resource-test.html");
+        assertExtract("/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 {
-        assertExtract("html/rdfa/xmlliteral-datatype-test.html");
+        assertExtract("/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 {
-        assertExtract("html/rdfa/rel-href.html");
+        assertExtract("/html/rdfa/rel-href.html");
         logger.debug(dumpModelToTurtle());
 
         assertContains(
@@ -110,7 +110,7 @@ public class RDFa11ExtractorTest extends
      */
     @Test
     public void testRelRevSupport() throws RepositoryException {
-        assertExtract("html/rdfa/rel-rev.html");
+        assertExtract("/html/rdfa/rel-rev.html");
         logger.debug(dumpModelToTurtle());
 
         assertContains(
@@ -132,7 +132,7 @@ public class RDFa11ExtractorTest extends
      */
     @Test
     public void testVocabSupport() throws RepositoryException {
-        assertExtract("html/rdfa/vocab.html");
+        assertExtract("/html/rdfa/vocab.html");
         logger.debug(dumpModelToTurtle());
 
         assertContains(
@@ -152,7 +152,7 @@ public class RDFa11ExtractorTest extends
      */
     @Test
     public void testTolerantParsing() {
-        assertExtract("html/rdfa/oreilly-invalid-datatype.html", false);
+        assertExtract("/html/rdfa/oreilly-invalid-datatype.html", false);
         assertIssue(IssueReport.IssueLevel.Warning, ".*Cannot map prefix \'mailto\'.*");
     }
 
@@ -170,7 +170,7 @@ public class RDFa11ExtractorTest extends
     throws RepositoryException, RDFHandlerException, IOException, RDFParseException {
         final int EXPECTED_STATEMENTS = 33;
 
-        assertExtract("html/rdfa/goodrelations-rdfa10.html");
+        assertExtract("/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;
 
-        assertExtract("html/rdfa/goodrelations-rdfa11.html");
+        assertExtract("/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 {
-        assertExtract("html/rdfa/opengraph-structured-properties.html");
+        assertExtract("/html/rdfa/opengraph-structured-properties.html");
         logger.info( dumpHumanReadableTriples() );
 
         Assert.assertEquals(8, getStatementsSize(null, null, null) );

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/RDFaExtractorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/RDFaExtractorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/RDFaExtractorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/extractor/rdfa/RDFaExtractorTest.java Fri Aug 10 00:40:13 2012
@@ -48,14 +48,14 @@ public class RDFaExtractorTest extends A
     throws RepositoryException, RDFHandlerException, IOException, RDFParseException {
         final int EXPECTED_STATEMENTS = 31;
 
-        assertExtract("html/rdfa/goodrelations-rdfa10.html");
+        assertExtract("/html/rdfa/goodrelations-rdfa10.html");
         logger.debug("Model 1 " + dumpHumanReadableTriples());
         Assert.assertEquals(EXPECTED_STATEMENTS, dumpAsListOfStatements().size());
         List<Statement> rdfa10Stmts = dumpAsListOfStatements();
 
         //assertContainsModel("/html/rdfa/goodrelations-rdfa10-expected.nq");
 
-        assertExtract("html/rdfa/goodrelations-rdfa11.html");
+        assertExtract("/html/rdfa/goodrelations-rdfa11.html");
         logger.debug("Model 2 " + dumpHumanReadableTriples());
         Assert.assertTrue(dumpAsListOfStatements().size() >= EXPECTED_STATEMENTS);
 
@@ -69,7 +69,7 @@ public class RDFaExtractorTest extends A
      */
     @Test
     public void testTolerantParsing() {
-        assertExtract("html/rdfa/oreilly-invalid-datatype.html");
+        assertExtract("/html/rdfa/oreilly-invalid-datatype.html");
     }
 
     @Override

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/mime/TikaMIMETypeDetectorTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/mime/TikaMIMETypeDetectorTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/mime/TikaMIMETypeDetectorTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/mime/TikaMIMETypeDetectorTest.java Fri Aug 10 00:40:13 2012
@@ -23,12 +23,16 @@ import org.junit.After;
 import org.junit.Before;
 import org.junit.Test;
 
+import java.io.BufferedInputStream;
 import java.io.ByteArrayInputStream;
 import java.io.ByteArrayOutputStream;
 import java.io.File;
 import java.io.FileInputStream;
 import java.io.IOException;
 import java.io.InputStream;
+import java.util.Arrays;
+import java.util.Collection;
+import java.util.List;
 
 /**
  * Test case for {@link TikaMIMETypeDetector} class.
@@ -115,67 +119,119 @@ public class TikaMIMETypeDetectorTest {
 
     @Test
     public void testDetectRSS1ByContent() throws Exception {
-        detectMIMEtypeByContent("application/rdf+xml", "src/test/resources/application/rss1");
+        detectMIMEtypeByContent("application/rdf+xml", manifestRss1());
+    }
+
+    private List<String> manifestRss1() {
+        return Arrays.asList("/application/rss1/test1");
     }
 
     @Test
     public void testDetectRSS2ByContent() throws Exception {
-        detectMIMEtypeByContent("application/rss+xml", "src/test/resources/application/rss2");
+        detectMIMEtypeByContent("application/rss+xml", manifestRss2());
+    }
+
+    private List<String> manifestRss2() {
+        return Arrays.asList("/application/rss2/index.html", "/application/rss2/rss2sample.xml", "/application/rss2/test1");
     }
 
     @Test
     public void testDetectRDFN3ByContent() throws Exception {
-        detectMIMEtypeByContent("text/n3", "src/test/resources/application/rdfn3");
+        detectMIMEtypeByContent("text/n3", manifestN3());
+    }
+
+    private List<String> manifestN3() {
+        return Arrays.asList("/application/rdfn3/test1", "/application/rdfn3/test2", "/application/rdfn3/test3");
     }
 
     @Test
     public void testDetectRDFNQuadsByContent() throws Exception {
-        detectMIMEtypeByContent("text/nq", "src/test/resources/application/nquads");
+        detectMIMEtypeByContent("text/nq", manifestNQuads());
+    }
+
+    private List<String> manifestNQuads() {
+        return Arrays.asList("/application/nquads/test1.nq", "/application/nquads/test2.nq");
     }
 
     @Test
     public void testDetectRDFXMLByContent() throws Exception {
-        detectMIMEtypeByContent("application/rdf+xml", "src/test/resources/application/rdfxml");
+        detectMIMEtypeByContent("application/rdf+xml", manifestRdfXml());
+    }
+
+    private List<String> manifestRdfXml() {
+        return Arrays.asList("/application/rdfxml/error.rdf", "/application/rdfxml/foaf", "/application/rdfxml/physics.owl", "/application/rdfxml/test1", "/application/rdfxml/test2", "/application/rdfxml/test3");
     }
 
     @Test
     public void testDetectTriXByContent() throws Exception {
-        detectMIMEtypeByContent("application/trix", "src/test/resources/application/trix");
+        detectMIMEtypeByContent("application/trix", manifestTrix());
+    }
+
+    private List<String> manifestTrix() {
+        return Arrays.asList("/application/trix/test1.trx");
     }
 
     @Test
     public void testDetectAtomByContent() throws Exception {
-        detectMIMEtypeByContent("application/atom+xml", "src/test/resources/application/atom");
+        detectMIMEtypeByContent("application/atom+xml", manifestAtom());
+    }
+
+    private List<String> manifestAtom() {
+        return Arrays.asList("/application/atom/atom.xml");
     }
 
     @Test
     public void testDetectHTMLByContent() throws Exception {
-        detectMIMEtypeByContent("text/html", "src/test/resources/text/html");
+        detectMIMEtypeByContent("text/html", manifestHtml());
+    }
+
+    private List<String> manifestHtml() {
+        return Arrays.asList("/text/html/test1");
     }
 
     @Test
     public void testDetectRDFaByContent() throws Exception {
-        detectMIMEtypeByContent("application/xhtml+xml", "src/test/resources/application/rdfa");
+        detectMIMEtypeByContent("application/xhtml+xml", manifestRdfa());
+    }
+
+    private List<String> manifestRdfa() {
+        return Arrays.asList("/application/rdfa/false.test", "/application/rdfa/london-gazette.html", "/application/rdfa/mic.xhtml", "/application/rdfa/test1.html");
     }
 
     @Test
     public void testDetectXHTMLByContent() throws Exception {
-        detectMIMEtypeByContent("application/xhtml+xml", "src/test/resources/application/xhtml");
+        detectMIMEtypeByContent("application/xhtml+xml", manifestXHtml());
+    }
+
+    private List<String> manifestXHtml() {
+        return Arrays.asList("/application/xhtml/blank-file-header.xhtml", "/application/xhtml/index.html", "/application/xhtml/test1");
     }
 
     @Test
     public void testDetectWSDLByContent() throws Exception {
-        detectMIMEtypeByContent("application/x-wsdl", "src/test/resources/application/wsdl");
+        detectMIMEtypeByContent("application/x-wsdl", manifestWsdl());
+    }
+
+    private List<String> manifestWsdl() {
+        return Arrays.asList("/application/wsdl/error.wsdl", "/application/wsdl/test1");
     }
 
     @Test
     public void testDetectZIPByContent() throws Exception {
-        detectMIMEtypeByContent("application/zip", "src/test/resources/application/zip");
+        detectMIMEtypeByContent("application/zip", manifestZip());
+    }
+
+    private List<String> manifestZip() {
+        return Arrays.asList("/application/zip/4_entries.zip", "/application/zip/test1.zip", "/application/zip/test2");
     }
 
     @Test
     public void testDetectCSVByContent() throws Exception {
-        detectMIMEtypeByContent("text/csv", "src/test/resources/org/apache/any23/extractor/csv/");
+        detectMIMEtypeByContent("text/csv", manifestCsv());
+    }
+
+    private List<String> manifestCsv() {
+        return Arrays.asList("/org/apache/any23/extractor/csv/test-comma.csv", "/org/apache/any23/extractor/csv/test-semicolon.csv", "/org/apache/any23/extractor/csv/test-tab.csv", "/org/apache/any23/extractor/csv/test-type.csv");
     }
 
     /* END: by content. */
@@ -258,67 +314,67 @@ public class TikaMIMETypeDetectorTest {
 
     @Test
     public void testRDFXMLByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("application/rdf+xml", "src/test/resources/application/rdfxml");
+        detectMIMETypeByContentAndName("application/rdf+xml", manifestRdfXml());
     }
 
     @Test
     public void testTriXByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("application/trix", "src/test/resources/application/trix");
+        detectMIMETypeByContentAndName("application/trix", manifestTrix());
     }
 
     @Test
     public void testRSS1ByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("application/rdf+xml", "src/test/resources/application/rss1");
+        detectMIMETypeByContentAndName("application/rdf+xml", manifestRss1());
     }
 
     @Test
     public void testRSS2ByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("application/rss+xml", "src/test/resources/application/rss2");
+        detectMIMETypeByContentAndName("application/rss+xml", manifestRss2());
     }
 
     @Test
     public void testDetectRDFN3ByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("text/n3", "src/test/resources/application/rdfn3");
+        detectMIMETypeByContentAndName("text/n3", manifestN3());
     }
 
     @Test
     public void testDetectRDFNQuadsByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("text/rdf+nq", "src/test/resources/application/nquads");
+        detectMIMETypeByContentAndName("text/rdf+nq", manifestNQuads());
     }
 
     @Test
     public void testAtomByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("application/atom+xml", "src/test/resources/application/atom");
+        detectMIMETypeByContentAndName("application/atom+xml", manifestAtom());
     }
 
     @Test
     public void testHTMLByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("text/html", "src/test/resources/text/html");
+        detectMIMETypeByContentAndName("text/html", manifestHtml());
     }
 
     @Test
     public void testXHTMLByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("application/xhtml+xml", "src/test/resources/application/xhtml");
+        detectMIMETypeByContentAndName("application/xhtml+xml", manifestXHtml());
     }
 
      @Test
     public void testWSDLByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("application/x-wsdl", "src/test/resources/application/wsdl");
+        detectMIMETypeByContentAndName("application/x-wsdl", manifestWsdl());
     }
 
     @Test
     public void testZipByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("application/zip", "src/test/resources/application/zip");
+        detectMIMETypeByContentAndName("application/zip", manifestZip());
     }
 
     @Test
     public void testRDFaByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("application/xhtml+xml", "src/test/resources/application/rdfa");
+        detectMIMETypeByContentAndName("application/xhtml+xml", manifestRdfa());
     }
 
     @Test
     public void testCSVByContentAndName() throws Exception {
-        detectMIMETypeByContentAndName("text/csv","src/test/resources/org/apache/any23/extractor/csv");
+        detectMIMETypeByContentAndName("text/csv", manifestCsv());
     }
 
     /* END: by content and name. */
@@ -350,23 +406,21 @@ public class TikaMIMETypeDetectorTest {
      * @param testDir the target file.
      * @throws IOException
      */
-    private void detectMIMEtypeByContent(String expectedMimeType, String testDir)
+    private void detectMIMEtypeByContent(String expectedMimeType, Collection<String> manifest)
     throws IOException {
-        File f = new File(testDir);
         String detectedMimeType;
-        for (File test : f.listFiles()) {
-            if (test.getName().startsWith(".")) continue;
-            InputStream is = getInputStream(test);
+        for (String test : manifest) {
+            InputStream is = new BufferedInputStream(this.getClass().getResourceAsStream(test));
             detectedMimeType = detector.guessMIMEType(
                     null,
                     is,
                     null
             ).toString();
-            if (test.getName().startsWith("error"))
+            if (test.contains("error"))
                 Assert.assertNotSame(expectedMimeType, detectedMimeType);
             else {
                 Assert.assertEquals(
-                        String.format("Error in mimetype detection for file %s", test.getAbsolutePath()),
+                        String.format("Error in mimetype detection for file %s", test),
                         expectedMimeType,
                         detectedMimeType
                 );
@@ -399,14 +453,12 @@ public class TikaMIMETypeDetectorTest {
      * @param testDir
      * @throws IOException
      */
-    private void detectMIMETypeByContentAndName(String expectedMimeType, String testDir) throws IOException {
-        File f = new File(testDir);
+    private void detectMIMETypeByContentAndName(String expectedMimeType, Collection<String> manifest) throws IOException {
         String detectedMimeType;
-        for (File test : f.listFiles()) {
-            if (test.getName().startsWith(".")) continue;
-            InputStream is = getInputStream(test);
-            detectedMimeType = detector.guessMIMEType(test.getName(), is, null).toString();
-            if (test.getName().startsWith("error"))
+        for (String test : manifest) {
+            InputStream is = new BufferedInputStream(this.getClass().getResourceAsStream(test));
+            detectedMimeType = detector.guessMIMEType(test, is, null).toString();
+            if (test.contains("error"))
                 Assert.assertNotSame(expectedMimeType, detectedMimeType);
             else {
                 Assert.assertEquals(
@@ -419,22 +471,4 @@ public class TikaMIMETypeDetectorTest {
         }
     }
 
-    /**
-     * @param file the file to be load.
-     * @return the input stream containing the file.
-     * @throws IOException
-     */
-    private InputStream getInputStream(File file) throws IOException {
-        FileInputStream fis = new FileInputStream(file);
-        ByteArrayOutputStream bos = new ByteArrayOutputStream();
-        byte[] buffer = new byte[4096];
-        while (fis.read(buffer) != -1) {
-            bos.write(buffer);
-        }
-        fis.close();
-        InputStream bais;
-        bais = new ByteArrayInputStream(bos.toByteArray());
-        return bais;
-    }
-    
 }

Modified: incubator/any23/trunk/core/src/test/java/org/apache/any23/mime/purifier/WhiteSpacesPurifierTest.java
URL: http://svn.apache.org/viewvc/incubator/any23/trunk/core/src/test/java/org/apache/any23/mime/purifier/WhiteSpacesPurifierTest.java?rev=1371537&r1=1371536&r2=1371537&view=diff
==============================================================================
--- incubator/any23/trunk/core/src/test/java/org/apache/any23/mime/purifier/WhiteSpacesPurifierTest.java (original)
+++ incubator/any23/trunk/core/src/test/java/org/apache/any23/mime/purifier/WhiteSpacesPurifierTest.java Fri Aug 10 00:40:13 2012
@@ -17,6 +17,7 @@
 
 package org.apache.any23.mime.purifier;
 
+import org.apache.tika.io.IOUtils;
 import org.junit.After;
 import org.junit.Assert;
 import org.junit.Before;
@@ -46,56 +47,18 @@ public class WhiteSpacesPurifierTest {
     @Test
     public void testPurification() throws IOException {
         InputStream inputStream =
-                getInputStream(new File("src/test/resources/application/xhtml/blank-file-header.xhtml"));
+                new BufferedInputStream(this.getClass().getResourceAsStream("/application/xhtml/blank-file-header.xhtml"));
         this.purifier.purify(inputStream);
         Assert.assertNotNull(inputStream);
         Assert.assertTrue(
                 validatePurification(
-                       readInputStreamAsString(inputStream)
+                       IOUtils.toString(inputStream)
                 )
         );
         
     }
 
     /**
-     * @param file the file to be load.
-     * @return the input stream containing the file.
-     * @throws java.io.IOException
-     */
-    private InputStream getInputStream(File file) throws IOException {
-        FileInputStream fis = new FileInputStream(file);
-        ByteArrayOutputStream bos = new ByteArrayOutputStream();
-        byte[] buffer = new byte[4096];
-        while (fis.read(buffer) != -1) {
-            bos.write(buffer);
-        }
-        fis.close();
-        InputStream bais;
-        bais = new ByteArrayInputStream(bos.toByteArray());
-        return bais;
-    }
-
-    /**
-     * Reads an {@link java.io.InputStream} as a {@link String}.
-     * 
-     * @param in
-     * @return
-     * @throws IOException
-     */
-    private String readInputStreamAsString(InputStream in)
-            throws IOException {
-        BufferedInputStream bis = new BufferedInputStream(in);
-        ByteArrayOutputStream buf = new ByteArrayOutputStream();
-        int result = bis.read();
-        while (result != -1) {
-            byte b = (byte) result;
-            buf.write(b);
-            result = bis.read();
-        }
-        return buf.toString();
-    }
-
-    /**
      * Checks if a {@link String} starts with a propert character.
      *  
      * @param string



Mime
View raw message