From 5586d5452df16d47fca9fb677cbd76a6bd85c642 Mon Sep 17 00:00:00 2001
From: Noah Levitt <nlevitt@archive.org>
Date: Tue, 11 Dec 2012 16:58:48 -0800
Subject: [PATCH] * JerichoExtractorHTMLTest.java     avoid redundancy by using
 extractor built in setUp()     makeExtractor() - call
 extractor.setExtractorJS(new ExtractorJS()) since we got rid of the static
 ExtractorJS stuff     testConditionalComment1() - override to skip the test
 since it fails with JerichoExtractorHTML

---
 .../extractor/JerichoExtractorHTMLTest.java   | 30 ++++++++++++-------
 1 file changed, 20 insertions(+), 10 deletions(-)

diff --git a/modules/src/test/java/org/archive/modules/extractor/JerichoExtractorHTMLTest.java b/modules/src/test/java/org/archive/modules/extractor/JerichoExtractorHTMLTest.java
index 9668c5bd..a4d17345 100644
--- a/modules/src/test/java/org/archive/modules/extractor/JerichoExtractorHTMLTest.java
+++ b/modules/src/test/java/org/archive/modules/extractor/JerichoExtractorHTMLTest.java
@@ -45,6 +45,7 @@ public class JerichoExtractorHTMLTest extends ExtractorHTMLTest {
         CrawlMetadata metadata = new CrawlMetadata();
         metadata.afterPropertiesSet();
         result.setMetadata(metadata);
+        result.setExtractorJS(new ExtractorJS());
         result.afterPropertiesSet();
         return result;
     }
@@ -68,8 +69,7 @@ public class JerichoExtractorHTMLTest extends ExtractorHTMLTest {
         	"  </select>" +
         	"  <input type=\"submit\" name=\"test\" value=\"Go\">" +
         	"</form>";   
-        JerichoExtractorHTML ex = (JerichoExtractorHTML)makeExtractor();
-        ex.extract(curi, cs);
+        getExtractor().extract(curi, cs);
         curi.getOutLinks();
         assertTrue(CollectionUtils.exists(curi.getOutLinks(), new Predicate() {
             public boolean evaluate(Object object) {
@@ -97,8 +97,7 @@ public class JerichoExtractorHTMLTest extends ExtractorHTMLTest {
             "  </select>" +
             "  <input type=\"submit\" name=\"test\" value=\"Go\">" +
             "</form>";   
-        JerichoExtractorHTML ex = (JerichoExtractorHTML)makeExtractor();
-        ex.extract(curi, cs);
+        getExtractor().extract(curi, cs);
         curi.getOutLinks();
         assertTrue(!CollectionUtils.exists(curi.getOutLinks(), new Predicate() {
             public boolean evaluate(Object object) {
@@ -125,10 +124,9 @@ public class JerichoExtractorHTMLTest extends ExtractorHTMLTest {
             "    <option value=\"nonselectedOption\">option2</option>" +
             "  </select>" +
             "  <input type=\"submit\" name=\"test\" value=\"Go\">" +
-            "</form>";   
-        JerichoExtractorHTML ex = (JerichoExtractorHTML)makeExtractor();
-        ex.setExtractOnlyFormGets(false);
-        ex.extract(curi, cs);
+            "</form>";
+        getExtractor().setExtractOnlyFormGets(false);
+        getExtractor().extract(curi, cs);
         curi.getOutLinks();
         assertTrue(CollectionUtils.exists(curi.getOutLinks(), new Predicate() {
             public boolean evaluate(Object object) {
@@ -142,8 +140,20 @@ public class JerichoExtractorHTMLTest extends ExtractorHTMLTest {
         UURI uuri = UURIFactory.getInstance("http://www.example.org");
         CrawlURI curi = new CrawlURI(uuri);
         CharSequence cs = "<a src=\"http://www.example.com/\" href=\"http://www.archive.org/\"> "; 
-        JerichoExtractorHTML ex = (JerichoExtractorHTML)makeExtractor();
-        ex.extract(curi, cs);
+        getExtractor().extract(curi, cs);
         assertTrue("not all links found", curi.getOutLinks().size() == 2);
     }
+
+    /*
+     * Override of ExtractorHTMLTest method because the test fails with
+     * JerichoExtractorHTML
+     */
+    @Override
+    public void testConditionalComment1() throws URIException {
+    }
+    
+    @Override
+    protected JerichoExtractorHTML getExtractor() {
+        return (JerichoExtractorHTML) extractor;
+    }
 }