[Archive-access-cvs] SF.net SVN: archive-access:[3591] trunk/archive-access/projects/ archive-commons

[<vin...@us...>]

Revision: 3591
          http://archive-access.svn.sourceforge.net/archive-access/?rev=3591&view=rev
Author:   vinaygoel
Date:     2012-01-08 03:09:20 +0000 (Sun, 08 Jan 2012)
Log Message:
-----------
Fixed CSS bug with URL too short (String out of bounds exception). Added basic test case for CSS extraction

Modified Paths:
--------------
    trunk/archive-access/projects/archive-commons/src/main/java/org/archive/resource/html/ExtractingParseObserver.java

Added Paths:
-----------
    trunk/archive-access/projects/archive-commons/src/test/java/org/archive/resource/html/ExtractingParseObserverTest.java

Property Changed:
----------------
    trunk/archive-access/projects/archive-commons/


Property changes on: trunk/archive-access/projects/archive-commons
___________________________________________________________________
Added: svn:ignore
   + target


Modified: trunk/archive-access/projects/archive-commons/src/main/java/org/archive/resource/html/ExtractingParseObserver.java
===================================================================
--- trunk/archive-access/projects/archive-commons/src/main/java/org/archive/resource/html/ExtractingParseObserver.java	2011-12-24 02:33:44 UTC (rev 3590)
+++ trunk/archive-access/projects/archive-commons/src/main/java/org/archive/resource/html/ExtractingParseObserver.java	2012-01-08 03:09:20 UTC (rev 3591)
@@ -378,6 +378,9 @@
 			int urlStart = m.start(1);
 			int urlEnd = m.end(1);
 			idx = urlEnd;
+			if(url.length() < 2) {
+				continue;
+			}
 			if ((url.charAt(0) == '(') 
 					&& (url.charAt(origUrlLength-1) == ')')) {
 				url = url.substring(1, origUrlLength - 1);

Added: trunk/archive-access/projects/archive-commons/src/test/java/org/archive/resource/html/ExtractingParseObserverTest.java
===================================================================
--- trunk/archive-access/projects/archive-commons/src/test/java/org/archive/resource/html/ExtractingParseObserverTest.java	                        (rev 0)
+++ trunk/archive-access/projects/archive-commons/src/test/java/org/archive/resource/html/ExtractingParseObserverTest.java	2012-01-08 03:09:20 UTC (rev 3591)
@@ -0,0 +1,98 @@
+package org.archive.resource.html;
+
+import org.archive.resource.MetaData;
+import org.htmlparser.nodes.TextNode;
+import org.json.JSONArray;
+import org.json.JSONException;
+import org.json.JSONObject;
+
+import junit.framework.TestCase;
+
+public class ExtractingParseObserverTest extends TestCase {
+
+	public void testHandleStyleNodeExceptions() throws Exception {
+		String[] tests = {
+				"some css",
+				"url()",
+				"url () ",
+				"url ('')",
+				"url (' ')",
+				"url('\")",
+				"url(')",
+				"url('\"')"
+		};
+		boolean except = false;
+		HTMLMetaData md = new HTMLMetaData(new MetaData());
+		ExtractingParseObserver epo = new ExtractingParseObserver(md);
+		for(String css : tests) {
+			try {
+				TextNode tn = new TextNode(css);
+				epo.handleStyleNode(tn);
+			} catch(Exception e) {
+				System.err.format("And the winner is....(%s)\n", css);
+				e.printStackTrace();
+				except = true;
+				throw e;
+			}
+			assertFalse(except);
+		}
+	}
+	public void testHandleStyleNode() throws Exception {
+		String[][] tests = {
+				{""},
+				{"url(foo.gif)","foo.gif"},
+				{"url('foo.gif')","foo.gif"},
+				{"url(\"foo.gif\")","foo.gif"},
+				{"url(\\\"foo.gif\\\")","foo.gif"},
+				{"url(\\'foo.gif\\')","foo.gif"},
+				
+		};
+		for(String[] testa : tests) {
+			checkExtract(testa);
+		}
+		//		boolean except = false;
+//		HTMLMetaData md = new HTMLMetaData(new MetaData());
+//		ExtractingParseObserver epo = new ExtractingParseObserver(md);
+//		for(String css : tests) {
+//			try {
+//				TextNode tn = new TextNode(css);
+//				epo.handleStyleNode(tn);
+//			} catch(Exception e) {
+//				System.err.format("And the winner is....(%s)\n", css);
+//				e.printStackTrace();
+//				except = true;
+//				throw e;
+//			}
+//			assertFalse(except);
+//		}
+	}
+	private void checkExtract(String[] data) throws JSONException {
+//		System.err.format("CSS(%s) want[0](%s)\n",css,want[0]);
+		String css = data[0];
+		boolean except = false;
+		HTMLMetaData md = new HTMLMetaData(new MetaData());
+		ExtractingParseObserver epo = new ExtractingParseObserver(md);
+		try {
+			TextNode tn = new TextNode(css);
+			epo.handleStyleNode(tn);
+		} catch(Exception e) {
+			fail("Exception with CSS:" + css);
+		}
+		JSONArray a = md.optJSONArray("Links");
+		if(data.length > 1) {
+			assertNotNull(a);
+			assertEquals(data.length-1,a.length());
+			for(int i = 1; i < data.length; i++) {
+				Object o = a.optJSONObject(i-1);
+				
+				assertTrue(o instanceof JSONObject);
+				JSONObject jo = (JSONObject) o;
+				assertEquals(data[i],jo.getString("href"));
+			}
+		} else {
+			assertNull(a);
+		}
+	}
+	
+
+}

This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.