Menu
▾
▴
[Archive-access-cvs] SF.net SVN: archive-access: [2115] trunk/archive-access/projects/wayback/ wayback-core/src
|
From: <bra...@us...> - 2007-12-15 02:00:47
|
Revision: 2115
http://archive-access.svn.sourceforge.net/archive-access/?rev=2115&view=rev
Author: bradtofel
Date: 2007-12-14 18:00:51 -0800 (Fri, 14 Dec 2007)
Log Message:
-----------
BUGFIX: (unreported) regex was not finding simple tags (ex: "<head>")...
Modified Paths:
--------------
trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/replay/TagMagix.java
trunk/archive-access/projects/wayback/wayback-core/src/test/java/org/archive/wayback/replay/TagMagixTest.java
Modified: trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/replay/TagMagix.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/replay/TagMagix.java 2007-12-12 03:34:13 UTC (rev 2114)
+++ trunk/archive-access/projects/wayback/wayback-core/src/main/java/org/archive/wayback/replay/TagMagix.java 2007-12-15 02:00:51 UTC (rev 2115)
@@ -100,7 +100,7 @@
Pattern pc = wholeTagPatterns.get(tagName);
if (pc == null) {
- String tagPatString = "<\\s*" + tagName + "\\s+[^>]+>";
+ String tagPatString = "<\\s*" + tagName + "((>)|(\\s+[^>]*>))";
pc = Pattern.compile(tagPatString, Pattern.CASE_INSENSITIVE);
wholeTagPatterns.put(tagName, pc);
Modified: trunk/archive-access/projects/wayback/wayback-core/src/test/java/org/archive/wayback/replay/TagMagixTest.java
===================================================================
--- trunk/archive-access/projects/wayback/wayback-core/src/test/java/org/archive/wayback/replay/TagMagixTest.java 2007-12-12 03:34:13 UTC (rev 2114)
+++ trunk/archive-access/projects/wayback/wayback-core/src/test/java/org/archive/wayback/replay/TagMagixTest.java 2007-12-15 02:00:51 UTC (rev 2115)
@@ -98,7 +98,24 @@
"author2","Bakri Abubakr http://bayanit.com/");
}
+ public void testFindEndOfFirst() {
+ findEndOf("<head>","head",6);
+ findEndOf("<html><head><body>","head",12);
+ findEndOf("<html><head goo=bar><body>","head",20);
+ findEndOf("<html><head goo=bar><body>full","body",26);
+ findEndOf("<html><head goo=bar><body >full","body",27);
+ findEndOf("<html><head goo=bar><body >full","body",27);
+ findEndOf("<html><head goo=bar><body yar=bam>full","body",34);
+ findEndOf("<html><head goo=bar><body yar='bam'>full","body",36);
+ findEndOf("<html><head goo=bar><body yar=\"bam\">full","body",36);
+ }
+ public void findEndOf(String page, String tag, int offset) {
+ StringBuilder sb = new StringBuilder(page);
+ int found = TagMagix.getEndOfFirstTag(sb,tag);
+ assertEquals("FAILED find end of " +tag+ " in ("+page+")",offset,found);
+ }
+
/**
* Test method for 'org.archive.wayback.archivalurl.TagMagix.markupTag(StringBuffer, String, String, String, String, String)'
*/
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
