Author: ab
Date: Thu Aug 5 19:34:30 2010
New Revision: 982745
URL: http://svn.apache.org/viewvc?rev=982745&view=rev
Log:
Undo a temporary fix introduced to accommodate parse-tika
(see NUTCH-794). We now use parse-html again, and the fix
is not needed.
Modified:
nutch/branches/branch-1.2/src/plugin/languageidentifier/src/test/org/apache/nutch/analysis/lang/TestHTMLLanguageParser.java
Modified:
nutch/branches/branch-1.2/src/plugin/languageidentifier/src/test/org/apache/nutch/analysis/lang/TestHTMLLanguageParser.java
URL:
http://svn.apache.org/viewvc/nutch/branches/branch-1.2/src/plugin/languageidentifier/src/test/org/apache/nutch/analysis/lang/TestHTMLLanguageParser.java?rev=982745&r1=982744&r2=982745&view=diff
==============================================================================
---
nutch/branches/branch-1.2/src/plugin/languageidentifier/src/test/org/apache/nutch/analysis/lang/TestHTMLLanguageParser.java
(original)
+++
nutch/branches/branch-1.2/src/plugin/languageidentifier/src/test/org/apache/nutch/analysis/lang/TestHTMLLanguageParser.java
Thu Aug 5 19:34:30 2010
@@ -41,7 +41,7 @@ public class TestHTMLLanguageParser exte
"<html><head><meta name=\"dc.language\" content=\"en\"><title>document 3
title</head><body>this is english</body></html>" };
// NUTCH-794 : temporarily replaced "fi" and "en" with null
- String metalanguages[] = { null, "en", "en" };
+ String metalanguages[] = { "fi", "en", "en" };
/**
* Test parsing of language identifiers from html