[jira] Updated: (WICKET-2061) interceptContinuationURL with umlauts not encoded

["Sven Meier (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/WICKET-2061?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sven Meier updated WICKET-2061:
-------------------------------

    Description: 
This is my second try to fix Wicket's encoding of redirects to intercepted URLs.
Instead of reopening WICKET-2007, I decided to create this new issue and make a clean start.

When Wicket redirects to an intercept page, it stores the original URL in PageMap#interceptContinuationURL.
		// The intercept continuation URL should be saved exactly as the
		// original request specified.
		...
			interceptContinuationURL = "/" + cycle.getRequest().getURL();

Note that comment and code are not in sync:
Instead of saving *exactly* the original request, a new URL is generated. ServletWebRequest#getURL() includes special characters non-encoded. Thus on a later continuation, the redirect to the original URL fails in case of special characters (umlauts in our case).

We're now using a subclass of ServletWebRequest, which utilizes the requestURI:

	public String getURL()
	{
		// servletPath is de-encoded, so use requestURI minus contextPath instead
		// String url = getServletPath();
		String url = httpServletRequest.getRequestURI().substring(
				httpServletRequest.getContextPath().length());

		final String pathInfo = getHttpServletRequest().getPathInfo();
		...
	}

I cannot say if this solution has unwanted side effects though.

  was:
This is my second try to fix Wicket's encoding of redirects to intercepted URLs.
Instead of reopening WICKET-2007, I decided to create this new issue and make a clean start.

When Wicket redirects to an intercept page, it stores the original URL in PageMap#interceptContinuationURL.
		// The intercept continuation URL should be saved exactly as the
		// original request specified.
		...
			interceptContinuationURL = "/" + cycle.getRequest().getURL();

Note that comment and code are not in sync:
Instead of saving *exactly* the original request, a new URL is generated. ServletWebRequest#getURL() does *not* encode special characters. Thus on a later continuation, the redirect to the original URL fails in case of special characters (umlauts in our case).

The following patch would save *almost* the original request (it has to remove the context path because RedirectRequestTarget prepends it later on).
I cannot say if this solution has unwanted side effects though:

Index: src/main/java/org/apache/wicket/PageMap.java
===================================================================
RCS file: /boss/dev/src/main/java/org/apache/wicket/PageMap.java,v
retrieving revision 1.1
diff -u -r1.1 PageMap.java
--- src/main/java/org/apache/wicket/PageMap.java	30 Jan 2009 12:54:41 -0000	1.1
+++ src/main/java/org/apache/wicket/PageMap.java	30 Jan 2009 12:55:02 -0000
@@ -262,7 +262,10 @@
 		}
 		else
 		{
-			interceptContinuationURL = "/" + cycle.getRequest().getURL();
+			WebRequest request = (WebRequest) cycle.getRequest();
+			String uri = request.getHttpServletRequest().getRequestURI();
+			String contextPath = request.getHttpServletRequest().getContextPath();
+			interceptContinuationURL = uri.substring(contextPath.length());
 		}
 
Note that the encoding done in WebResponse#redirect(String) appends a session id only, it doesn't handle special characters (e.g. umlauts).



> interceptContinuationURL with umlauts not encoded
> -------------------------------------------------
>
>                 Key: WICKET-2061
>                 URL: https://issues.apache.org/jira/browse/WICKET-2061
>             Project: Wicket
>          Issue Type: Bug
>          Components: wicket
>    Affects Versions: 1.3.5, 1.3.6, 1.4-RC1
>            Reporter: Sven Meier
>         Attachments: encodingtest.zip
>
>
> This is my second try to fix Wicket's encoding of redirects to intercepted URLs.
> Instead of reopening WICKET-2007, I decided to create this new issue and make a clean start.
> When Wicket redirects to an intercept page, it stores the original URL in PageMap#interceptContinuationURL.
> 		// The intercept continuation URL should be saved exactly as the
> 		// original request specified.
> 		...
> 			interceptContinuationURL = "/" + cycle.getRequest().getURL();
> Note that comment and code are not in sync:
> Instead of saving *exactly* the original request, a new URL is generated. ServletWebRequest#getURL() includes special characters non-encoded. Thus on a later continuation, the redirect to the original URL fails in case of special characters (umlauts in our case).
> We're now using a subclass of ServletWebRequest, which utilizes the requestURI:
> 	public String getURL()
> 	{
> 		// servletPath is de-encoded, so use requestURI minus contextPath instead
> 		// String url = getServletPath();
> 		String url = httpServletRequest.getRequestURI().substring(
> 				httpServletRequest.getContextPath().length());
> 		final String pathInfo = getHttpServletRequest().getPathInfo();
> 		...
> 	}
> I cannot say if this solution has unwanted side effects though.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.