Details
-
Type:
Bug
-
Status:
Closed
(View Workflow)
-
Priority:
Major
-
Resolution: Done
-
Affects Version/s: None
-
Fix Version/s: tomcat_1.0.0.CR3
-
Component/s: Tomcat Containers
-
Security Level: Public (Everyone can see)
-
Labels:None
-
Environment:Win 7 32-bit (Locale: en_IN, file.encoding = Cp1252), Apache Tomcat 7.0.14
-
Forum Reference:
-
Steps to Reproduce:
-
Git Pull Request:
-
Similar Issues:
Description
Tomcat decodes the Authorization HTTP header containing the HTTP BASIC authentication credentials, using the ISO-8859-1 charset. On the contrary, Arquillian uses Charset.defaultCharset to encode the supplied credentials, before Base64 encoding the resulting byte array. This can result in a 401 error, when accessing the manager application on Tomcat; the failure will occur when the credentials contain characters that are encoded differently in ISO-8859-1 and the default charset of the environment. For example, if the default charset is UTF-8, then credentials containing ö will not be accepted by Tomcat as ö is encoded as 0xC3 0xB6 in UTF-8 and 0xF6 in ISO-8859-1.
The suggested fix is to use ISO-8859-1 in CommonTomcatManager.constructHttpBasicAuthHeader() as listed:
// Set up an authorization header with our credentials
String credentials = configuration.getUser() + ":" + configuration.getPass();
return "Basic "
+ new String(Base64.encodeBase64(credentials.getBytes(Charset.forName("ISO-8859-1"))), Charset.defaultCharset());
instead of the existing
// Set up an authorization header with our credentials
String credentials = configuration.getUser() + ":" + configuration.getPass();
return "Basic "
+ new String(Base64.encodeBase64(credentials.getBytes(Charset.defaultCharset())), Charset.defaultCharset());
Issue Links
- depends on
-
SHRINKDESC-97
XmlDomNodeDescriptorImporterImpl.from(String) uses the platform encoding to parse the String argument resulting in XML parser exceptions when parsing the XML Document
-
Activity
Pull request submitted. An improvement of the proposed fix in JIRA is present in the pull request.
Unit tests have also been amended to use ISO-8859-1 credentials in arquillian.xml for testing; credentials in ASCII will continue to work.
This issue depends on SHRINKDESC-97. Until the parent issue is fixed, the workaround mentioned in SHRINKDESC-97 can be used to ensure that Arquillian does not fail to parse the arquillian.xml file.
pushed upstream
Added the steps to reproduce the bug.