[XERCESJ-909] XInclude processor treats malformed bytes sequences as resource errors.

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 2.6.2
• Component/s: XInclude 1.0
• Environment: Operating System: All; Platform: All
• Bugzilla Id: 27422
• Assignee: Michael Glavassevich
• Reporter: Michael Glavassevich
• Created: 04/Mar/04 07:20
• Updated: 23/Apr/04 17:42
• Resolved: 23/Apr/04 17:42

Description

While parsing a resource at an include location (both parse=xml and
parse=text), the XInclude processor is unable to detect whether the resource it
was parsing contained malformed byte sequences [1]. This is because the
IOException thrown by the Reader does not provide enough information to
determine the cause of the exception. It may very well have been something else
such as not being able to open an input stream. Instead of reporting a fatal
error, the processor treats it as a resource error and then looks for a
fallback. If a fallback exists the resulting stream will be malformed.

See Bug #27083, it's somewhat related to this problem. Although we cannot
change the behaviour of Java's readers, we should at least catch malformed byte
sequences with our own UTF8Reader and ASCIIReader and report them as with other
well-formedness violations.

[1] http://www.w3.org/TR/2004/REC-xml-20040204/#charencoding