Applying fix for XALANJ-2205, which gets Xalan out of the business of parsing base URIs of stylesheets and include/import href values provided that:
1) The user has provided a URIResolver
2) The Source returned by a URIResolver has its system ID set.
If the system ID is not set, or there is no URI resolver the behavior is as before, namely that Xalan-J will call its SystemIDResolver class to figure out how to get a base URI for the included stylesheet. This does fine for most users, but has issues if the URIs are not valid. One way to be invalid is to have non-ASCII characters in the URI. For example, suppose one is running on an operating system that allows non-ASCII characters in its directories or file names (e.g. Chinese characters). To use URIs such as:
file:///dir1/dirXX2/part.xsl
where XX are some Chinese characters, then the user must provide 1) and 2) above.
2 files changed
