Fix #11815 HttpServletMapping #11829
Conversation
Implemented a test that uses the examples from the javadoc Fixed the matchValue for PREFIX matches
gregw
added
the
Specification
|
I'm inclined not to port back to EE10 as nobody has complained and we passed the TCK. |
|
@janbartel @lachlan-roberts the CI failures are just the asciidoc stuff... so can I get a review please. |
…1.x/11815/HttpServletMapping
|
@janbartel @lachlan-roberts nudge! |
| _pathInfo = matchedPath != null | ||
| ? matchedPath.getPathInfo() | ||
| : _servletPath.length() == pathInContext.length() ? null : pathInContext.substring(_servletPath.length()); | ||
| _matchValue = _pathInfo == null ? "" : _pathInfo.startsWith("/") ? _pathInfo.substring(1) : _pathInfo; |
There was a problem hiding this comment.
Why is this using _pathInfo and not pathInContext?
There was a problem hiding this comment.
The spec requires that for in a context "/ctx", with a pattern "/foo/*" then for a request to "/ctx/foo/info", the contextPath will be "/ctx", the servletPath will be "/foo" and the pathInfo "/info". We had previously implemented that the match value was based on the servletPath (so "foo"), but it needs to be based on the part that matched the *, i.e. the pathInfo without the leading slash (so "info").
The path in context is "/foo/info", so that is servletPath + pathInfo and is not what we want for the matchValue.
Implemented a test that uses the examples from the javadoc Fixed the matchValue for PREFIX matches