Hi,
I met a problem when I used MultipartScanner#scan() at Grizzly 2.3.8 or
2.3.9.
--
MultipartScanner.java:85
84: final String contentType = request.getContentType(); //
HttpHeader#getContentType()
85: String[] contentTypeParams = contentType.split(";"); // NPE because of
contentType was null.
--
When I debugged the
Requests(org.glassfish.grizzly.http.server.Request.java), the object already
contained the content-type header(both key and value) in parameters field
and raw HttpRequestPacket object, but maybe HttpHeader#getContentType()
returned a wrong value(null).
I think that the content-type had the chance which could be set at
ServerHttpRequestImp#parseContentTypeHeader() but contentType.isSet() was
true because contentType.isMimeTypeSet() was true.
--
In HttpServerFilter.java
..
private void parseContentTypeHeader() {
contentTypeParsed = true;
if (!contentType.isSet()) {
..
// the logic of setting content type
}
}
--
In ContentType.java
..
public boolean isMimeTypeSet() {
return !isParsed || mimeType != null; // isParsed is default true.
}
--
When I tested the same logic at Grizzly 2.3.7, I didn't meet this issue.
Perhaps, could this be the side-effect of Issue
#1032(https://java.net/jira/browse/GRIZZLY-1032)? :-)
Regards,
Bongjae Chang