Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 307847

Summary:	Illegal Content-Type computation if combined with custom charset
Product:	[RT] Jetty	Reporter:	Philipe Mulet <philippe_mulet>
Component:	server	Assignee:	Greg Wilkins <gregw>
Status:	RESOLVED FIXED	QA Contact:
Severity:	critical
Priority:	P3	CC:	jerome_lanneluc, jetty-inbox, simon_kaegi
Version:	unspecified
Target Milestone:	7.0.2.RC0
Hardware:	PC
OS:	Windows XP
Whiteboard:

Description Philipe Mulet

2010-04-01 06:55:05 EDT

This got discovered in v6.1.15

When setting a custom character encoding on a servlet response (org.mortbay.jetty.Response) for which the content type already contains a semicolon (but not at the end), it ends up producing an illegal content type, which isn't parsed by other libs.

e.g. set charset to UTF-8 on 'multipart/mixed; boundary="frontier"'
does produce: 'multipart/mixed; boundary="frontier" charset=UTF-8'

This seems to come from the following line in org.mortbay.jetty.Response :

_contentType = _contentType+" charset="+QuotedStringTokenizer.quote(_characterEncoding,";= "); 

when checking for presence of a semicolon, it doesn't consider whether the semicolon was in trailing position or not.

Comment 1 Greg Wilkins

2010-04-05 16:44:32 EDT

Fixed for 6.1.23   r5927

Comment 2 Greg Wilkins

2010-04-05 16:58:10 EDT

also fixed for 7.1.0 in r1466

Comment 3 Philipe Mulet

2010-04-09 04:07:05 EDT

Thanks for quick reaction.