Bug 28654

Summary:

[Qt] MIME handling in qtwebkit network layer case-sensitivity

Product:

WebKit

Reporter:

Richard Moore <rich>

Component:

Page Loading

Assignee:

QtWebKit Unassigned <webkit-qt-unassigned>

Status:

RESOLVED FIXED

Severity:

Normal

CC:

abarth, commit-queue, eric, hausmann, jesus, kent.hansen, laszlo.gombos, luiz, markus, robert, tonikitoo, webkit.review.bot

Priority:

Keywords:

Qt, QtTriaged

Version:

528+ (Nightly build)

Hardware:

All

OS:

All

URL:

http://www.buienradar.nl/images.aspx?jaar=-3&bliksem=0&voor=&soort=verwachting1x1

Attachments:

Description	Flags
Patch to fix the bug	none
Patch	none

Richard Moore

Reported 2009-08-22 05:26:51 PDT

Created attachment 38432 [details] Patch to fix the bug The code in webkit assumes MIME types will be lower case, but QNetworkAccessManager returns the mime type exactly as it is sent by the remote server. The RFC says mime types are case-insensitive, so in the case of the URL above which sends image/GIF things should work just fine. Unfortunately that is not the case, qtwebkit believes this mime type to be unsupported wheras image/gif is fine. The fix is simply to lowercase the mimetype as it is passed from QNAM to webkit.

Attachments
Patch to fix the bug (804 bytes, patch) 2009-08-22 05:26 PDT, Richard Moore	no flags	Details Formatted Diff Diff
Patch (3.25 KB, patch) 2010-07-12 13:11 PDT, Robert Hogan	no flags	Details Formatted Diff Diff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.

Simon Hausmann

Comment 1 2009-11-23 05:56:26 PST

Richard, your patch needs a ChangeLog and a testcase/unit-test, and the patch should probably be marked for review :)

Richard Moore

Comment 2 2009-11-23 12:42:10 PST

Yeah, the difficulty here is that in order to avoid the test case being more complex than the issue itself we need a server. I was hoping that the server used by the QNAM performance tests would be available for testing qtwebkit stuff but that never happened. This leaves me with the webkit layout test server which i'm told supports php so should be doable. I don't have that setup right now, so this is waiting until I or someone else gets it setup and can write the test.

Simon Hausmann

Comment 3 2010-03-08 04:28:48 PST

One way of testing it is through a synthetic QNetworkAccessManager instance, like it's done in some of the unit tests, returning a fake http response.

Markus Goetz

Comment 4 2010-05-14 02:02:40 PDT

Check out tst_qnetworkreply the MiniHttpServer. You don't need a custom QNAM.

Robert Hogan

Comment 5 2010-07-12 13:11:58 PDT

Created attachment 61258 [details] Patch

Simon Hausmann

Comment 6 2010-07-12 13:34:41 PDT

Comment on attachment 61258 [details] Patch Excellent, thanks Robert! r=me. I hope it'll work with the other ports :)

WebKit Commit Bot

Comment 7 2010-07-13 11:56:52 PDT

Comment on attachment 61258 [details] Patch Clearing flags on attachment: 61258 Committed r63229: <http://trac.webkit.org/changeset/63229>

WebKit Commit Bot

Comment 8 2010-07-13 11:56:57 PDT

All reviewed patches have been landed. Closing bug.

WebKit Review Bot

Comment 9 2010-07-13 12:17:28 PDT

http://trac.webkit.org/changeset/63229 might have broken GTK Linux 32-bit Release The following changes are on the blame list: http://trac.webkit.org/changeset/63228 http://trac.webkit.org/changeset/63229

Note You need to log in before you can comment on or make changes to this bug.