210229 - Helper app Window for text/html document (two comma-delimited charsets in http content-type header)

Status: VERIFIED FIXED

(Core :: Networking: HTTP, defect, P2)

Description

[Matthias Versen [:Matti]]

win2k and todays CVS build and sirLurxalot confirmed it on Linux

a) open the URL
b) get a helper app window (see the attachment)

wget -S http://members.aol.com/elgrognard/successors/pbem1.htm
--02:09:52--  http://members.aol.com/elgrognard/successors/pbem1.htm
           => `pbem1.htm.2'
Resolving members.aol.com... done.
Connecting to members.aol.com[205.188.226.216]:80... connected.
HTTP request sent, awaiting response...
 1 HTTP/1.1 200 OK
 2 Date: Sun, 22 Jun 2003 00:09:53 GMT
 3 Server: Apache/1.3.26 (Unix) Resin/2.0.5
 4 Connection: close
 5 Content-Type: text/html; charset=Shift_JIS,ISO-8859-1


This is a regression because 1.3 displays it as HTML
(works as well in a 1 week old FB build from sirLurxalot )

Comment 1

[Matthias Versen [:Matti]]

Attachment: screenshot

Comment 2

[Boris Zbarsky [:bzbarsky]]

Sounds like a regression from those recent content-type parsing changes.... 
Confirming with a current CVS build on Linux.

Comment 3

[Boris Zbarsky [:bzbarsky]]

Oh, the MIME type we come out with for this site is "iso-8859-1", which is
indeed what the Content-Type header says....  Depending on how common this
particular server bug is we _may_ want to evangelize it....

Comment 4

[Bill Mason]

*** Bug 210921 has been marked as a duplicate of this bug. ***

Comment 5

[Bob Clary [:bc] (inactive)]

*** Bug 212714 has been marked as a duplicate of this bug. ***

Comment 6

[Matthias Versen [:Matti]]

*** Bug 213900 has been marked as a duplicate of this bug. ***

Comment 7

[Bill Mason]

*** Bug 214022 has been marked as a duplicate of this bug. ***

Comment 8

[Bill Mason]

Tweaking summary to try and help make finding this bug easier, as the dup's keep
coming in.

Comment 9

[Matthias Versen [:Matti]]

*** Bug 214277 has been marked as a duplicate of this bug. ***

Comment 10

[Michael Lefevre]

someone posting in the forums - http://forums.mozillazine.org/viewtopic.php?t=17752

reports a similar issue which I guess is caused by the same change.  The example
is a Netscape server with an even more screwed up content-type:

Server: NetWare-Enterprise-Web-Server/5.1
Date: Wed, 30 Jul 2003 16:07:09 GMT
Content-type: text/html;�l]ÕÀuÅÕE,DÕ`0Ñp.•Ò`0Ñô ÑÐ Ñ�l]ÕÀuÅÕ(DÕp.•Ò
Connection: close

and also results in the download/helper app dialog appearing.

Comment 11

[Darin Fisher]

this is probably related to my changes in nsHttpResponseHead::ParseContentType.

Comment 12

[Darin Fisher]

perhaps we can reject a "mime-type" if it does not contain a single '/' ??

Comment 13

[Boris Zbarsky [:bzbarsky]]

That seems like a reasonable course of action to me... I can't think of any
pages sending correct info that would get broken by that.  ;)

Comment 14

[Matthias Versen [:Matti]]

*** Bug 214536 has been marked as a duplicate of this bug. ***

Comment 15

[Darin Fisher]

Attachment: v1 patch

Comment 16

[Darin Fisher]

fixed-on-trunk, with this change:

  strchr(type, '/')

instead of

  strchr(type, "/")

which was just a typo ;-)

Comment 17

[Bill Mason]

*** Bug 215322 has been marked as a duplicate of this bug. ***

Comment 18

[Jesse Ruderman]

*** Bug 216611 has been marked as a duplicate of this bug. ***

Comment 19

[Dan Tobias]

Why was it parsing the "ISO-8859-1" as the content type, anyway?  It looks like
that is part of the "charset" attribute... a broken value (since there's no
logical reason to have two charsets separated by a comma; a document can only be
in one character encoding at a time), but not affecting the content type of
"text/html".

Somebody should evangelize AOL about their members.aol.com server returning a
broken header, anyway.

Comment 20

[Christian :Biesinger (don't email me, ping me on IRC)]

Dan: The HTTP Spec requires parsing that header just like mozilla did parse it.

Content-Type: foo,bar
is supposed to be treated exactly like:
Content-Type: foo
Content-Type: bar

Comment 21

[Simon Paquet]

*** Bug 218759 has been marked as a duplicate of this bug. ***

Comment 22

[Martha Palotay]

Just adding some of the things real people use to search for this bug:

  Firebird won't open page, keeps wanting to download it
  Unable to open iso-8859-1?
  Error about displaying ISO-8859-1 (does not know how to handle filetype ""
   (Null)
  won't view the url, instead shows the does not know how to handle this file 
   type dialog
  A window asking me where to save the page appears instead of rendering the
   page.
  Commas in Content-Type headers cause great confusion
  Messed up Content-Type header leads to problems
  HTTP Header includes two entries following CHARSET
  states that index.html is of type iso-8859-1 and that Mozilla Firebird does
   not know how to open this type of file.

(Because I spent over an hour searching through Bugzilla, but still ended up 
entering a duplicate of this bug.)

Comment 23

[Bob Firth]

*** Bug 219076 has been marked as a duplicate of this bug. ***

Comment 24

[Matthias Versen [:Matti]]

*** Bug 219158 has been marked as a duplicate of this bug. ***

Comment 25

[Bill Mason]

*** Bug 219224 has been marked as a duplicate of this bug. ***

Comment 26

[Bill Mason]

v.

Comment 27

[Dan Tobias]

Somebody should still evangelize AOL about changing their bogus header; even
though Mozilla has been "fixed" to do better error-correction on this, it still
causes problems.  All members.aol.com pages are now regarded by Mozilla as being
of charset Shift_JIS, a Japanese character set, which produces odd font choices
and misrendering of non-Japanese special characters (any use of iso-8859-1 or
windows-1252 characters causes Japanese characters to display instead).

Comment 28

[Martha Palotay]

The other problem is that the latest 'milestone' version of Firebird (the one 
that is recommended for new users who want a more stable version than the 
nightly releases) still has this bug. Until a newer milestone is released, 
duplicates of this will likely keep popping up.

(I've tried telling AOL about this, but it's almost impossible to find a 
customer support contact that actually knows what the heck an http header is.)

Comment 29

[Boris Zbarsky [:bzbarsky]]

Feel free to file an evangelism bug on the issue, but this bug was about the 
technical problem, which is resolved.

Comment 30

[Ronald Tilby]

The evangelism issue with AOL is the subject of bug 212714 which I just re-
opened.  (It was previously closed as a dup of this bug.)