Closed
Bug 187868
Opened 22 years ago
Closed 17 years ago
<segment>/../ is converted to <segment>/ in absolute URIs
Categories
(Core :: Networking, defect)
Core
Networking
Tracking
()
RESOLVED
WONTFIX
Future
People
(Reporter: dsmutil, Unassigned)
References
()
Details
(Keywords: verifyme)
User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.3a) Gecko/20021212
Build Identifier: Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.3a) Gecko/20021212
In an absolute URL, such as the one listed above, the "/.." is parsed out of the
path as if it were a relative link. RFC 2396 says that "." and ".." only have a
special meaning when "interpreting a relative path" and not in an absolute path.
As I understand this, if we're at http://www.google.com/images/index.html and
have an image at ../logo.gif it should go to http://www.google.com/logo.gif but
if the image was at http://www.google.com/images/../logo.gif then that's the
path that should be sent. (See section 5.2 of this RFC.)
Reproducible: Always
Steps to Reproduce:
This was prompted in part by bug 87501 and bug 187845.
Comment 1•22 years ago
|
||
Yes, correct, but we always mormalize urls to make them compareable and that
includes collapsing the path by removing .. segments as much as possible, so
http://www.google.com/images/../logo.gif ends up as
http://www.google.com/logo.gif on the client side. It should make no difference
on the page/image that gets returned but helps compare urls.
Reporter | ||
Comment 2•22 years ago
|
||
That makes since, but ".." doesn't need to mean the parent folder to the server.
I've never seen a web server that uses ".." as something other then the parent
folder but it could be used that way.
Comment 3•22 years ago
|
||
dan: do you have a real testcase for this bug? what do other browsers do?
Reporter | ||
Comment 4•22 years ago
|
||
I did some experiments with IE 5.5 and Opera 7 beta demo (I'm not sure what else
to try; I only use Mozilla). As a test, I created a web page with a single link
to http://www.pusd.org/it/../public_index.asp . The page also worked when I sent
the entire path to the server (with the "..") via Telnet to port 80.
MSIE 5.5 - Showed the entire path in the address bar when typed (including the
".." in the path) but it actually requested the page with the truncated path.
Opera 7 beta - Shortened the path like Mozilla does and then requested the
shorter path.
Comment 6•18 years ago
|
||
*** This bug has been marked as a duplicate of 51169 ***
Status: NEW → RESOLVED
Closed: 18 years ago
Resolution: --- → DUPLICATE
Reporter | ||
Comment 7•18 years ago
|
||
Not the same bug. Bug 51169 deals with relative URIs that include a scheme; this bug is for an absolute URI that includes "/../" in the path.
Status: RESOLVED → REOPENED
Resolution: DUPLICATE → ---
SUMMARY:
Dan's point about the strict reading of the use of ".." is probably correct.
However, mozilla implements the URL passing as Andreas described, and having tested a lot of the URL fixes he wrote, I doubt that anyone is going to come up with a better solution.
Also, the concern was theoretical. Nobody has come back with an example (or a crazy webserver/file system) that wanted ".." to mean anything else.
Status: NEW → RESOLVED
Closed: 18 years ago → 17 years ago
Keywords: verifyme
Resolution: --- → WONTFIX
You need to log in
before you can comment on or make changes to this bug.
Description
•