[plt-scheme] Cookies to web servlet getting lost when __utmz cookie present

From: Todd O'Bryan (toddobryan at gmail.com)
Date: Mon May 3 22:09:50 EDT 2010

Nadeem,

You might try get-cookie from the net/cookie module.

#lang scheme
(require net/cookie)
(require web-server/http/request-structs)

(get-cookie "teaching-order" (headers-assq* #"cookie"
(request-headers/raw req)))

It looks like it works for your cookie, but it doesn't do so well with Google's.

While I sympathize with Jay's desire to be true to the spec, it seems
like there should be some way to deal with evil, non-compliant
cookies, especially given that you have no control over them.

Maybe use request-cookies first and then, if that ends up empty, check
to see if there's a Cookie header and try a more forgiving parsing
algorithm like the one Jay provided. If you come up with something
that doesn't feel terribly hacky, please share!

Todd

On Fri, Apr 30, 2010 at 12:33 PM, Jay McCarthy <jay.mccarthy at gmail.com> wrote:
> On Fri, Apr 30, 2010 at 10:12 AM, Nadeem Abdul Hamid <nadeem at acm.org> wrote:
>> Thanks, Jay.
>>
>> That's kind of annoying though -- I mean I can't do anything about the
>> cookies that Google requests the browser to set when other pages on
>> the domain are visited. And I tried different browsers -- the same
>> problem. Why would Google Analytics be setting invalid cookies? (A
>> search doesn't seem to turn up any relevant hits, or anyone else every
>> experience this issue.)
>
> I don't know why they write buggy programs. I've submitted a bug
> report the last time this exact problem came up.
>
>> So basically, I need to parse the headers for cookies in my own program?
>
> It's not that bad. Here's something that does what Ruby does for
> parsing cookies. [Note that Ruby allows a value like the one've sent
> to be interpreted as a cookie, even though it is not.]
>
> #lang scheme
> (define ex #"teaching-order=course;
> __utmz=165257760.1272597702.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none)\r\n")
>
> (define (not-cookie-parse s)
>  (for/fold ([cookies (make-immutable-hash empty)])
>    ([key*val (in-list (regexp-split #rx"[;,][ \t]*" s))])
>    (match-define (list _ key val) (regexp-match #rx"^([^=]*)=(.*)$" key*val))
>    (hash-update cookies key
>                 (curry append (regexp-split #rx"&" val))
>                 empty)))
>
> (not-cookie-parse ex)
> =>
> #hash((#"teaching-order" . (#"course")) (#"__utmz" .
> (#"165257760.1272597702.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none)\r\n")))
>
> Jay
>
>>
>> --- nadeem
>>
>>
>> On Fri, Apr 30, 2010 at 12:26 AM, Jay McCarthy <jay.mccarthy at gmail.com> wrote:
>>> Hi Nadeem,
>>>
>>> The problem is that this:
>>>
>>> "teaching-order=course;
>>> __utmz=165257760.1272597702.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none)\r\n"
>>>
>>> is not a valid cookie. In particular, the characters to the right of
>>> the second "=":
>>>
>>> "165257760.1272597702.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none)\r\n"
>>>
>>> should be a VALUE, which is a value, which is either a token or a quoted-string.
>>>
>>> It is not a token because it contains "(", ")", and "=".
>>>
>>> It is not a quoted-string because it is not wrapped in "".
>>>
>>> The Web Server does not throw an exception when it is asked to parse
>>> an invalid cookie string, instead it returns the empty list of
>>> cookies. You can look at the request's headers directly to do
>>> something to this header, but since it is not a cookie, the Web
>>> Server's cookie parsing cannot do anything with it.
>>>
>>> For Reference:
>>>
>>> http://tools.ietf.org/html/rfc2965 [for VALUE and value]
>>> http://tools.ietf.org/html/rfc2616 [for token and quoted-string]
>>>
>>> Jay
>>>
>>> On Thu, Apr 29, 2010 at 10:05 PM, Nadeem Abdul Hamid <nadeem at acm.org> wrote:
>>>> Hello all,
>>>>
>>>> I've been experimenting with a simple servlet to drive my website, but am having a strange problem with cookies not getting through when a Google Analytics cookie (__utmz) is present in the browser's request to the servlet. I've included a step-by-step trace below, but briefly, here's the problem: Starting from an empty browser cache of cookies, I request a page from my servlet that sets a cookie; when I then request a page again, I see that the cookie is sent by the browser and received by the servlet. However, then I visit another web page on the same domain as my servlet, which sets a __utmz (apparently a Google Analytics cookie). After this cookie is set, when I request pages from my servlet, I see that the browser is sending all the cookies, but none are getting through to the servlet! If I clear the __utmz cookie from the browser, and then request pages from the servlet, the servlet receives the cookie again.
>>>>
>>>> So, I haven't peeked into the webserver code (no time yet), but the question is: is there some reason that the __utmz cookie (and not any other __utma, __utmc, cookies) seems to cause interference with all other cookies getting through to it?
>>>>
>>>> Below are the steps that I take to replicate this problem, as well as snippets from my code. I added some terminal output to the servlet dispatch function that is passed to serve/servlet so that it displays the cookies it receives and the URL being requested before doing anything else.
>>>>
>>>> I appreciate any help/insight. Thanks,
>>>>
>>>> --- nadeem
>>>>
>>>>
>>>> **************************************************************
>>>> (Step 1)
>>>> Browser request (sniffed using Wireshark):
>>>>   GET /~nhamid/teaching/ HTTP/1.1\r\n
>>>>   (no cookies sent)
>>>>
>>>> Servlet output:
>>>>   ()/~nhamid/teaching/
>>>>
>>>>
>>>> (Step 2)
>>>>   request page that sets a cookie (and redirects to /~nhamid/teaching/course…)
>>>>
>>>>
>>>> (Step 3)
>>>> Browser request:
>>>>   GET /~nhamid/teaching/course HTTP/1.1\r\n
>>>>   Cookie: teaching-order=course\r\n
>>>>
>>>> Servlet output:
>>>>   ((teaching-order course #f #f))/~nhamid/teaching/course
>>>> (See the cookie is received.)
>>>>
>>>>
>>>> (Step 4) visit another page (www.berry.edu) -- sets __utmz (google analytics cookie) and a bunch of others, which I cleared out, except for the __utmz one.
>>>>
>>>>
>>>> (Step 5)
>>>> Browser request:
>>>>   GET /~nhamid/teaching/ HTTP/1.1\r\n
>>>>   Cookie: teaching-order=course; __utmz=165257760.1272597702.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none)\r\n
>>>>
>>>> Servlet output:
>>>>   ()/~nhamid/teaching/
>>>> (Note: no cookies received at all!)
>>>>
>>>>
>>>> (Step 6) delete the __utmz cookie from the browser. This is the *only* thing I change.
>>>>
>>>>
>>>> (Step 7)
>>>> Browser request:
>>>>   GET /~nhamid/teaching/ HTTP/1.1\r\n
>>>>   Cookie: teaching-order=course\r\n
>>>>
>>>> Servlet output:
>>>>   ((teaching-order course #f #f))/~nhamid/teaching/
>>>> (Cookie is received again!)
>>>>
>>>> **************************************************************
>>>>
>>>> Scheme code:
>>>>
>>>> (serve/servlet my-dispatch
>>>>               #:listen-ip #f
>>>>               #:launch-browser? #f
>>>>               #:servlet-path "/nhamid/index.ss"
>>>>               #:servlet-regexp #rx""
>>>>               #:extra-files-paths (list htdocs)
>>>>               #:stateless? false)
>>>>
>>>> (define-values (web-dispatch web-url)
>>>>  (dispatch-rules
>>>>   [("~nhamid" "index.ss") render-home]
>>>>   [("~nhamid") render-home]
>>>>   [("~nhamid" "") render-home]
>>>>   [("~nhamid" "editlinks") render-editlinks]
>>>>   [("~nhamid" "teaching" "") render-teaching]
>>>>   [("~nhamid" "teaching" "semester") render-teaching-by-semester]
>>>>   [("~nhamid" "teaching" "course") render-teaching-by-course]
>>>> ))
>>>>
>>>> (define (display-cookies req)
>>>>  (let ([cookies (request-cookies req)])
>>>>    (display
>>>>    (map (lambda (c) (list (client-cookie-name  c)
>>>>                           (client-cookie-value c)
>>>>                           (client-cookie-domain c)
>>>>                           (client-cookie-path c)))
>>>>         cookies))))
>>>>
>>>> ;; my-dispatch : request -> response
>>>> (define (my-dispatch req)
>>>>  (display-cookies req)
>>>>  (display (string-append (url->string (request-uri req)) "\n"))
>>>>  (web-dispatch req))
>>>>
>>>>
>>>> _________________________________________________
>>>>  For list-related administrative tasks:
>>>>  http://list.cs.brown.edu/mailman/listinfo/plt-scheme
>>>>
>>>
>>>
>>>
>>> --
>>> Jay McCarthy <jay at cs.byu.edu>
>>> Assistant Professor / Brigham Young University
>>> http://teammccarthy.org/jay
>>>
>>> "The glory of God is Intelligence" - D&C 93
>>>
>>
>>
>>
>> --
>> Nadeem Abdul Hamid
>> Assistant Professor, Computer Science
>> Berry College
>> PO Box 5014
>> 2277 Martha Berry Hwy NW
>> Mount Berry, GA 30149-5014
>> (706) 368-5632
>> http://cs.berry.edu/~nhamid/
>>
>
>
>
> --
> Jay McCarthy <jay at cs.byu.edu>
> Assistant Professor / Brigham Young University
> http://teammccarthy.org/jay
>
> "The glory of God is Intelligence" - D&C 93
> _________________________________________________
>  For list-related administrative tasks:
>  http://list.cs.brown.edu/mailman/listinfo/plt-scheme
>


Posted on the users mailing list.