Disecting Set-Cookie header with StringRegExp (half-solved)

[BAM5]

I'm attempting to take the string a server returns in the Set-Cookie header and break it up into its bits and pieces and have come pretty close with this.

Set-Cookie string looks like this: "Name=value; expires=Thu, 02-Sep-2010 16:44:12 GMT; path=/ Name=value; expires=Thu, 02-Sep-2010 16:44:12 GMT; path=/ secure Name2=Value2; expires=Wed, 02-Sep-2010 16:44:12 GMT; path=/ HttpOnly name3=value3; expires=Thu, 02-Sep-2010 16:44:12 GMT; path=/ secure HttpOnly"

([^=]+)=([^;]+); (?:(path)=(\S+)\s?)?

This will work for all cookies without the HttpOnly and secure flag.

([^=]+)=([^;]+); (?:(path)=(\S+)\s?)?(?:(secure)?\s?|(HttpOnly)?\s?){0,2}

Is what I came up with for the flags, but for some reason it doesn't work.

Could some sre master out there point out the error of my ways?

Thanks

BAM5

Edited August 3, 2010 by BAM5

[jchd]

This pattern seems to do what you want in the example you give, but would probably need some sophistication to work with pathnames having whitespaces in them.

(?i)([^=]+)=([^;]+);\s([^=]+)=([^;]+);\s([^=]+)=([^ ]+)\s(secure)?\s?(HttpOnly)?\s*

(Use mode 3 of course)

EDIT: I forgot to mention that I take the presence of namexxx=yyy expires=ttt and path=zzz for obligatory.

Edited August 3, 2010 by jchd

[BAM5]

Ah, namexxx=yyy expires=ttt and path=zzz are all required for cookies, but there are other values that can be added in optionally as well (such as domain and max-age), so unfortunately, your sre will not work for me.

A modification of the second pattern I posted almost works, except for when the httponly flag is up and secure isn't, it'll cause an blank element in the array.

Modification: ([^=]+)=([^;]+); (?:(path)=(\S+)\s?)?(?:(secure)\s?|(HttpOnly)\s?){0,2}

[BAM5]

After messing around a bit I figured out this pattern:

([^=]+)=([^;]+); (?:(path)=(\S+)\s?)?(secure|HttpOnly)?\s?(secure|HttpOnly)?\s?

It isn't exactly what I was looking for, but it works.

[jchd]

The more optional entities are to be parsed (furthermore using varying syntax) the more complicated and potentially fragile your SRE will become.

If I had to do it myself, I wouldn't rely on a single SRE to do it entirely in one shot, as some new option/syntax is likely to make it fail almost anytime soon. I would dissect things with a more pedestrian approach. Except if you have millions of such cookies to routinely process, the added time spent (if any) is most certainly unnoticeable.