From: Glenn Parker <glenn.parker@...>
Date: 2004-12-26T13:13:54+09:00
Subject: Re: Net::HTTPResponse extracting name=value from "set-cookies"

Arturo 'Buanzo' Busleiman wrote:
> 
> I have this String, gotten from the set-cookies value of an HTTPResponse 
> (from a post call):
> 
> KEY1=VALUE1; Path=/, KEY2=VALUE2; Domain=.somedomain.net; Expires=Mon, 
> 24-Jan-2005 19:40:10 GMT;
> Path=/, KEY3=VALUE3; Domain=e-messenger.net; Expires=Mon, 24-Jan-2005 
> 19:40:10 GMT; Path=/
> 
> How can I extract each KEY,VALUE pair? I'm currently TRYING to use the 
> scan String method, but I've
> never been that good with regular expressions... any help?

input = 'KEY1=VALUE1; Path=/, KEY2=VALUE2; Domain=.somedomain.net; ' +
'Expires=Mon, 24-Jan-2005 19:40:10 GMT; Path=/, KEY3=VALUE3; ' +
'Domain=e-messenger.net; Expires=Mon, 24-Jan-2005 19:40:10 GMT; Path=/'

pkv = Regexp.new(';?\s*(?:Path=([^,]+),?)?\s*(?:([^=]+)=([^;]+))?')
input.scan(pkv) do |path, key, value|
   puts "path=#{path} key=#{key} value=#{value}"
end

Produces:

path= key=KEY1 value=VALUE1
path=/ key=KEY2 value=VALUE2
path= key=Domain value=.somedomain.net
path= key=Expires value=Mon, 24-Jan-2005 19:40:10 GMT
path=/ key=KEY3 value=VALUE3
path= key=Domain value=e-messenger.net
path= key=Expires value=Mon, 24-Jan-2005 19:40:10 GMT
path=/ key= value=
path= key= value=

The last line is annoying, but hard to fix.  Since you can have either 
Path alone, or Key=Value alone, or both together, it's hard to avoid 
matching the fourth option, which is neither Path nor Key=Value, i.e. 
the empty string.  That's what I get at the end of the scan, but it's 
easy to ignore that case.

-- 
Glenn Parker | glenn.parker-AT-comcast.net | <http://www.tetrafoil.com/>