[ruby-core:26223] [Bug #2251] URI.parse accepts strings with invalid characters

Bug #2251: URI.parse accepts strings with invalid characters
http://redmine.ruby-lang.org/issues/show/2251

Author: Sam Quigley
Status: Open, Priority: Normal
Category: lib
ruby -v: ruby 1.9.1p243 (2009-07-16 revision 24175) [i386-darwin10.0.0] 

The regexes used in URI::Parser's initialize_regexp use ^ and $ rather than \A and \Z:

   399       # for URI::split
   400       ret[:ABS_URI] = Regexp.new('^' + pattern[:X_ABS_URI] + '$', Regexp::EXTENDED)
   401       ret[:REL_URI] = Regexp.new('^' + pattern[:X_REL_URI] + '$', Regexp::EXTENDED)

The result is that URI.parse matches on any URI separated by newlines, rather than on its argument as a whole:

   irb(main):001:0> require 'uri'
   => true
   irb(main):002:0> URI.parse("blah\nhttp://www.foo.com/\nblahblah")
   => #<URI::HTTP:0x000001010aac78 URL:http://www.foo.com/>

I think programmers would expect URI.parse to only successfully parse strings that *are* URIs, rather than any string that *contains* a URI surrounded by a particular kind of whitespace.  This issue has apparently caused at least one security vulnerability in the real world: http://schmoil.blogspot.com/2009/10/mainlining-new-lines-feel-burn.html

Replacing the ^ and $ with \A and \Z should fix the issue, and is unlikely to break any existing code.  The Rubyspec project does not seem to have any tests for this behavior.  This behavior is present in at least versions 1.8.6, 1.8.7, and 1.9.1.

-sq


----------------------------------------
http://redmine.ruby-lang.org