Comments on: URL Validation in Ruby/Rails

By: Dean

Dean — Wed, 16 Jul 2008 01:32:47 +0000

Excellent. Thanks for the info!

By: jaren

jaren — Mon, 23 Jun 2008 05:38:09 +0000

f4hvYk dfv078fnw8f934ndvkg2l

By: JR

JR — Thu, 10 Apr 2008 04:13:36 +0000

Very informative.

By: alan

alan — Thu, 27 Mar 2008 23:26:29 +0000

in case its hard to see... PORT = /(([:]\d+)?)/ DOMAIN = /([a-z0-9\-]+\.?)*([a-z0-9]{2,})\.[a-z]{2,}/ NUMERIC_IP = /(?>(?:1?\d?\d|2[0-4]\d|25[0-5])\.){3}(?:1?\d?\d|2[0-4]\d|25[0-5])(?:\/(?:[12]?\d|3[012])|-(?>(?:1?\d?\d|2[0-4]\d|25[0-5])\.){3}(?:1?\d?\d|2[0-4]\d|25[0-5]))?/ validates_format_of :name, :with => /^((localhost)|#{DOMAIN}|#{NUMERIC_IP})#{PORT}$/

By: alan

alan — Thu, 27 Mar 2008 23:25:32 +0000

Check this model, towards the bottom there is a pretty good regex. It only needs to recognise urls with http|https, but that is very easy to do.

http://sample.caboo.se/weed2/app/models/domain.rb

By: jacques

jacques — Tue, 11 Mar 2008 20:58:54 +0000

in a description text i have to find and replace urls…
hi how can i search for reg in a string? How should the pattern look like?

thank you

By: here it is

here it is — Fri, 07 Mar 2008 02:32:06 +0000

When I said domain names up to 5 characters I meant domain extensions, like .info, .com, .org, .tv etc

By: here it is

here it is — Fri, 07 Mar 2008 02:31:07 +0000

This is the best regex i’ve found so far

/^(http|https):\/\/[a-z0-9]+([\-\.]{1}[a-z0-9]+)*\.[a-z]{2,5}(:[0-9]{1,5})?(\/.*)?$/ix

It handles both http, https, ip addresses, domain names, port numbers, domain names up to 5 characters, and even domains like the one Tom Harrison said could not be matched, were matched correctly by the above regex.

If you want to try it yourself

go into the console mode:

ruby ./script/console

url = “http://www.target.com/gp/detail.html/602-4045909-4263801?ASIN=B000NPCK3W&AFID=Froogle&LNM=B000NPCK3W|Lexmark_AllInOne_Printer_with_Scanner_and_Copier__X1240&ci_src=14110944&ci_sku=B000NPCK3W&ref=tgt_adv_XSG10001″

reg = /^(http|https):\/\/[a-z0-9]+([\-\.]{1}[a-z0-9]+)*\.[a-z]{2,5}(:[0-9]{1,5})?(\/.*)?$/ix

reg.match(url) ? true : false

you’ll see it’ll return true

By: Tom Harrison

Tom Harrison — Wed, 23 Jan 2008 17:38:44 +0000

Didn’t find any alternatives: the problem in the URL above is the vertical bar. If I escape it with %7D URI is happy as a clam.

So reading the URI specification, it does appear that this character falls into a group that while specifically excluded are in a class that should not be used in URLs (or escaped if they are). So I would be fine with Ruby’s picky URI class, except that in Rails, URLs are sometimes generated with ids in [123] square brackets, which are also not allowed by the spec, but which URI seems fine with.

So anyway, if anyone runs into this, just gsub replace out any characters like carat, backtick, tilde and possibly others with their CGI encoded variants.

By: Tom Harrison

Tom Harrison — Wed, 23 Jan 2008 17:03:38 +0000

Ah Google. It led me here, but I have found that URI is very, very picky about URLs.

For example, this one from target.com cannot be parsed:

http://www.target.com/gp/detail.html/602-4045909-4263801?ASIN=B000NPCK3W&AFID=Froogle&LNM=B000NPCK3W|Lexmark_AllInOne_Printer_with_Scanner_and_Copier__X1240&ci_src=14110944&ci_sku=B000NPCK3W&ref=tgt_adv_XSG10001

I think it is the vertical bar in the URL, but we have found numerous other characters (e.g. carat) that URI wont accept.

Seeking alternatives…