Ruby on Rails URL Validation (regex)

Question

I'm trying to use a regular expression to validate the format of a URL in my Rails model. I've tested the regex in Rubular with the URL http://trentscott.com and it matched.

Any idea why it fails validation when I test it in my Rails app (it says "name is invalid").

Code:

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

  validates :serial, :presence => true
  validates :name, :presence => true,
                   :format    => {  :with => url_regex  }

Solution 1

Your input ( http://trentscott.com) does not have a subdomain but the regex is checking for one.

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

Update

You also don't need the ? after ((http|https):\/\/) unless the protocol is sometimes missing. I've also escaped . because that will match any character. I'm not sure what the grouping above is for, but here is a better version that supports dashes and groups by section

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

Solution 2

You don't need to use a regexp here. Ruby has a much more reliable way to do that:

# Use the URI module distributed with Ruby:

require 'uri'

unless (url =~ URI::regexp).nil?
    # Correct URL
end

(this answer comes from this post:)

Solution 3

(I like Thomas Hupkens' answer, but for other people viewing, I'll recommend Addressable)

It's not recommended to use regex to validate URLs.

Use Ruby's URI library or a replacement like Addressable, both of which making URL validation trivial. Unlike URI, Addressable can also handle international characters and tlds.

Example Usage:

require 'addressable/uri'

Addressable::URI.parse("кц.рф") # Works

uri = Addressable::URI.parse("http://example.com/path/to/resource/")
uri.scheme
#=> "http"
uri.host
#=> "example.com"
uri.path
#=> "/path/to/resource/"

And you could build a custom validation like:

class Example
  include ActiveModel::Validations

  ##
  # Validates a URL
  #
  # If the URI library can parse the value, and the scheme is valid
  # then we assume the url is valid
  #
  class UrlValidator < ActiveModel::EachValidator
    def validate_each(record, attribute, value)
      begin
        uri = Addressable::URI.parse(value)

        if !["http","https","ftp"].include?(uri.scheme)
          raise Addressable::URI::InvalidURIError
        end
      rescue Addressable::URI::InvalidURIError
        record.errors[attribute] << "Invalid URL"
      end
    end
  end

  validates :field, :url => true
end

Code Source

Solution 4

Try this.

It's working for me.

/(ftp|http|https):\/\/(\w+:{0,1}\w*@)?(\S+)(:[0-9]+)?(\/|\/([\w#!:.?+=&%@!\-\/]))?/

View more solutions

Comments

[Trent Scott]

Thanks. That fixed the error but now an entry like "abcd" is valid. Any idea on how to fix that?

[cordsen]

The update should work. One more thing I removed was the [-.] and replaced it with \.

[stephenmurdoch]

after looking at addressable, I think it wins hands down, thanks

[David Keener]

This does not handle international domain names, which can be represented in ASCII like: www.xn--b1akcweg3a.xn--p1ai. Yes, this gives you double dashes in your domain, which is legal, as well as top-level domains (the right-most component) that are longer than 3 characters.

[onetwopunch]

+1 for addressable BUT don't assume that it will raise any exceptions because it won't. Addressable::URI.parse will fail silently trying its best to figure out the URI. For example, say you want to validate an incorrect URI such as: http://thing.com. Addressable will call the scheme http and the domain http as well since it views the colon as a port delimiter. No error will be raised

[huzefa biyawarwala]

@cordsen: what if I want to write a regex in Ruby for a URL which includes any non-ASCII characters or Chinese characters as well? For example, http://www.詹姆斯.com/ Can you please let me know how to figure this out?

[Machisuji]

"https://foo;bar.com" =~ URI::regexp yields a successful match. So this isn't very useful.