Validate URLs using Addressable gem to allow non-ascii characters (closes #111)

README.md

@@ -128,7 +128,7 @@ Exhaustive list of supported validators and their implementation:
 * `ssn` : Social Security Number (only for USA).
 * `tracking_number`: based on a set of predefined masks
 * `twitter` : based on a regular expression
-* `url`   : based on a regular expression
+* `url`   : based on [`Addressable`](https://github.com/sporkmonger/addressable) gem
 
 
 ### Handling error messages

activevalidators.gemspec

@@ -17,6 +17,7 @@ Gem::Specification.new do |s|
   s.add_dependency 'rake'
   s.add_dependency 'mail'
   s.add_dependency 'date_validator'
+  s.add_dependency 'addressable'            , '~> 2.7'
   s.add_dependency 'activemodel'            , '>= 3.0'
   s.add_dependency 'phony'                  , '~> 2.0'
   s.add_dependency 'countries'              , '>= 1.2', '< 4.0'

lib/active_validators/active_model/validations/url_validator.rb

@@ -1,10 +1,10 @@
 require 'active_support/core_ext/array/wrap'
-require 'uri'
+require 'addressable'
 
 module ActiveModel
   module Validations
 
-    # Public: Uses `URI.regexp` to validate URLs, by default only allows
+    # Public: Uses `Addressable::URI.parse` to validate URLs, by default only allows
     # the http and https protocols.
     #
     # Examples
@@ -46,7 +46,7 @@ def initialize(options)
       def validate_each(record, attribute, value)
         uri = as_uri(value)
         tld_requirement_fullfilled = check_tld_requirement(value)
-        record.errors.add(attribute) unless uri && value.to_s =~ uri_regexp && tld_requirement_fullfilled
+        record.errors.add(attribute) unless uri && uri.scheme.in?(protocols) && tld_requirement_fullfilled
       end
 
       private
@@ -68,22 +68,14 @@ def protocols
         Array.wrap(options[:protocols] || %w{http https})
       end
 
-      # Internal: Constructs the regular expression to check
-      # the URI for the configured protocols.
-      #
-      # Returns the Regexp.
-      def uri_regexp
-        @uri_regexp ||= /\A#{URI::Parser.new.make_regexp(protocols)}\z/
-      end
-
       # Internal: Checks if the tld requirements are fullfilled
       #
       # When :require_tld option is set to true, the url will be searched for
       # a dot anywhere inside the host except the first or last position
       #
       # Returns a boolean value.
       def check_tld_requirement(value)
-        host = URI.parse(value.to_s).host rescue value
+        host = Addressable::URI.parse(value.to_s).host rescue value
         options[:require_tld] === true ? host =~ /.(\.)\w+/ : true
       end
 
@@ -92,7 +84,7 @@ def check_tld_requirement(value)
       #
       # Returns the URI or nil.
       def as_uri(value)
-        URI.parse(value.to_s) rescue nil if value.present?
+        Addressable::URI.parse(value.to_s) rescue nil if value.present?
       end
     end
   end

test/validations/url_test.rb

@@ -1,4 +1,5 @@
 require 'test_helper'
+
 ActiveValidators.activate(:url)
 
 describe "Url Validation" do
@@ -49,6 +50,20 @@ def build_ftp_record
       _(subject.errors.size).must_equal(0)
     end
 
+    it "accepts valid urls with non-ascii domain name" do
+      subject = build_url_record
+      subject.url = 'https://www.詹姆斯.com'
+      _(subject.valid?).must_equal(true)
+      _(subject.errors.size).must_equal(0)
+    end
+
+    it "accepts valid urls with non-ascii path" do
+      subject = build_url_record
+      subject.url = 'https://www.example.com/ουτοπία'
+      _(subject.valid?).must_equal(true)
+      _(subject.errors.size).must_equal(0)
+    end
+
     it "accepts ftp if defined" do
       subject = build_ftp_record
       subject.url = 'ftp://ftp.verrot.fr'
@@ -67,7 +82,7 @@ def build_ftp_record
   describe "for invalid urls" do
     it "rejects invalid urls" do
       subject = build_url_record
-      subject.url = 'http://^^^^.fr'
+      subject.url = 'http://in va lid.fr'
       _(subject.valid?).must_equal(false)
       _(subject.errors.size).must_equal(1)
     end
@@ -81,7 +96,7 @@ def build_ftp_record
 
     it "generates an error message of type invalid" do
       subject = build_url_record
-      subject.url = 'http://^^^^.fr'
+      subject.url = 'http://in va lid.fr'
       _(subject.valid?).must_equal(false)
       _(subject.errors[:url].include?(subject.errors.generate_message(:url, :invalid))).must_equal(true)
     end