Add verify and clean methods to ActiveSupport::Multibyte.

When accepting character input from outside of your application you can't
blindly trust that all strings are properly encoded. With these methods
you can check incoming strings and clean them up if necessary.

Signed-off-by: Michael Koziarski <michael@koziarski.com>

Conflicts:

	activesupport/lib/active_support/multibyte.rb

Author: NZKoz

activesupport/lib/active_support/multibyte.rb

@@ -29,7 +29,35 @@ module Multibyte
     #
     # Example:
     #   ActiveSupport::Multibyte.proxy_class = CharsForUTF32
-    mattr_accessor :proxy_class
-    self.proxy_class = ActiveSupport::Multibyte::Chars
+    def self.proxy_class=(klass)
+      @proxy_class = klass
+    end
+
+    # Returns the currect proxy class
+    def self.proxy_class
+      @proxy_class ||= ActiveSupport::Multibyte::Chars
+    end
+
+    # Regular expressions that describe valid byte sequences for a character
+    VALID_CHARACTER = {
+      # Borrowed from the Kconv library by Shinji KONO - (also as seen on the W3C site)
+      'UTF-8' => /\A(?:
+                  [\x00-\x7f]                                         |
+                  [\xc2-\xdf] [\x80-\xbf]                             |
+                  \xe0        [\xa0-\xbf] [\x80-\xbf]                 |
+                  [\xe1-\xef] [\x80-\xbf] [\x80-\xbf]                 |
+                  \xf0        [\x90-\xbf] [\x80-\xbf] [\x80-\xbf]     |
+                  [\xf1-\xf3] [\x80-\xbf] [\x80-\xbf] [\x80-\xbf]     |
+                  \xf4        [\x80-\x8f] [\x80-\xbf] [\x80-\xbf])\z /xn,
+      # Quick check for valid Shift-JIS characters, disregards the odd-even pairing
+      'Shift_JIS' => /\A(?:
+                  [\x00-\x7e \xa1-\xdf]                                     |
+                  [\x81-\x9f \xe0-\xef] [\x40-\x7e \x80-\x9e \x9f-\xfc])\z /xn
+    }
   end
 end
+
+require 'active_support/multibyte/chars'
+require 'active_support/multibyte/exceptions'
+require 'active_support/multibyte/unicode_database'
+require 'active_support/multibyte/utils'

activesupport/lib/active_support/multibyte/chars.rb

@@ -74,16 +74,7 @@ def self.codepoints_to_pattern(array_of_codepoints) #:nodoc:
       UNICODE_TRAILERS_PAT = /(#{codepoints_to_pattern(UNICODE_LEADERS_AND_TRAILERS)})+\Z/
       UNICODE_LEADERS_PAT = /\A(#{codepoints_to_pattern(UNICODE_LEADERS_AND_TRAILERS)})+/
 
-      # Borrowed from the Kconv library by Shinji KONO - (also as seen on the W3C site)
-      UTF8_PAT = /\A(?:
-                     [\x00-\x7f]                                     |
-                     [\xc2-\xdf] [\x80-\xbf]                         |
-                     \xe0        [\xa0-\xbf] [\x80-\xbf]             |
-                     [\xe1-\xef] [\x80-\xbf] [\x80-\xbf]             |
-                     \xf0        [\x90-\xbf] [\x80-\xbf] [\x80-\xbf] |
-                     [\xf1-\xf3] [\x80-\xbf] [\x80-\xbf] [\x80-\xbf] |
-                     \xf4        [\x80-\x8f] [\x80-\xbf] [\x80-\xbf]
-                    )*\z/xn
+      UTF8_PAT = ActiveSupport::Multibyte::VALID_CHARACTER['UTF-8']
 
       attr_reader :wrapped_string
       alias to_s wrapped_string
@@ -308,31 +299,31 @@ def center(integer, padstr=' ')
       def rstrip
         chars(@wrapped_string.gsub(UNICODE_TRAILERS_PAT, ''))
       end
-      
+
       # Strips entire range of Unicode whitespace from the left of the string.
       def lstrip
         chars(@wrapped_string.gsub(UNICODE_LEADERS_PAT, ''))
       end
-      
+
       # Strips entire range of Unicode whitespace from the right and left of the string.
       def strip
         rstrip.lstrip
       end
-      
+
       # Returns the number of codepoints in the string
       def size
         self.class.u_unpack(@wrapped_string).size
       end
       alias_method :length, :size
-      
+
       # Reverses all characters in the string.
       #
       # Example:
       #   'Café'.mb_chars.reverse.to_s #=> 'éfaC'
       def reverse
         chars(self.class.u_unpack(@wrapped_string).reverse.pack('U*'))
       end
-      
+
       # Implements Unicode-aware slice with codepoints. Slicing on one point returns the codepoints for that
       # character.
       #
@@ -647,7 +638,7 @@ def tidy_bytes(string)
           string.split(//u).map do |c|
             c.force_encoding(Encoding::ASCII) if c.respond_to?(:force_encoding)
 
-            if !UTF8_PAT.match(c)
+            if !ActiveSupport::Multibyte::VALID_CHARACTER['UTF-8'].match(c)
               n = c.unpack('C')[0]
               n < 128 ? n.chr :
               n < 160 ? [UCD.cp1252[n] || n].pack('U') :

activesupport/lib/active_support/multibyte/utils.rb

@@ -0,0 +1,61 @@
+# encoding: utf-8
+
+module ActiveSupport #:nodoc:
+  module Multibyte #:nodoc:
+    if Kernel.const_defined?(:Encoding)
+      # Returns a regular expression that matches valid characters in the current encoding
+      def self.valid_character
+        VALID_CHARACTER[Encoding.default_internal.to_s]
+      end
+    else
+      def self.valid_character
+        case $KCODE
+        when 'UTF8'
+          VALID_CHARACTER['UTF-8']
+        when 'SJIS'
+          VALID_CHARACTER['Shift_JIS']
+        end
+      end
+    end
+
+    if 'string'.respond_to?(:valid_encoding?)
+      # Verifies the encoding of a string
+      def self.verify(string)
+        string.valid_encoding?
+      end
+    else
+      def self.verify(string)
+        if expression = valid_character
+          for c in string.split(//)
+            return false unless valid_character.match(c)
+          end
+        end
+        true
+      end
+    end
+
+    # Verifies the encoding of the string and raises an exception when it's not valid
+    def self.verify!(string)
+      raise EncodingError.new("Found characters with invalid encoding") unless verify(string)
+    end
+
+    if 'string'.respond_to?(:force_encoding)
+      # Removes all invalid characters from the string.
+      #
+      # Note: this method is a no-op in Ruby 1.9
+      def self.clean(string)
+        string
+      end
+    else
+      def self.clean(string)
+        if expression = valid_character
+          stripped = []; for c in string.split(//)
+            stripped << c if valid_character.match(c)
+          end; stripped.join
+        else
+          string
+        end
+      end
+    end
+  end
+end

activesupport/test/multibyte_utils_test.rb

@@ -0,0 +1,141 @@
+# encoding: utf-8
+
+require 'abstract_unit'
+require 'multibyte_test_helpers'
+
+class MultibyteUtilsTest < ActiveSupport::TestCase
+  include MultibyteTestHelpers
+
+  test "valid_character returns an expression for the current encoding" do
+    with_encoding('None') do
+      assert_nil ActiveSupport::Multibyte.valid_character
+    end
+    with_encoding('UTF8') do
+      assert_equal ActiveSupport::Multibyte::VALID_CHARACTER['UTF-8'], ActiveSupport::Multibyte.valid_character
+    end
+    with_encoding('SJIS') do
+      assert_equal ActiveSupport::Multibyte::VALID_CHARACTER['Shift_JIS'], ActiveSupport::Multibyte.valid_character
+    end
+  end
+
+  test "verify verifies ASCII strings are properly encoded" do
+    with_encoding('None') do
+      examples.each do |example|
+        assert ActiveSupport::Multibyte.verify(example)
+      end
+    end
+  end
+
+  test "verify verifies UTF-8 strings are properly encoded" do
+    with_encoding('UTF8') do
+      assert ActiveSupport::Multibyte.verify(example('valid UTF-8'))
+      assert !ActiveSupport::Multibyte.verify(example('invalid UTF-8'))
+    end
+  end
+
+  test "verify verifies Shift-JIS strings are properly encoded" do
+    with_encoding('SJIS') do
+      assert ActiveSupport::Multibyte.verify(example('valid Shift-JIS'))
+      assert !ActiveSupport::Multibyte.verify(example('invalid Shift-JIS'))
+    end
+  end
+
+  test "verify! raises an exception when it finds an invalid character" do
+    with_encoding('UTF8') do
+      assert_raises(ActiveSupport::Multibyte::EncodingError) do
+        ActiveSupport::Multibyte.verify!(example('invalid UTF-8'))
+      end
+    end
+  end
+
+  test "verify! doesn't raise an exception when the encoding is valid" do
+    with_encoding('UTF8') do
+      assert_nothing_raised do
+        ActiveSupport::Multibyte.verify!(example('valid UTF-8'))
+      end
+    end
+  end
+
+  if RUBY_VERSION < '1.9'
+    test "clean leaves ASCII strings intact" do
+      with_encoding('None') do
+        [
+          'word', "\270\236\010\210\245"
+        ].each do |string|
+          assert_equal string, ActiveSupport::Multibyte.clean(string)
+        end
+      end
+    end
+
+    test "clean cleans invalid characters from UTF-8 encoded strings" do
+      with_encoding('UTF8') do
+        cleaned_utf8 = [8].pack('C*')
+        assert_equal example('valid UTF-8'), ActiveSupport::Multibyte.clean(example('valid UTF-8'))
+        assert_equal cleaned_utf8, ActiveSupport::Multibyte.clean(example('invalid UTF-8'))
+      end
+    end
+
+    test "clean cleans invalid characters from Shift-JIS encoded strings" do
+      with_encoding('SJIS') do
+        cleaned_sjis = [184, 0, 136, 165].pack('C*')
+        assert_equal example('valid Shift-JIS'), ActiveSupport::Multibyte.clean(example('valid Shift-JIS'))
+        assert_equal cleaned_sjis, ActiveSupport::Multibyte.clean(example('invalid Shift-JIS'))
+      end
+    end
+  else
+    test "clean is a no-op" do
+      with_encoding('UTF8') do
+        assert_equal example('invalid Shift-JIS'), ActiveSupport::Multibyte.clean(example('invalid Shift-JIS'))
+      end
+    end
+  end
+
+  private
+
+  STRINGS = {
+    'valid ASCII'       => [65, 83, 67, 73, 73].pack('C*'),
+    'invalid ASCII'     => [128].pack('C*'),
+    'valid UTF-8'       => [227, 129, 147, 227, 129, 171, 227, 129, 161, 227, 130, 143].pack('C*'),
+    'invalid UTF-8'     => [184, 158, 8, 136, 165].pack('C*'),
+    'valid Shift-JIS'   => [131, 122, 129, 91, 131, 128].pack('C*'),
+    'invalid Shift-JIS' => [184, 158, 8, 0, 255, 136, 165].pack('C*')
+  }
+
+  if Kernel.const_defined?(:Encoding)
+    def example(key)
+      STRINGS[key].force_encoding(Encoding.default_internal)
+    end
+
+    def examples
+      STRINGS.values.map { |s| s.force_encoding(Encoding.default_internal) }
+    end
+  else
+    def example(key)
+      STRINGS[key]
+    end
+
+    def examples
+      STRINGS.values
+    end
+  end
+
+  if 'string'.respond_to?(:encoding)
+    def with_encoding(enc)
+      before = Encoding.default_internal
+
+      case enc
+      when 'UTF8'
+        Encoding.default_internal = Encoding::UTF_8
+      when 'SJIS'
+        Encoding.default_internal = Encoding::Shift_JIS
+      else
+        Encoding.default_internal = Encoding::BINARY
+      end
+      yield
+
+      Encoding.default_internal = before
+    end
+  else
+    alias with_encoding with_kcode
+  end
+end