Add support for []= on ActiveSupport::Multibyte::Chars. Closes #9142. [ewan, manfred]

git-svn-id: http://svn-commit.rubyonrails.org/rails/trunk@7257 5ecf4fe2-1ee6-0310-87b1-e25e094e27de

3 files changed, 74 insertions, 0 deletions

diff --git a/activesupport/CHANGELOG b/activesupport/CHANGELOG
index 1b539f52bf..35a7d77859 100644
--- a/activesupport/CHANGELOG
+++ b/activesupport/CHANGELOG
@@ -1,5 +1,7 @@
 *SVN*
 
+* Add support for []= on ActiveSupport::Multibyte::Chars. Closes #9142. [ewan, manfred]
+
 * Added Array#extract_options! to encapsulate the pattern of getting an options hash out of a variable number of parameters #8759 [norbert].
 
 * Let alias_attribute work with attributes with initial capital letters (legacy columns etc).  Closes #8596 [mpalmer]
diff --git a/activesupport/lib/active_support/multibyte/handlers/utf8_handler.rb b/activesupport/lib/active_support/multibyte/handlers/utf8_handler.rb
index 009950d33e..02fc7b3e2b 100644
--- a/activesupport/lib/active_support/multibyte/handlers/utf8_handler.rb
+++ b/activesupport/lib/active_support/multibyte/handlers/utf8_handler.rb
@@ -140,6 +140,44 @@ module ActiveSupport::Multibyte::Handlers #:nodoc:
         bidx ? (u_unpack(str.slice(0...bidx)).size) : nil
       end
       
+      # Works just like the indexed replace method on string, except instead of byte offsets you specify
+      # character offsets.
+      #
+      # Example:
+      #
+      #   s = "Müller"
+      #   s.chars[2] = "e" # Replace character with offset 2
+      #   s
+      #   #=> "Müeler"
+      #
+      #   s = "Müller"
+      #   s.chars[1, 2] = "ö" # Replace 2 characters at character offset 1
+      #   s
+      #   #=> "Möler"
+      def []=(str, *args)
+        replace_by = args.pop
+        # Indexed replace with regular expressions already works
+        return str[*args] = replace_by if args.first.is_a?(Regexp)
+        result = u_unpack(str)
+        if args[0].is_a?(Fixnum)
+          raise IndexError, "index #{args[0]} out of string" if args[0] >= result.length
+          min = args[0]
+          max = args[1].nil? ? min : (min + args[1] - 1)
+          range = Range.new(min, max)
+          replace_by = [replace_by].pack('U') if replace_by.is_a?(Fixnum)
+        elsif args.first.is_a?(Range)
+          raise RangeError, "#{args[0]} out of range" if args[0].min >= result.length
+          range = args[0]
+        else
+          needle = args[0].to_s
+          min = index(str, needle)
+          max = min + length(needle) - 1
+          range = Range.new(min, max)
+        end
+        result[range] = u_unpack(replace_by)
+        str.replace(result.pack('U*'))
+      end
+      
       # Does Unicode-aware rstrip
       def rstrip(str)
         str.gsub(UNICODE_TRAILERS_PAT, '')
diff --git a/activesupport/test/multibyte_handler_test.rb b/activesupport/test/multibyte_handler_test.rb
index ea728aa555..e4744def6c 100644
--- a/activesupport/test/multibyte_handler_test.rb
+++ b/activesupport/test/multibyte_handler_test.rb
@@ -199,6 +199,40 @@ module UTF8HandlingTest
      assert_raise(ActiveSupport::Multibyte::Handlers::EncodingError) { @handler.index(@bytestring, "\010") }
   end
   
+  def test_indexed_insert
+    s = "Καλη!"
+    @handler[s, 2] = "a"
+    assert_equal "Καaη!", s
+    @handler[s, 2] = "ηη"
+    assert_equal "Καηηη!", s
+    assert_raises(IndexError) { @handler[s, 10] = 'a' }
+    assert_equal "Καηηη!", s
+    @handler[s, 2] = 32
+    assert_equal "Κα ηη!", s
+    @handler[s, 3, 2] = "λλλ"
+    assert_equal "Κα λλλ!", s
+    @handler[s, 1, 0] = "λ"
+    assert_equal "Κλα λλλ!", s
+    assert_raises(IndexError) { @handler[s, 10, 4] = 'a' }
+    assert_equal "Κλα λλλ!", s
+    @handler[s, 4..6] = "ηη"
+    assert_equal "Κλα ηη!", s
+    assert_raises(RangeError) { @handler[s, 10..12] = 'a' }
+    assert_equal "Κλα ηη!", s
+    @handler[s, /ηη/] = "λλλ"
+    assert_equal "Κλα λλλ!", s
+    assert_raises(IndexError) { @handler[s, /ii/] = 'a' }
+    assert_equal "Κλα λλλ!", s
+    @handler[s, /(λλ)(.)/, 2] = "α"
+    assert_equal "Κλα λλα!", s
+    assert_raises(IndexError) { @handler[s, /()/, 10] = 'a' }
+    assert_equal "Κλα λλα!", s
+    @handler[s, "α"] = "η"
+    assert_equal "Κλη λλα!", s
+    @handler[s, "λλ"] = "ααα"
+    assert_equal "Κλη αααα!", s
+  end
+  
   def test_strip
     # A unicode aware version of strip should strip all 26 types of whitespace. This includes the NO BREAK SPACE
     # aka BOM (byte order mark). The byte order mark has no place in UTF-8 because it's used to detect LE and BE.