* lib/uri/common.rb (URI#{en,de}code_www_form_component):

renamed from URI#{en,de}code_www_component. [ruby-dev:40672]

* lib/uri/common.rb (URI#encode_www_form_component): %-encoded
  element should have always two hex.

* lib/uri/common.rb (URI#encode_www_form_component):
  better treatment for ASCII incompatible encodings and
  encodings whose lead byte may use 7bit.

* lib/uri/common.rb (URI#decode_www_form_component): add %20.

* lib/uri/common.rb (URI#decode_www_form_component): add
  result's encoding as 2nd argument.

* lib/uri/common.rb (URI#decode_www_form): added.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@26962 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,22 @@
+Thu Mar 18 00:00:58 2010  NARUSE, Yui  <naruse@ruby-lang.org>
+
+	* lib/uri/common.rb (URI#{en,de}code_www_form_component):
+	  renamed from URI#{en,de}code_www_component. [ruby-dev:40672]
+
+	* lib/uri/common.rb (URI#encode_www_form_component): %-encoded
+	  element should have always two hex.
+
+	* lib/uri/common.rb (URI#encode_www_form_component):
+	  better treatment for ASCII incompatible encodings and
+	  encodings whose lead byte may use 7bit.
+
+	* lib/uri/common.rb (URI#decode_www_form_component): add %20.
+
+	* lib/uri/common.rb (URI#decode_www_form_component): add
+	  result's encoding as 2nd argument.
+
+	* lib/uri/common.rb (URI#decode_www_form): added.
+
 Wed Mar 17 16:25:53 2010  Nobuyoshi Nakada  <nobu@ruby-lang.org>
 
 	* hash.c (rb_hash_aset): allow recursive key.  [ruby-core:24648]

lib/uri/common.rb

@@ -729,49 +729,60 @@ module URI
   #
   # This refers http://www.w3.org/TR/html5/forms.html#url-encoded-form-data
   #
-  # See URI.decode_www_component(str), URI.encode_www_form(enum)
+  # See URI.decode_www_form_component, URI.encode_www_form
-  def self.encode_www_component(str)
+  def self.encode_www_form_component(str)
     if TBLENCWWWCOMP_.empty?
       256.times do |i|
         case i
         when 0x20
           TBLENCWWWCOMP_[' '] = '+'
-        when 0x2A, 0x2D, 0x2E, 0x30..0x39, 0x41..0x5A, 0x5F, 0x61..0x7A
+        # when 0x2A, 0x2D, 0x2E, 0x30..0x39, 0x41..0x5A, 0x5F, 0x61..0x7A
         else
-          TBLENCWWWCOMP_[i.chr] = '%%%X' % i
+          TBLENCWWWCOMP_[i.chr] = '%%%02X' % i
         end
       end
       TBLENCWWWCOMP_.freeze
     end
-    str = str.to_s.dup
+    str = str.to_s
-    enc = str.encoding
+    case str.encoding
-    str.force_encoding(Encoding::ASCII_8BIT)
+    when Encoding::ASCII_8BIT, Encoding::US_ASCII, Encoding::UTF_8
-    str.gsub!(/[^*\-.0-9A-Z_a-z]/, TBLENCWWWCOMP_)
+      str = str.dup.force_encoding(Encoding::ASCII_8BIT)
-    str.force_encoding(enc)
+      str.gsub!(/[^*\-.0-9A-Z_a-z]/, TBLENCWWWCOMP_)
+    when Encoding::UTF_16BE, Encoding::UTF_16LE, Encoding::UTF_32BE, Encoding::UTF_32LE
+      reg = Regexp.new('[^*\-.0-9A-Z_a-z]+'.encode(str.encoding))
+      str = str.gsub(reg){
+        $&.force_encoding(Encoding::ASCII_8BIT).gsub(/./, TBLENCWWWCOMP_).
+        force_encoding(str.encoding)
+      }
+    else
+      if str.encoding.ascii_compatible?
+        str = str.gsub(/[^*\-.0-9A-Z_a-z]+/){
+          $&.force_encoding(Encoding::ASCII_8BIT).gsub(/./, TBLENCWWWCOMP_)}
+      else
+        str = str.force_encoding(Encoding::ASCII_8BIT).gsub(/./, TBLENCWWWCOMP_)
+      end
+    end
+    str.force_encoding(Encoding::US_ASCII)
   end
 
   # Decode given +str+ of URL-encoded form data.
   #
   # This decods + to SP.
   #
-  # See URI.encode_www_component(str)
+  # See URI.encode_www_form_component, URI.decode_www_form
-  def self.decode_www_component(str)
+  def self.decode_www_form_component(str, enc=Encoding::UTF_8)
     if TBLDECWWWCOMP_.empty?
       256.times do |i|
-        case i
+        h, l = i>>4, i&15
-        when 0x20
+        TBLDECWWWCOMP_['%%%X%X' % [h, l]] = i.chr
-          TBLDECWWWCOMP_['+'] = ' '
+        TBLDECWWWCOMP_['%%%x%X' % [h, l]] = i.chr
-        else
+        TBLDECWWWCOMP_['%%%X%x' % [h, l]] = i.chr
-          h, l = i>>4, i&15
+        TBLDECWWWCOMP_['%%%x%x' % [h, l]] = i.chr
-          TBLDECWWWCOMP_['%%%X%X' % [h, l]] = i.chr
+        TBLDECWWWCOMP_['+'] = ' ' if i == 0x20
-          TBLDECWWWCOMP_['%%%x%X' % [h, l]] = i.chr
-          TBLDECWWWCOMP_['%%%X%x' % [h, l]] = i.chr
-          TBLDECWWWCOMP_['%%%x%x' % [h, l]] = i.chr
-        end
       end
       TBLDECWWWCOMP_.freeze
     end
-    str.gsub(/\+|%\h\h/, TBLDECWWWCOMP_)
+    str.gsub(/\+|%\h\h/, TBLDECWWWCOMP_).force_encoding(Encoding::UTF_8)
   end
 
   # Generate URL-encoded form data from given +enum+.
@@ -779,7 +790,7 @@ module URI
   # This generates application/x-www-form-urlencoded data defined in HTML5
   # from given an Enumerable object.
   #
-  # This internally uses URI.encode_www_component(str).
+  # This internally uses URI.encode_www_form_component(str).
   #
   # This doesn't convert encodings of give items, so convert them before call
   # this method if you want to send data as other than original encoding or
@@ -789,7 +800,7 @@ module URI
   #
   # This refers http://www.w3.org/TR/html5/forms.html#url-encoded-form-data
   #
-  # See URI.encode_www_component(str)
+  # See URI.encode_www_form_component, URI.decode_www_form
   def self.encode_www_form(enum)
     str = nil
     enum.each do |k,v|
@@ -798,12 +809,43 @@ module URI
       else
         str = ''.force_encoding(Encoding::US_ASCII)
       end
-      str << encode_www_component(k)
+      str << encode_www_form_component(k)
       str << '='
-      str << encode_www_component(v)
+      str << encode_www_form_component(v)
     end
     str
   end
+
+  # Decode URL-encoded form data from given +str+.
+  #
+  # This decodes application/x-www-form-urlencoded data
+  # and returns array of key-value array.
+  # This internally uses URI.decode_www_form_component.
+  #
+  # _charset_ hack is not supported now because the mapping from given charset
+  # to Ruby's encoding is not clear yet.
+  # see also http://www.w3.org/TR/html5/syntax.html#character-encodings-0
+  #
+  # This refers http://www.w3.org/TR/html5/forms.html#url-encoded-form-data
+  #
+  # ary = URI.decode_www_form("a=1&a=2&b=3")
+  # p ary                  #=> [['a', '1'], ['a', '2'], ['b', '3']]
+  # p ary.assoc('a').last  #=> '1'
+  # p ary.assoc('b').last  #=> '3'
+  # p ary.rassoc('a').last #=> '2'
+  # p Hash[ary]            # => {"a"=>"2", "b"=>"3"}
+  #
+  # See URI.decode_www_form_component, URI.encode_www_form
+  def self.decode_www_form(str, enc=Encoding::UTF_8)
+    ary = []
+    unless /\A\??(?<query>[^=;&]*=[^;&]*(?:[;&][^=;&]*=[^;&]*)*)\z/ =~ str
+      raise ArgumentError, "invalid data of application/x-www-form-urlencoded (#{str})"
+    end
+    query.scan(/([^=;&]+)=([^;&]*)/) do
+      ary << [decode_www_form_component($1, enc), decode_www_form_component($2, enc)]
+    end
+    ary
+  end
 end
 
 module Kernel

test/uri/test_common.rb

@@ -50,16 +50,22 @@ class TestCommon < Test::Unit::TestCase
     assert_raise(NoMethodError) { Object.new.URI("http://www.ruby-lang.org/") }
   end
 
-  def test_encode_www_component
+  def test_encode_www_form_component
-    assert_equal("+%21%22%23%24%25%26%27%28%29*%2B%2C-.%2F09%3A%3B%3C%3D%3E%3F%40" \
+    assert_equal("%00+%21%22%23%24%25%26%27%28%29*%2B%2C-.%2F09%3A%3B%3C%3D%3E%3F%40" \
                  "AZ%5B%5C%5D%5E_%60az%7B%7C%7D%7E",
-                 URI.encode_www_component(" !\"\#$%&'()*+,-./09:;<=>?@AZ[\\]^_`az{|}~"))
+                 URI.encode_www_form_component("\x00 !\"\#$%&'()*+,-./09:;<=>?@AZ[\\]^_`az{|}~"))
+    assert_equal("%95%41", URI.encode_www_form_component(
+                   "\x95\x41".force_encoding(Encoding::Shift_JIS)))
+    assert_equal("%30%42", URI.encode_www_form_component(
+                   "\x30\x42".force_encoding(Encoding::UTF_16BE)))
+    assert_equal("%30%42", URI.encode_www_form_component(
+                   "\x30\x42".force_encoding(Encoding::ISO_2022_JP)))
   end
 
-  def test_decode_www_component
+  def test_decode_www_form_component
-    assert_equal(" !\"\#$%&'()*+,-./09:;<=>?@AZ[\\]^_`az{|}~",
+    assert_equal("  !\"\#$%&'()*+,-./09:;<=>?@AZ[\\]^_`az{|}~",
-                 URI.decode_www_component(
+                 URI.decode_www_form_component(
-                   "+%21%22%23%24%25%26%27%28%29*%2B%2C-.%2F09%3A%3B%3C%3D%3E%3F%40" \
+                   "%20+%21%22%23%24%25%26%27%28%29*%2B%2C-.%2F09%3A%3B%3C%3D%3E%3F%40" \
                    "AZ%5B%5C%5D%5E_%60az%7B%7C%7D%7E"))
   end
 
@@ -74,6 +80,12 @@ class TestCommon < Test::Unit::TestCase
     assert_equal(expected, URI.encode_www_form([["a", "1"], ["\u3042", "\u6F22"]]))
     assert_equal(expected, URI.encode_www_form([[:a, 1], [:"\u3042", "\u6F22"]]))
   end
+
+  def test_decode_www_form
+    assert_equal([%w[a 1], %w[a 2]], URI.decode_www_form("a=1&a=2"))
+    assert_equal([%w[a 1], ["\u3042", "\u6F22"]],
+                 URI.decode_www_form("a=1&%E3%81%82=%E6%BC%A2"))
+  end
 end