Add basic encoding support and String byte operations

opal · Oct 23, 2013 · 423af6c · 423af6c
1 parent 77b6f82
commit 423af6c
Show file tree

Hide file tree

Showing 3 changed files with 158 additions and 1 deletion.
diff --git a/corelib/encoding.rb b/corelib/encoding.rb
@@ -0,0 +1,123 @@
+class Encoding
+  def self.register(name, options = {}, &block)
+    names    = [name] + (options[:aliases] || [])
+    encoding = Class.new(self, &block).
+      new(name, names, options[:ascii] || false, options[:dummy] || false)
+
+    names.each {|name|
+      const_set name.sub('-', '_'), encoding
+    }
+  end
+
+  def self.find(name)
+    return name if self === name
+
+    constants.each {|const|
+      encoding = const_get(const)
+
+      if encoding.name == name || encoding.names.include?(name)
+        return encoding
+      end
+    }
+
+    raise ArgumentError, "unknown encoding name - #{name}"
+  end
+
+  attr_reader :name, :names
+
+  def initialize(name, names, ascii, dummy)
+    @name  = name
+    @names = names
+    @ascii = ascii
+    @dummy = dummy
+  end
+
+  def ascii_compatible?
+    @ascii
+  end
+
+  def dummy?
+    @dummy
+  end
+
+  def to_s
+    @name
+  end
+
+  def inspect
+    "#<Encoding:#{@name}#{" (dummy)" if @dummy}>"
+  end
+
+  # methods to implement per encoding
+  def each_byte(*)
+    raise NotImplementedError
+  end
+
+  def getbyte(*)
+    raise NotImplementedError
+  end
+
+  def bytesize(*)
+    raise NotImplementedError
+  end
+end
+
+Encoding.register "UTF-8", aliases: ["CP65001"], ascii: true do
+  def each_byte(string, &block)
+    %x{
+      for (var i = 0, length = string.length; i < length; i++) {
+        var code = string.charCodeAt(i);
+
+        if (code <= 0x7f) {
+          #{yield `code`};
+        }
+        else {
+          var encoded = encodeURIComponent(string.charAt(i)).substr(1).split('%');
+
+          for (var j = 0, encoded_length = encoded.length; j < encoded_length; j++) {
+            #{yield `parseInt(encoded[j], 16)`};
+          }
+        }
+      }
+    }
+  end
+
+  def bytesize
+    bytes.length
+  end
+end
+
+Encoding.register "UTF-16LE" do
+  def each_byte(string, &block)
+    %x{
+      for (var i = 0, length = string.length; i < length; i++) {
+        var code = string.charCodeAt(i);
+
+        #{yield `code & 0xff`};
+        #{yield `code >> 8`};
+      }
+    }
+  end
+
+  def bytesize
+    bytes.length
+  end
+end
+
+Encoding.register "ASCII-8BIT", aliases: ["BINARY"], ascii: true do
+  def each_byte(string, &block)
+    %x{
+      for (var i = 0, length = string.length; i < length; i++) {
+        #{yield `string.charCodeAt(i) & 0xff`};
+      }
+    }
+  end
+
+  def bytesize
+    bytes.length
+  end
+end
+
+class String
+  `def.encoding = #{Encoding::UTF_16LE}`
+end
diff --git a/corelib/opal.rb b/corelib/opal.rb
@@ -14,6 +14,7 @@
 require 'hash'
 require 'string'
 require 'match_data'
+require 'encoding'
 require 'numeric'
 require 'proc'
 require 'range'

diff --git a/corelib/string.rb b/corelib/string.rb
@@ -142,6 +142,14 @@ def [](index, length = undefined)
     }
   end
 
+  def bytes
+    each_byte.to_a
+  end
+
+  def bytesize
+    @encoding.bytesize(self)
+  end
+
   def capitalize
     `self.charAt(0).toUpperCase() + self.substr(1).toLowerCase()`
   end
@@ -227,6 +235,14 @@ def downcase
     `self.toLowerCase()`
   end
 
+  def each_byte(&block)
+    return enum_for :each_byte unless block_given?
+
+    @encoding.each_byte(self, &block)
+
+    self
+  end
+
   def each_char(&block)
     return enum_for :each_char unless block_given?
 
@@ -264,6 +280,10 @@ def empty?
     `self.length === 0`
   end
 
+  def encoding
+    @encoding
+  end
+
   def end_with?(*suffixes)
     %x{
       for (var i = 0, length = suffixes.length; i < length; i++) {
@@ -281,8 +301,21 @@ def end_with?(*suffixes)
   alias eql? ==
   alias equal? ===
 
+  def force_encoding(encoding)
+    encoding = Encoding.find(encoding)
+
+    return self if encoding == @encoding
+
+    %x{
+      var result = new native_string(self);
+      result.encoding = encoding;
+
+      return result;
+    }
+  end
+
   def getbyte(idx)
-    `#{self}.charCodeAt(idx)`
+    @encoding.getbyte(self, idx)
   end
 
   def gsub(pattern, replace = undefined, &block)