bert 1.0.0 → 1.1.0

data/ext/bert/c/extconf.rb

@@ -0,0 +1,11 @@
+# Loads mkmf which is used to make makefiles for Ruby extensions
+require 'mkmf'
+
+# Give it a name
+extension_name = 'decode'
+
+# The destination
+dir_config(extension_name)
+
+# Do the work
+create_makefile(extension_name)

data/lib/bert.rb

@@ -1,33 +1,24 @@
 require 'rubygems'
-require 'erlectricity'
 
-require 'bert/encoder'
-require 'bert/decoder'
+$:.unshift File.join(File.dirname(__FILE__), *%w[.. ext])
 
-module BERT
-  def self.encode(ruby)
-    Encoder.encode(ruby)
-  end
+require 'bert/bert'
+require 'bert/types'
 
-  def self.decode(bert)
-    Decoder.decode(bert)
-  end
-
-  def self.ebin(str)
-    bytes = []
-    str.each_byte { |b| bytes << b.to_s }
-    "<<" + bytes.join(',') + ">>"
-  end
+begin
+  # try to load the C extension
+  require 'bert/c/decode'
+rescue LoadError
+  # fall back on the pure ruby version
+  require 'bert/decode'
 end
 
-module BERT
-  class Tuple < Array
-    def inspect
-      "t#{super}"
-    end
-  end
-end
+require 'bert/encode'
+
+require 'bert/encoder'
+require 'bert/decoder'
 
+# Global method for specifying that an array should be encoded as a tuple.
 def t
   BERT::Tuple
 end

data/lib/bert/bert.rb

@@ -0,0 +1,21 @@
+module BERT
+  def self.encode(ruby)
+    Encoder.encode(ruby)
+  end
+
+  def self.decode(bert)
+    Decoder.decode(bert)
+  end
+
+  def self.ebin(str)
+    bytes = []
+    str.each_byte { |b| bytes << b.to_s }
+    "<<" + bytes.join(',') + ">>"
+  end
+
+  class Tuple < Array
+    def inspect
+      "t#{super}"
+    end
+  end
+end

data/lib/bert/decode.rb

@@ -0,0 +1,248 @@
+module BERT
+  class Decode
+    attr_accessor :in
+    include Types
+
+    def self.impl
+      'Ruby'
+    end
+
+    def self.decode(string)
+      new(StringIO.new(string)).read_any
+    end
+
+    def initialize(ins)
+      @in = ins
+      @peeked = ""
+    end
+
+    def read_any
+      fail("Bad Magic") unless read_1 == MAGIC
+      read_any_raw
+    end
+
+    def read_any_raw
+      case peek_1
+        when ATOM then read_atom
+        when SMALL_INT then read_small_int
+        when INT then read_int
+        when SMALL_BIGNUM then read_small_bignum
+        when LARGE_BIGNUM then read_large_bignum
+        when FLOAT then read_float
+        when SMALL_TUPLE then read_small_tuple
+        when LARGE_TUPLE then read_large_tuple
+        when NIL then read_nil
+        when STRING then read_erl_string
+        when LIST then read_list
+        when BIN then read_bin
+        else
+          fail("Unknown term tag: #{peek_1}")
+      end
+    end
+
+    def read(length)
+      if length < @peeked.length
+        result = @peeked[0...length]
+        @peeked = @peeked[length..-1]
+        length = 0
+      else
+        result = @peeked
+        @peeked = ''
+        length -= result.length
+      end
+
+      if length > 0
+        result << @in.read(length)
+      end
+      result
+    end
+
+    def peek(length)
+      if length <= @peeked.length
+        @peeked[0...length]
+      else
+        read_bytes = @in.read(length - @peeked.length)
+        @peeked << read_bytes if read_bytes
+        @peeked
+      end
+    end
+
+    def peek_1
+      peek(1).unpack("C").first
+    end
+
+    def peek_2
+      peek(2).unpack("n").first
+    end
+
+    def read_1
+      read(1).unpack("C").first
+    end
+
+    def read_2
+      read(2).unpack("n").first
+    end
+
+    def read_4
+      read(4).unpack("N").first
+    end
+
+    def read_string(length)
+      read(length)
+    end
+
+    def read_atom
+      fail("Invalid Type, not an atom") unless read_1 == ATOM
+      length = read_2
+      a = read_string(length)
+      case a
+        when ""
+          Marshal.load("\004\b:\005") # Workaround for inability to do ''.to_sym
+        else
+          a.to_sym
+      end
+    end
+
+    def read_small_int
+      fail("Invalid Type, not a small int") unless read_1 == SMALL_INT
+      read_1
+    end
+
+    def read_int
+      fail("Invalid Type, not an int") unless read_1 == INT
+      value = read_4
+      negative = (value >> 31)[0] == 1
+      value = (value - (1 << 32)) if negative
+      value = Fixnum.induced_from(value)
+    end
+
+    def read_small_bignum
+      fail("Invalid Type, not a small bignum") unless read_1 == SMALL_BIGNUM
+      size = read_1
+      sign = read_1
+      bytes = read_string(size).unpack("C" * size)
+      added = bytes.zip((0..bytes.length).to_a).inject(0) do |result, byte_index|
+        byte, index = *byte_index
+        value = (byte * (256 ** index))
+        sign != 0 ? (result - value) : (result + value)
+      end
+      Bignum.induced_from(added)
+    end
+
+    def read_large_bignum
+      fail("Invalid Type, not a large bignum") unless read_1 == LARGE_BIGNUM
+      size = read_4
+      sign = read_1
+      bytes = read_string(size).unpack("C" * size)
+      added = bytes.zip((0..bytes.length).to_a).inject(0) do |result, byte_index|
+        byte, index = *byte_index
+        value = (byte * (256 ** index))
+        sign != 0 ? (result - value) : (result + value)
+      end
+      Bignum.induced_from(added)
+    end
+
+    def read_float
+      fail("Invalid Type, not a float") unless read_1 == FLOAT
+      string_value = read_string(31)
+      result = string_value.to_f
+    end
+
+    def read_small_tuple
+      fail("Invalid Type, not a small tuple") unless read_1 == SMALL_TUPLE
+      read_tuple(read_1)
+    end
+
+    def read_large_tuple
+      fail("Invalid Type, not a small tuple") unless read_1 == LARGE_TUPLE
+      read_tuple(read_4)
+    end
+
+    def read_tuple(arity)
+      if arity > 0
+        tag = read_any_raw
+        if tag == :bert
+          read_complex_type(arity)
+        else
+          tuple = Tuple.new(arity)
+          tuple[0] = tag
+          (arity - 1).times { |i| tuple[i + 1] = read_any_raw }
+          tuple
+        end
+      else
+        Tuple.new
+      end
+    end
+
+    def read_complex_type(arity)
+      case read_any_raw
+        when :nil
+          nil
+        when :true
+          true
+        when :false
+          false
+        when :time
+          Time.at(read_any_raw * 1_000_000 + read_any_raw, read_any_raw)
+        when :regex
+          source = read_any_raw
+          opts = read_any_raw
+          options = 0
+          options |= Regexp::EXTENDED if opts.include?(:extended)
+          options |= Regexp::IGNORECASE if opts.include?(:caseless)
+          options |= Regexp::MULTILINE if opts.include?(:multiline)
+          Regexp.new(source, options)
+        when :dict
+          read_dict
+        else
+          nil
+      end
+    end
+
+    def read_dict
+      type = read_1
+      fail("Invalid dict spec, not an erlang list") unless [LIST, NIL].include?(type)
+      if type == LIST
+        length = read_4
+      else
+        length = 0
+      end
+      hash = {}
+      length.times do |i|
+        pair = read_any_raw
+        hash[pair[0]] = pair[1]
+      end
+      read_1 if type == LIST
+      hash
+    end
+
+    def read_nil
+      fail("Invalid Type, not a nil list") unless read_1 == NIL
+      []
+    end
+
+    def read_erl_string
+      fail("Invalid Type, not an erlang string") unless read_1 == STRING
+      length = read_2
+      read_string(length).unpack('C' * length)
+    end
+
+    def read_list
+      fail("Invalid Type, not an erlang list") unless read_1 == LIST
+      length = read_4
+      list = (0...length).map { |i| read_any_raw }
+      read_1
+      list
+    end
+
+    def read_bin
+      fail("Invalid Type, not an erlang binary") unless read_1 == BIN
+      length = read_4
+      read_string(length)
+    end
+
+    def fail(str)
+      raise str
+    end
+  end
+end

data/lib/bert/decoder.rb

@@ -5,59 +5,7 @@ module BERT
     #
     # Returns a Ruby object
     def self.decode(bert)
-      simple_ruby = Erlectricity::Decoder.decode(bert)
-      convert(simple_ruby)
-    end
-
-    # Convert Erlectricity representation of BERT complex types into
-    # corresponding Ruby types.
-    #   +item+ is the Ruby object to convert
-    #
-    # Returns the converted Ruby object
-    def self.convert(item)
-      case item
-        when TrueClass, FalseClass
-          item.to_s.to_sym
-        when Erl::List
-          item.map { |x| convert(x) }
-        when Array
-          if item[0] == :bert
-            convert_bert(item)
-          else
-            Tuple.new(item.map { |x| convert(x) })
-          end
-        else
-          item
-      end
-    end
-
-    # Convert complex types.
-    #   +item+ is the complex type array
-    #
-    # Returns the converted Ruby object
-    def self.convert_bert(item)
-      case item[1]
-        when :nil
-          nil
-        when :dict
-          item[2].inject({}) do |acc, x|
-            acc[convert(x[0])] = convert(x[1]); acc
-          end
-        when TrueClass
-          true
-        when FalseClass
-          false
-        when :time
-          Time.at(item[2] * 1_000_000 + item[3], item[4])
-        when :regex
-          options = 0
-          options |= Regexp::EXTENDED if item[3].include?(:extended)
-          options |= Regexp::IGNORECASE if item[3].include?(:caseless)
-          options |= Regexp::MULTILINE if item[3].include?(:multiline)
-          Regexp.new(item[2], options)
-        else
-          nil
-      end
+      Decode.decode(bert)
     end
   end
 end

data/lib/bert/encode.rb

@@ -0,0 +1,142 @@
+module BERT
+  class Encode
+    include Types
+
+    attr_accessor :out
+
+    def initialize(out)
+      self.out = out
+    end
+
+    def self.encode(data)
+      io = StringIO.new
+      self.new(io).write_any(data)
+      io.string
+    end
+
+    def write_any obj
+      write_1 MAGIC
+      write_any_raw obj
+    end
+
+    def write_any_raw obj
+      case obj
+        when Symbol then write_symbol(obj)
+        when Fixnum, Bignum then write_fixnum(obj)
+        when Float then write_float(obj)
+        when Tuple then write_tuple(obj)
+        when Array then write_list(obj)
+        when String then write_binary(obj)
+        else
+          fail(obj)
+      end
+    end
+
+    def write_1(byte)
+      out.write([byte].pack("C"))
+    end
+
+    def write_2(short)
+      out.write([short].pack("n"))
+    end
+
+    def write_4(long)
+      out.write([long].pack("N"))
+    end
+
+    def write_string(string)
+      out.write(string)
+    end
+
+    def write_boolean(bool)
+      write_symbol(bool.to_s.to_sym)
+    end
+
+    def write_symbol(sym)
+      fail(sym) unless sym.is_a?(Symbol)
+      data = sym.to_s
+      write_1 ATOM
+      write_2 data.length
+      write_string data
+    end
+
+    def write_fixnum(num)
+      if num >= 0 && num < 256
+        write_1 SMALL_INT
+        write_1 num
+      elsif num <= MAX_INT && num >= MIN_INT
+        write_1 INT
+        write_4 num
+      else
+        write_bignum num
+      end
+    end
+
+    def write_float(float)
+      write_1 FLOAT
+      write_string format("%15.15e", float).ljust(31, "\000")
+    end
+
+    def write_bignum(num)
+      if num.is_a?(Bignum)
+        n = num.size
+      else
+        n = (num.to_s(2).size / 8.0).ceil
+      end
+      if n <= 256
+        write_1 SMALL_BIGNUM
+        write_1 n
+        write_bignum_guts(num)
+      else
+        write_1 LARGE_BIGNUM
+        write_4 n
+        write_bignum_guts(num)
+      end
+    end
+
+    def write_bignum_guts(num)
+      write_1 (num >= 0 ? 0 : 1)
+      num = num.abs
+      while num != 0
+        rem = num % 256
+        write_1 rem
+        num = num >> 8
+      end
+    end
+
+    def write_tuple(data)
+      fail(data) unless data.is_a? Array
+
+      if data.length < 256
+        write_1 SMALL_TUPLE
+        write_1 data.length
+      else
+        write_1 LARGE_TUPLE
+        write_4 data.length
+      end
+
+      data.each { |e| write_any_raw e }
+    end
+
+    def write_list(data)
+      fail(data) unless data.is_a? Array
+      write_1 NIL and return if data.empty?
+      write_1 LIST
+      write_4 data.length
+      data.each{|e| write_any_raw e }
+      write_1 NIL
+    end
+
+    def write_binary(data)
+      write_1 BIN
+      write_4 data.length
+      write_string data
+    end
+
+    private
+
+    def fail(obj)
+      raise "Cannot encode to erlang external format: #{obj.inspect}"
+    end
+  end
+end