Merge pull request #2243 from herwinw/marshal_string_encoding

Fixes for Marshal.load and Marshal.dump related to binary strings

spec/core/marshal/dump_spec.rb

@@ -319,21 +319,15 @@ def _dump(level)
 
   describe "with a String" do
     it "dumps a blank String" do
-      NATFIXME 'dumps a blank String', exception: SpecFailedException do
-        Marshal.dump("".dup.force_encoding("binary")).should == "\004\b\"\000"
-      end
+      Marshal.dump("".dup.force_encoding("binary")).should == "\004\b\"\000"
     end
 
     it "dumps a short String" do
-      NATFIXME 'dumps a short String', exception: SpecFailedException do
-        Marshal.dump("short".dup.force_encoding("binary")).should == "\004\b\"\012short"
-      end
+      Marshal.dump("short".dup.force_encoding("binary")).should == "\004\b\"\012short"
     end
 
     it "dumps a long String" do
-      NATFIXME 'dumps a long String', exception: SpecFailedException do
-        Marshal.dump(("big" * 100).force_encoding("binary")).should == "\004\b\"\002,\001#{"big" * 100}"
-      end
+      Marshal.dump(("big" * 100).force_encoding("binary")).should == "\004\b\"\002,\001#{"big" * 100}"
     end
 
     it "dumps a String extended with a Module" do

spec/core/marshal/shared/load.rb

@@ -484,9 +484,7 @@
     arr = ArraySubPush.new
     arr[0] = '1'
     arr_dump = Marshal.dump(arr)
-    NATFIXME 'loads subclasses of Array with overridden << and push correctly', exception: ArgumentError, message: 'marshal data too short' do
-      Marshal.send(@method, arr_dump).should == arr
-    end
+    Marshal.send(@method, arr_dump).should == arr
   end
 
   it "raises a TypeError with bad Marshal version" do
@@ -612,7 +610,7 @@
       h = { key: s }
       h.instance_variable_set :@hash_ivar, 'hash ivar'
 
-      NATFIXME 'preserves hash ivars when hash contains a string having ivar', exception: ArgumentError, message: 'marshal data too short' do
+      NATFIXME 'preserves hash ivars when hash contains a string having ivar', exception: SpecFailedException do
         unmarshalled = Marshal.send(@method, Marshal.dump(h))
         unmarshalled.instance_variable_get(:@hash_ivar).should == 'hash ivar'
         unmarshalled[:key].instance_variable_get(:@string_ivar).should == 'string ivar'
@@ -746,9 +744,7 @@
     it "loads a string through StringIO stream" do
       require 'stringio'
       obj = "This is a string which should be unmarshalled through StringIO stream!"
-      NATFIXME 'loads a string through StringIO stream', exception: ArgumentError, message: 'marshal data too short' do
-        Marshal.send(@method, StringIO.new(Marshal.dump(obj))).should == obj
-      end
+      Marshal.send(@method, StringIO.new(Marshal.dump(obj))).should == obj
     end
 
     it "sets binmode if it is loading through StringIO stream" do
@@ -802,21 +798,17 @@ def io.binmode; raise "binmode"; end
     it "loads a String as BINARY if no encoding is specified at the end" do
       str = "\xC3\xB8".dup.force_encoding("BINARY")
       data = "\x04\b\"\a\xC3\xB8".dup.force_encoding("UTF-8")
-      NATFIXME 'loads a String as BINARY if no encoding is specified at the end', exception: ArgumentError, message: 'marshal data too short' do
-        result = Marshal.send(@method, data)
-        result.encoding.should == Encoding::BINARY
-        result.should == str
-      end
+      result = Marshal.send(@method, data)
+      result.encoding.should == Encoding::BINARY
+      result.should == str
     end
 
     it "raises ArgumentError when end of byte sequence reached before string characters end" do
-      NATFIXME 'raises ArgumentError when end of byte sequence reached before string characters end', exception: SpecFailedException do
-        Marshal.dump("hello").should == "\x04\b\"\nhello"
+      Marshal.dump("hello").should == "\x04\b\"\nhello"
 
-        -> {
-          Marshal.send(@method, "\x04\b\"\nhel")
-        }.should raise_error(ArgumentError, "marshal data too short")
-      end
+      -> {
+        Marshal.send(@method, "\x04\b\"\nhel")
+      }.should raise_error(ArgumentError, "marshal data too short")
     end
   end
 

src/marshal.rb

@@ -140,10 +140,10 @@ def write_integer(value)
     end
 
     def write_string(value)
-      write_char('I')
+      write_char('I') if value.encoding != Encoding::ASCII_8BIT
       write_char('"')
       write_string_bytes(value)
-      write_encoding_bytes(value)
+      write_encoding_bytes(value) if value.encoding != Encoding::ASCII_8BIT
     end
 
     def write_symbol(value)
@@ -565,10 +565,10 @@ def read_byte
     end
 
     def read_bytes(integer)
-      if @source.length - @offset >= integer
-        string = @source.slice(@offset, integer)
+      if @source.bytesize - @offset >= integer
+        string = @source.byteslice(@offset, integer)
         @offset += integer
-        string
+        string.b
       else
         raise ArgumentError, 'marshal data too short'
       end