Fix Marshal.load and preserves Regexp source encoding

Author: andrykonchin

spec/ruby/core/marshal/shared/load.rb

@@ -776,7 +776,7 @@
         [Meths, MethsMore, Regexp]
     end
 
-    it "loads a extended_user_regexp having ivar" do
+    it "loads a Regexp subclass instance variables when it is extended with a module" do
       obj = UserRegexp.new('').extend(Meths)
       obj.instance_variable_set(:@noise, 'much')
 
@@ -799,6 +799,14 @@
         new_obj.instance_variable_get(:@regexp_ivar).should == [42]
       end
     end
+
+    it "preserves Regexp encoding" do
+      source_object = Regexp.new("a".encode("utf-32le"))
+      regexp = Marshal.send(@method, Marshal.dump(source_object))
+
+      regexp.encoding.should == Encoding::UTF_32LE
+      regexp.source.should == "a".encode("utf-32le")
+    end
   end
 
   describe "for a Float" do

spec/tags/core/marshal/load_tags.txt

@@ -1,8 +1,5 @@
-fails:Marshal.load loads an array containing objects having _dump method, and with proc
-fails:Marshal.load loads an array containing objects having marshal_dump method, and with proc
 fails:Marshal.load loads a Random
 fails:Marshal.load when called with a proc returns the value of the proc
-fails:Marshal.load when called with a proc calls the proc for recursively visited data
 fails:Marshal.load when called with a proc loads an Array with proc
 fails:Marshal.load for an Array loads an array containing the same objects
 fails:Marshal.load for an Object raises ArgumentError if the object from an 'o' stream is not dumpable as 'o' type user class
@@ -23,3 +20,4 @@ fails:Marshal.load when called with freeze: true when called with a proc does no
 fails:Marshal.load when called with freeze: true does freeze extended objects
 fails:Marshal.load when called with freeze: true does freeze extended objects with instance variables
 fails:Marshal.load for a Regexp restore the regexp instance variables
+fails:Marshal.load for a Regexp loads a Regexp subclass instance variables when it is extended with a module

spec/tags/core/marshal/restore_tags.txt

@@ -1,8 +1,5 @@
-fails:Marshal.restore loads an array containing objects having _dump method, and with proc
-fails:Marshal.restore loads an array containing objects having marshal_dump method, and with proc
 fails:Marshal.restore loads a Random
 fails:Marshal.restore when called with a proc returns the value of the proc
-fails:Marshal.restore when called with a proc calls the proc for recursively visited data
 fails:Marshal.restore when called with a proc loads an Array with proc
 fails:Marshal.restore for an Array loads an array containing the same objects
 fails:Marshal.restore for an Object raises ArgumentError if the object from an 'o' stream is not dumpable as 'o' type user class
@@ -23,3 +20,4 @@ fails:Marshal.restore when called with freeze: true when called with a proc does
 fails:Marshal.restore when called with freeze: true does freeze extended objects
 fails:Marshal.restore when called with freeze: true does freeze extended objects with instance variables
 fails:Marshal.restore for a Regexp restore the regexp instance variables
+fails:Marshal.restore for a Regexp loads a Regexp subclass instance variables when it is extended with a module

src/main/java/org/truffleruby/core/regexp/RubyRegexp.java

@@ -75,9 +75,6 @@ public static RubyRegexp create(RubyLanguage language,
     public final TRegexCache tregexCache;
 
     private RubyRegexp(Regex regex, RegexpOptions options) {
-        // The RegexpNodes.compile operation may modify the encoding of the source rope. This modified copy is stored
-        // in the Regex object as the "user object". Since ropes are immutable, we need to take this updated copy when
-        // constructing the final regexp.
         this.regex = regex;
         final TStringWithEncoding tstringWithEncoding = (TStringWithEncoding) regex.getUserObject();
         this.source = tstringWithEncoding.tstring;

src/main/ruby/truffleruby/core/marshal.rb

@@ -641,7 +641,7 @@ def construct(ivar_index = nil, call_proc = true)
             when 34   # ?"
               construct_string
             when 47   # ?/
-              construct_regexp
+              construct_regexp(ivar_index)
             when 91   # ?[
               construct_array
             when 123  # ?{
@@ -903,13 +903,23 @@ def construct_integer
       end
     end
 
-    def construct_regexp
-      s = get_byte_sequence
+    def construct_regexp(ivar_index)
+      source = get_byte_sequence
+      options = consume_byte
+
+      # A Regexp instance variables are ignored by CRuby,
+      # but we need to know the encoding before building the Regexp
+      if ivar_index and @has_ivar[ivar_index]
+        # This sets the encoding of the String
+        set_instance_variables source
+        @has_ivar[ivar_index] = false
+      end
+
       if user_class?
-        obj = user_class.new s, consume_byte
+        obj = user_class.new source, options
         clear_user_class
       else
-        obj = Regexp.new s, consume_byte
+        obj = Regexp.new source, options
       end
 
       store_unique_object obj