Freeze all instances of Regexp, except subclass instances

* Instances of a subclass of Regexp are not frozen for compatibility.
* [Feature #8948]
* Use a less confusing example in test_regexp2,
  the ivar was named @encoding but had no effect on Regexp#encoding.

Author: eregon

core/marshal/dump_spec.rb

@@ -473,14 +473,26 @@ def _dump(level)
       Marshal.dump(//im).should == "\x04\bI/\x00\x05\x06:\x06EF"
     end
 
-    it "dumps a Regexp with instance variables" do
-      o = Regexp.new("")
+    it "dumps a Regexp subclass with instance variables" do
+      o = UserRegexp.new("")
       o.instance_variable_set(:@ivar, :ivar)
-      Marshal.dump(o).should == "\x04\bI/\x00\x00\a:\x06EF:\n@ivar:\tivar"
+      Marshal.dump(o).should == "\x04\bIC:\x0FUserRegexp/\x00\x00\a:\x06EF:\n@ivar:\tivar"
     end
 
-    it "dumps an extended Regexp" do
-      Marshal.dump(Regexp.new("").extend(Meths)).should == "\x04\bIe:\nMeths/\x00\x00\x06:\x06EF"
+    it "dumps an extended Regexp subclass" do
+      Marshal.dump(UserRegexp.new("").extend(Meths)).should == "\x04\bIe:\nMethsC:\x0FUserRegexp/\x00\x00\x06:\x06EF"
+    end
+
+    ruby_version_is ""..."4.1" do
+      it "dumps a Regexp with instance variables" do
+        o = Regexp.new("")
+        o.instance_variable_set(:@ivar, :ivar)
+        Marshal.dump(o).should == "\x04\bI/\x00\x00\a:\x06EF:\n@ivar:\tivar"
+      end
+
+      it "dumps an extended Regexp" do
+        Marshal.dump(Regexp.new("").extend(Meths)).should == "\x04\bIe:\nMeths/\x00\x00\x06:\x06EF"
+      end
     end
 
     it "dumps a Regexp subclass" do

core/marshal/shared/load.rb

@@ -901,14 +901,52 @@ def io.binmode; raise "binmode"; end
   end
 
   describe "for a Regexp" do
-    it "loads an extended Regexp" do
-      obj = /[a-z]/.dup.extend(Meths, MethsMore)
-      new_obj = Marshal.send(@method, "\004\be:\nMethse:\016MethsMore/\n[a-z]\000")
+    ruby_version_is "4.1" do
+      it "raises FrozenError for an extended Regexp" do
+        -> {
+          Marshal.send(@method, "\004\be:\nMethse:\016MethsMore/\n[a-z]\000")
+        }.should raise_error(FrozenError)
+      end
+
+      it "raises FrozenError when regexp has instance variables" do
+        -> {
+          Marshal.send(@method, "\x04\bI/\nhello\x00\a:\x06EF:\x11@regexp_ivar[\x06i/")
+        }.should raise_error(FrozenError)
+      end
+    end
+
+    ruby_version_is ""..."4.1" do
+      it "loads an extended Regexp" do
+        obj = /[a-z]/.dup.extend(Meths, MethsMore)
+        new_obj = Marshal.send(@method, "\004\be:\nMethse:\016MethsMore/\n[a-z]\000")
+
+        new_obj.should == obj
+        new_obj_metaclass_ancestors = class << new_obj; ancestors; end
+        new_obj_metaclass_ancestors[@num_self_class, 3].should ==
+          [Meths, MethsMore, Regexp]
+      end
+
+      it "restore the regexp instance variables" do
+        obj = Regexp.new("hello")
+        obj.instance_variable_set(:@regexp_ivar, [42])
+
+        new_obj = Marshal.send(@method, "\x04\bI/\nhello\x00\a:\x06EF:\x11@regexp_ivar[\x06i/")
+        new_obj.instance_variables.should == [:@regexp_ivar]
+        new_obj.instance_variable_get(:@regexp_ivar).should == [42]
+      end
+    end
+
+    it "loads a Regexp subclass instance variables" do
+      obj = UserRegexp.new('abc')
+      obj.instance_variable_set(:@noise, 'much')
+
+      new_obj = Marshal.send(@method, Marshal.dump(obj))
 
       new_obj.should == obj
+      new_obj.instance_variable_get(:@noise).should == 'much'
       new_obj_metaclass_ancestors = class << new_obj; ancestors; end
-      new_obj_metaclass_ancestors[@num_self_class, 3].should ==
-        [Meths, MethsMore, Regexp]
+      new_obj_metaclass_ancestors[@num_self_class, 2].should ==
+        [UserRegexp, Regexp]
     end
 
     it "loads a Regexp subclass instance variables when it is extended with a module" do
@@ -924,15 +962,6 @@ def io.binmode; raise "binmode"; end
         [Meths, UserRegexp, Regexp]
     end
 
-    it "restore the regexp instance variables" do
-      obj = Regexp.new("hello")
-      obj.instance_variable_set(:@regexp_ivar, [42])
-
-      new_obj = Marshal.send(@method, "\x04\bI/\nhello\x00\a:\x06EF:\x11@regexp_ivar[\x06i/")
-      new_obj.instance_variables.should == [:@regexp_ivar]
-      new_obj.instance_variable_get(:@regexp_ivar).should == [42]
-    end
-
     it "preserves Regexp encoding" do
       source_object = Regexp.new("a".encode("utf-32le"))
       regexp = Marshal.send(@method, Marshal.dump(source_object))

core/regexp/initialize_spec.rb

@@ -9,7 +9,21 @@
     -> { //.send(:initialize, "") }.should raise_error(FrozenError)
   end
 
-  it "raises a TypeError on an initialized non-literal Regexp" do
-    -> { Regexp.new("").send(:initialize, "") }.should raise_error(TypeError)
+  ruby_version_is "4.1" do
+    it "raises a FrozenError on an initialized non-literal Regexp" do
+      regexp = Regexp.new("")
+      -> { regexp.send(:initialize, "") }.should raise_error(FrozenError)
+    end
+  end
+
+  ruby_version_is ""..."4.1" do
+    it "raises a TypeError on an initialized non-literal Regexp" do
+      -> { Regexp.new("").send(:initialize, "") }.should raise_error(TypeError)
+    end
+  end
+
+  it "raises a TypeError on an initialized non-literal Regexp subclass" do
+    r = Class.new(Regexp).new("")
+    -> { r.send(:initialize, "") }.should raise_error(TypeError)
   end
 end

core/regexp/shared/new.rb

@@ -5,6 +5,12 @@
     Regexp.send(@method, '').is_a?(Regexp).should == true
   end
 
+  ruby_version_is "4.1" do
+    it "is frozen" do
+      Regexp.send(@method, '').should.frozen?
+    end
+  end
+
   it "works by default for subclasses with overridden #initialize" do
     class RegexpSpecsSubclass < Regexp
       def initialize(*args)

core/string/match_spec.rb

@@ -137,9 +137,17 @@ def obj.method_missing(*args) "." end
   end
 
   it "calls match on the regular expression" do
-    regexp = /./.dup
-    regexp.should_receive(:match).and_return(:foo)
-    'hello'.match(regexp).should == :foo
+    # Can't use regexp.should_receive(:match).and_return(:foo) since regexps are frozen
+    ScratchPad.clear
+    regexp = Class.new(Regexp) {
+      def match(*args)
+        ScratchPad.record [:match, *args]
+        super(*args)
+      end
+    }.new('.')
+
+    'hello'.match(regexp)
+    ScratchPad.recorded.should == [:match, 'hello']
   end
 end
 

optional/capi/encoding_spec.rb

@@ -489,12 +489,20 @@
       @s.rb_enc_copy("string", @obj).encoding.should == Encoding::US_ASCII
     end
 
-    it "raises a RuntimeError if the second argument is a Symbol" do
+    it "raises a RuntimeError if the first argument is a Symbol" do
       -> { @s.rb_enc_copy(:symbol, @obj) }.should raise_error(RuntimeError)
     end
 
-    it "sets the encoding of a Regexp to that of the second argument" do
-      @s.rb_enc_copy(/regexp/.dup, @obj).encoding.should == Encoding::US_ASCII
+    ruby_version_is "4.1" do
+      it "raises a FrozenError if the first argument is a Regexp" do
+        -> { @s.rb_enc_copy(/regexp/.dup, @obj) }.should raise_error(FrozenError)
+      end
+    end
+
+    ruby_version_is ""..."4.1" do
+      it "sets the encoding of a Regexp to that of the second argument" do
+        @s.rb_enc_copy(/regexp/.dup, @obj).encoding.should == Encoding::US_ASCII
+      end
     end
   end
 
@@ -544,8 +552,16 @@
       -> { @s.rb_enc_associate(:symbol, "US-ASCII") }.should raise_error(RuntimeError)
     end
 
-    it "sets the encoding of a Regexp to the encoding" do
-      @s.rb_enc_associate(/regexp/.dup, "BINARY").encoding.should == Encoding::BINARY
+    ruby_version_is "4.1" do
+      it "raises a FrozenError if the argument is a Regexp" do
+        -> { @s.rb_enc_associate(/regexp/.dup, "BINARY") }.should raise_error(FrozenError)
+      end
+    end
+
+    ruby_version_is ""..."4.1" do
+      it "sets the encoding of a Regexp to the encoding" do
+        @s.rb_enc_associate(/regexp/.dup, "BINARY").encoding.should == Encoding::BINARY
+      end
     end
 
     it "sets the encoding of a String to a default when the encoding is NULL" do
@@ -560,10 +576,19 @@
       enc.should == Encoding::BINARY
     end
 
-    it "sets the encoding of a Regexp to the encoding" do
-      index = @s.rb_enc_find_index("UTF-8")
-      enc = @s.rb_enc_associate_index(/regexp/.dup, index).encoding
-      enc.should == Encoding::UTF_8
+    ruby_version_is "4.1" do
+      it "raises a FrozenError if the argument is a Regexp" do
+        index = @s.rb_enc_find_index("UTF-8")
+        -> { @s.rb_enc_associate_index(/regexp/.dup, index) }.should raise_error(FrozenError)
+      end
+    end
+
+    ruby_version_is ""..."4.1" do
+      it "sets the encoding of a Regexp to the encoding" do
+        index = @s.rb_enc_find_index("UTF-8")
+        enc = @s.rb_enc_associate_index(/regexp/.dup, index).encoding
+        enc.should == Encoding::UTF_8
+      end
     end
 
     it "sets the encoding of a Symbol to the encoding" do