1 files changed, 60 insertions, 4 deletions

diff --git a/spec/ruby/core/string/scrub_spec.rb b/spec/ruby/core/string/scrub_spec.rb
index 86fd4e85ba..9dc55dbef7 100644
--- a/spec/ruby/core/string/scrub_spec.rb
+++ b/spec/ruby/core/string/scrub_spec.rb
@@ -1,5 +1,7 @@
 # -*- encoding: utf-8 -*-
+# frozen_string_literal: false
 require_relative '../../spec_helper'
+require_relative 'fixtures/classes'
 
 describe "String#scrub with a default replacement" do
   it "returns self for valid strings" do
@@ -13,18 +15,33 @@ describe "String#scrub with a default replacement" do
     "abc\u3042#{x81}".scrub.should == "abc\u3042\uFFFD"
   end
 
+  it "replaces invalid byte sequences in lazy substrings" do
+    x81 = [0x81].pack('C').force_encoding('utf-8')
+    "abc\u3042#{x81}def"[1...-1].scrub.should == "bc\u3042\uFFFDde"
+  end
+
   it "returns a copy of self when the input encoding is BINARY" do
     input = "foo".encode('BINARY')
 
     input.scrub.should == "foo"
   end
 
-
   it "replaces invalid byte sequences when using ASCII as the input encoding" do
     xE3x80 = [0xE3, 0x80].pack('CC').force_encoding 'utf-8'
     input = "abc\u3042#{xE3x80}".force_encoding('ASCII')
     input.scrub.should == "abc?????"
   end
+
+  it "returns a String in the same encoding as self" do
+    x81 = [0x81].pack('C').force_encoding('utf-8')
+    "abc\u3042#{x81}".scrub.encoding.should == Encoding::UTF_8
+  end
+
+  it "returns String instances when called on a subclass" do
+    StringSpecs::MyString.new("foo").scrub.should.instance_of?(String)
+    input = [0x81].pack('C').force_encoding('utf-8')
+    StringSpecs::MyString.new(input).scrub.should.instance_of?(String)
+  end
 end
 
 describe "String#scrub with a custom replacement" do
@@ -42,7 +59,9 @@ describe "String#scrub with a custom replacement" do
   it "replaces invalid byte sequences in frozen strings" do
     x81 = [0x81].pack('C').force_encoding('utf-8')
     (-"abc\u3042#{x81}").scrub("*").should == "abc\u3042*"
-    utf16_str = ("abc".encode('UTF-16LE').bytes + [0x81]).pack('c*').force_encoding('UTF-16LE')
+
+    leading_surrogate = [0x00, 0xD8]
+    utf16_str = ("abc".encode('UTF-16LE').bytes + leading_surrogate).pack('c*').force_encoding('UTF-16LE')
     (-(utf16_str)).scrub("*".encode('UTF-16LE')).should == "abc*".encode('UTF-16LE')
   end
 
@@ -56,14 +75,25 @@ describe "String#scrub with a custom replacement" do
     xE4 = [0xE4].pack('C').force_encoding('utf-8')
     block = -> { "foo#{x81}".scrub(xE4) }
 
-    block.should raise_error(ArgumentError)
+    block.should.raise(ArgumentError)
+  end
+
+  it "returns a String in the same encoding as self" do
+    x81 = [0x81].pack('C').force_encoding('utf-8')
+    "abc\u3042#{x81}".scrub("*").encoding.should == Encoding::UTF_8
   end
 
   it "raises TypeError when a non String replacement is given" do
     x81 = [0x81].pack('C').force_encoding('utf-8')
     block = -> { "foo#{x81}".scrub(1) }
 
-    block.should raise_error(TypeError)
+    block.should.raise(TypeError)
+  end
+
+  it "returns String instances when called on a subclass" do
+    StringSpecs::MyString.new("foo").scrub("*").should.instance_of?(String)
+    input = [0x81].pack('C').force_encoding('utf-8')
+    StringSpecs::MyString.new(input).scrub("*").should.instance_of?(String)
   end
 end
 
@@ -89,6 +119,12 @@ describe "String#scrub with a block" do
 
     replaced.should == "€€"
   end
+
+  it "returns String instances when called on a subclass" do
+    StringSpecs::MyString.new("foo").scrub { |b| "*" }.should.instance_of?(String)
+    input = [0x81].pack('C').force_encoding('utf-8')
+    StringSpecs::MyString.new(input).scrub { |b| "<#{b.unpack("H*")[0]}>" }.should.instance_of?(String)
+  end
 end
 
 describe "String#scrub!" do
@@ -105,4 +141,24 @@ describe "String#scrub!" do
     input.scrub! { |b| "<?>" }
     input.should == "a<?>"
   end
+
+  it "maintains the state of frozen strings that are already valid" do
+    input = "a"
+    input.freeze
+    input.scrub!
+    input.frozen?.should == true
+  end
+
+  it "preserves the instance variables of already valid strings" do
+    input = "a"
+    input.instance_variable_set(:@a, 'b')
+    input.scrub!
+    input.instance_variable_get(:@a).should == 'b'
+  end
+
+  it "accepts a frozen string as a replacement" do
+    input = "a\xE2"
+    input.scrub!('.'.freeze)
+    input.should == 'a.'
+  end
 end