1 files changed, 41 insertions, 14 deletions
diff --git a/spec/ruby/core/string/shared/length.rb b/spec/ruby/core/string/shared/length.rb
index 0e6e66ee1c..ae572ba755 100644
--- a/spec/ruby/core/string/shared/length.rb
+++ b/spec/ruby/core/string/shared/length.rb
@@ -10,19 +10,46 @@ describe :string_length, shared: true do
     "four".send(@method).should == 4
   end
 
-  with_feature :encoding do
-    it "returns the length of a string in different encodings" do
-      utf8_str = 'こにちわ' * 100
-      utf8_str.size.should == 400
-      utf8_str.encode(Encoding::UTF_32BE).size.should == 400
-      utf8_str.encode(Encoding::SHIFT_JIS).size.should == 400
-    end
-
-    it "returns the length of the new self after encoding is changed" do
-      str = 'こにちわ'
-      str.send(@method)
-
-      str.force_encoding('ASCII-8BIT').send(@method).should == 12
-    end
+  it "returns the length of a string in different encodings" do
+    utf8_str = 'こにちわ' * 100
+    utf8_str.send(@method).should == 400
+    utf8_str.encode(Encoding::UTF_32BE).send(@method).should == 400
+    utf8_str.encode(Encoding::SHIFT_JIS).send(@method).should == 400
+  end
+
+  it "returns the length of the new self after encoding is changed" do
+    str = +'こにちわ'
+    str.send(@method)
+
+    str.force_encoding('BINARY').send(@method).should == 12
+  end
+
+  it "returns the correct length after force_encoding(BINARY)" do
+    utf8 = "あ"
+    ascii = "a"
+    concat = utf8 + ascii
+
+    concat.encoding.should == Encoding::UTF_8
+    concat.bytesize.should == 4
+
+    concat.send(@method).should == 2
+    concat.force_encoding(Encoding::ASCII_8BIT)
+    concat.send(@method).should == 4
+  end
+
+  it "adds 1 for every invalid byte in UTF-8" do
+    "\xF4\x90\x80\x80".send(@method).should == 4
+    "a\xF4\x90\x80\x80b".send(@method).should == 6
+    "é\xF4\x90\x80\x80è".send(@method).should == 6
+  end
+
+  it "adds 1 (and not 2) for a incomplete surrogate in UTF-16" do
+    "\x00\xd8".dup.force_encoding("UTF-16LE").send(@method).should == 1
+    "\xd8\x00".dup.force_encoding("UTF-16BE").send(@method).should == 1
+  end
+
+  it "adds 1 for a broken sequence in UTF-32" do
+    "\x04\x03\x02\x01".dup.force_encoding("UTF-32LE").send(@method).should == 1
+    "\x01\x02\x03\x04".dup.force_encoding("UTF-32BE").send(@method).should == 1
   end
 end