5 files changed, 532 insertions, 25 deletions

diff --git a/spec/ruby/core/string/capitalize_spec.rb b/spec/ruby/core/string/capitalize_spec.rb
index e279f7d061..10f9ab00a1 100644
--- a/spec/ruby/core/string/capitalize_spec.rb
+++ b/spec/ruby/core/string/capitalize_spec.rb
@@ -26,8 +26,65 @@ describe "String#capitalize" do
   end
 
   ruby_version_is '2.4' do
-    it "works for all of Unicode" do
-      "äöü".capitalize.should == "Äöü"
+    describe "full Unicode case mapping" do
+      it "works for all of Unicode with no option" do
+        "äöÜ".capitalize.should == "Äöü"
+      end
+
+      it "only capitalizes the first resulting character when upcasing a character produces a multi-character sequence" do
+        "ß".capitalize.should == "Ss"
+      end
+
+      it "updates string metadata" do
+        capitalized = "ßeT".capitalize
+
+        capitalized.should == "Sset"
+        capitalized.size.should == 4
+        capitalized.bytesize.should == 4
+        capitalized.ascii_only?.should be_true
+      end
+    end
+
+    describe "ASCII-only case mapping" do
+      it "does not capitalize non-ASCII characters" do
+        "ßet".capitalize(:ascii).should == "ßet"
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Turkic languages" do
+      it "capitalizes ASCII characters according to Turkic semantics" do
+        "iSa".capitalize(:turkic).should == "İsa"
+      end
+
+      it "allows Lithuanian as an extra option" do
+        "iSa".capitalize(:turkic, :lithuanian).should == "İsa"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { "iSa".capitalize(:turkic, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Lithuanian" do
+      it "currently works the same as full Unicode case mapping" do
+        "iß".capitalize(:lithuanian).should == "Iß"
+      end
+
+      it "allows Turkic as an extra option (and applies Turkic semantics)" do
+        "iß".capitalize(:lithuanian, :turkic).should == "İß"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { "iß".capitalize(:lithuanian, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    it "does not allow the :fold option for upcasing" do
+      lambda { "abc".capitalize(:fold) }.should raise_error(ArgumentError)
+    end
+
+    it "does not allow invalid options" do
+      lambda { "abc".capitalize(:invalid_option) }.should raise_error(ArgumentError)
     end
   end
 
@@ -45,10 +102,80 @@ describe "String#capitalize!" do
   end
 
   ruby_version_is '2.4' do
-    it "capitalizes self in place for all of Unicode" do
-      a = "äöü"
-      a.capitalize!.should equal(a)
-      a.should == "Äöü"
+    describe "full Unicode case mapping" do
+      it "modifies self in place for all of Unicode with no option" do
+        a = "äöÜ"
+        a.capitalize!
+        a.should == "Äöü"
+      end
+
+      it "only capitalizes the first resulting character when upcasing a character produces a multi-character sequence" do
+        a = "ß"
+        a.capitalize!
+        a.should == "Ss"
+      end
+
+      it "updates string metadata" do
+        capitalized = "ßeT"
+        capitalized.capitalize!
+
+        capitalized.should == "Sset"
+        capitalized.size.should == 4
+        capitalized.bytesize.should == 4
+        capitalized.ascii_only?.should be_true
+      end
+    end
+
+    describe "modifies self in place for ASCII-only case mapping" do
+      it "does not capitalize non-ASCII characters" do
+        a = "ßet"
+        a.capitalize!(:ascii)
+        a.should == "ßet"
+      end
+    end
+
+    describe "modifies self in place for full Unicode case mapping adapted for Turkic languages" do
+      it "capitalizes ASCII characters according to Turkic semantics" do
+        a = "iSa"
+        a.capitalize!(:turkic)
+        a.should == "İsa"
+      end
+
+      it "allows Lithuanian as an extra option" do
+        a = "iSa"
+        a.capitalize!(:turkic, :lithuanian)
+        a.should == "İsa"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { a = "iSa"; a.capitalize!(:turkic, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    describe "modifies self in place for full Unicode case mapping adapted for Lithuanian" do
+      it "currently works the same as full Unicode case mapping" do
+        a = "iß"
+        a.capitalize!(:lithuanian)
+        a.should == "Iß"
+      end
+
+      it "allows Turkic as an extra option (and applies Turkic semantics)" do
+        a = "iß"
+        a.capitalize!(:lithuanian, :turkic)
+        a.should == "İß"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { a = "iß"; a.capitalize!(:lithuanian, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    it "does not allow the :fold option for upcasing" do
+      lambda { a = "abc"; a.capitalize!(:fold) }.should raise_error(ArgumentError)
+    end
+
+    it "does not allow invalid options" do
+      lambda { a = "abc"; a.capitalize!(:invalid_option) }.should raise_error(ArgumentError)
     end
   end
 
diff --git a/spec/ruby/core/string/casecmp_spec.rb b/spec/ruby/core/string/casecmp_spec.rb
index 9e89e1314d..87be999964 100644
--- a/spec/ruby/core/string/casecmp_spec.rb
+++ b/spec/ruby/core/string/casecmp_spec.rb
@@ -37,6 +37,10 @@ describe "String#casecmp independent of case" do
     end
   end
 
+  it "returns nil if incompatible encodings" do
+    "あれ".casecmp("れ".encode(Encoding::EUC_JP)).should be_nil
+  end
+
   describe "in UTF-8 mode" do
     describe "for non-ASCII characters" do
       before :each do
@@ -96,6 +100,12 @@ describe "String#casecmp independent of case" do
     it "returns 1 when numerically greater than other" do
       @lower_a_tilde.casecmp(@upper_a_tilde).should == 1
     end
+
+    ruby_version_is "2.4" do
+      it "does not case fold" do
+        "ß".casecmp("ss").should == 1
+      end
+    end
   end
 
   describe "when comparing a subclass instance" do
@@ -138,6 +148,10 @@ ruby_version_is "2.4" do
       "abc".casecmp?(other).should == true
     end
 
+    it "returns nil if incompatible encodings" do
+      "あれ".casecmp?("れ".encode(Encoding::EUC_JP)).should be_nil
+    end
+
     describe 'for UNICODE characters' do
       it 'returns true when downcase(:fold) on unicode' do
         'äöü'.casecmp?('ÄÖÜ').should == true
@@ -181,6 +195,12 @@ ruby_version_is "2.4" do
       end
     end
 
+    ruby_version_is "2.4" do
+      it "case folds" do
+        "ß".casecmp?("ss").should be_true
+      end
+    end
+
     ruby_version_is "2.4" ... "2.5" do
       it "raises a TypeError if other can't be converted to a string" do
         lambda { "abc".casecmp?(mock('abc')) }.should raise_error(TypeError)
diff --git a/spec/ruby/core/string/downcase_spec.rb b/spec/ruby/core/string/downcase_spec.rb
index 30116db5e6..9fb93902b1 100644
--- a/spec/ruby/core/string/downcase_spec.rb
+++ b/spec/ruby/core/string/downcase_spec.rb
@@ -23,8 +23,64 @@ describe "String#downcase" do
   end
 
   ruby_version_is '2.4' do
-    it "works for all of Unicode" do
-      "ÄÖÜ".downcase.should == "äöü"
+    describe "full Unicode case mapping" do
+      it "works for all of Unicode with no option" do
+        "ÄÖÜ".downcase.should == "äöü"
+      end
+
+      it "updates string metadata" do
+        downcased = "\u{212A}ING".downcase
+
+        downcased.should == "king"
+        downcased.size.should == 4
+        downcased.bytesize.should == 4
+        downcased.ascii_only?.should be_true
+      end
+    end
+
+    describe "ASCII-only case mapping" do
+      it "does not downcase non-ASCII characters" do
+        "CÅR".downcase(:ascii).should == "cÅr"
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Turkic languages" do
+      it "downcases characters according to Turkic semantics" do
+        "İ".downcase(:turkic).should == "i"
+      end
+
+      it "allows Lithuanian as an extra option" do
+        "İ".downcase(:turkic, :lithuanian).should == "i"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { "İ".downcase(:turkic, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Lithuanian" do
+      it "currently works the same as full Unicode case mapping" do
+        "İS".downcase(:lithuanian).should == "i\u{307}s"
+      end
+
+      it "allows Turkic as an extra option (and applies Turkic semantics)" do
+        "İS".downcase(:lithuanian, :turkic).should == "is"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { "İS".downcase(:lithuanian, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    describe "case folding" do
+      it "case folds special characters" do
+        "ß".downcase.should == "ß"
+        "ß".downcase(:fold).should == "ss"
+      end
+    end
+
+    it "does not allow invalid options" do
+      lambda { "ABC".downcase(:invalid_option) }.should raise_error(ArgumentError)
     end
   end
 
@@ -47,10 +103,81 @@ describe "String#downcase!" do
   end
 
   ruby_version_is '2.4' do
-    it "modifies self in place for all of Unicode" do
-      a = "ÄÖÜ"
-      a.downcase!.should equal(a)
-      a.should == "äöü"
+    describe "full Unicode case mapping" do
+      it "modifies self in place for all of Unicode with no option" do
+        a = "ÄÖÜ"
+        a.downcase!
+        a.should == "äöü"
+      end
+
+      it "updates string metadata" do
+        downcased = "\u{212A}ING"
+        downcased.downcase!
+
+        downcased.should == "king"
+        downcased.size.should == 4
+        downcased.bytesize.should == 4
+        downcased.ascii_only?.should be_true
+      end
+    end
+
+    describe "ASCII-only case mapping" do
+      it "does not downcase non-ASCII characters" do
+        a = "CÅR"
+        a.downcase!(:ascii)
+        a.should == "cÅr"
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Turkic languages" do
+      it "downcases characters according to Turkic semantics" do
+        a = "İ"
+        a.downcase!(:turkic)
+        a.should == "i"
+      end
+
+      it "allows Lithuanian as an extra option" do
+        a = "İ"
+        a.downcase!(:turkic, :lithuanian)
+        a.should == "i"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { a = "İ"; a.downcase!(:turkic, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Lithuanian" do
+      it "currently works the same as full Unicode case mapping" do
+        a = "İS"
+        a.downcase!(:lithuanian)
+        a.should == "i\u{307}s"
+      end
+
+      it "allows Turkic as an extra option (and applies Turkic semantics)" do
+        a = "İS"
+        a.downcase!(:lithuanian, :turkic)
+        a.should == "is"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { a = "İS"; a.downcase!(:lithuanian, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    describe "case folding" do
+      it "case folds special characters" do
+        a = "ß"
+        a.downcase!
+        a.should == "ß"
+
+        a.downcase!(:fold)
+        a.should == "ss"
+      end
+    end
+
+    it "does not allow invalid options" do
+      lambda { a = "ABC"; a.downcase!(:invalid_option) }.should raise_error(ArgumentError)
     end
   end
 
diff --git a/spec/ruby/core/string/swapcase_spec.rb b/spec/ruby/core/string/swapcase_spec.rb
index 6d00906aad..bb89dee48f 100644
--- a/spec/ruby/core/string/swapcase_spec.rb
+++ b/spec/ruby/core/string/swapcase_spec.rb
@@ -23,8 +23,61 @@ describe "String#swapcase" do
   end
 
   ruby_version_is '2.4' do
-    it "works for all of Unicode" do
-      "äÖü".swapcase.should == "ÄöÜ"
+    describe "full Unicode case mapping" do
+      it "works for all of Unicode with no option" do
+        "äÖü".swapcase.should == "ÄöÜ"
+      end
+
+      it "updates string metadata" do
+        swapcased = "Aßet".swapcase
+
+        swapcased.should == "aSSET"
+        swapcased.size.should == 5
+        swapcased.bytesize.should == 5
+        swapcased.ascii_only?.should be_true
+      end
+    end
+
+    describe "ASCII-only case mapping" do
+      it "does not swapcase non-ASCII characters" do
+        "aßet".swapcase(:ascii).should == "AßET"
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Turkic languages" do
+      it "swaps case of ASCII characters according to Turkic semantics" do
+        "aiS".swapcase(:turkic).should == "Aİs"
+      end
+
+      it "allows Lithuanian as an extra option" do
+        "aiS".swapcase(:turkic, :lithuanian).should == "Aİs"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { "aiS".swapcase(:turkic, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Lithuanian" do
+      it "currently works the same as full Unicode case mapping" do
+        "Iß".swapcase(:lithuanian).should == "iSS"
+      end
+
+      it "allows Turkic as an extra option (and applies Turkic semantics)" do
+        "iS".swapcase(:lithuanian, :turkic).should == "İs"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { "aiS".swapcase(:lithuanian, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    it "does not allow the :fold option for upcasing" do
+      lambda { "abc".swapcase(:fold) }.should raise_error(ArgumentError)
+    end
+
+    it "does not allow invalid options" do
+      lambda { "abc".swapcase(:invalid_option) }.should raise_error(ArgumentError)
     end
   end
 
@@ -42,10 +95,74 @@ describe "String#swapcase!" do
   end
 
   ruby_version_is '2.4' do
-    it "modifies self in place for all of Unicode" do
-      a = "äÖü"
-      a.swapcase!.should equal(a)
-      a.should == "ÄöÜ"
+    describe "full Unicode case mapping" do
+      it "modifies self in place for all of Unicode with no option" do
+        a = "äÖü"
+        a.swapcase!
+        a.should == "ÄöÜ"
+      end
+
+      it "updates string metadata" do
+        swapcased = "Aßet"
+        swapcased.swapcase!
+
+        swapcased.should == "aSSET"
+        swapcased.size.should == 5
+        swapcased.bytesize.should == 5
+        swapcased.ascii_only?.should be_true
+      end
+    end
+
+    describe "modifies self in place for ASCII-only case mapping" do
+      it "does not swapcase non-ASCII characters" do
+        a = "aßet"
+        a.swapcase!(:ascii)
+        a.should == "AßET"
+      end
+    end
+
+    describe "modifies self in place for full Unicode case mapping adapted for Turkic languages" do
+      it "swaps case of ASCII characters according to Turkic semantics" do
+        a = "aiS"
+        a.swapcase!(:turkic)
+        a.should == "Aİs"
+      end
+
+      it "allows Lithuanian as an extra option" do
+        a = "aiS"
+        a.swapcase!(:turkic, :lithuanian)
+        a.should == "Aİs"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { a = "aiS"; a.swapcase!(:turkic, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Lithuanian" do
+      it "currently works the same as full Unicode case mapping" do
+        a = "Iß"
+        a.swapcase!(:lithuanian)
+        a.should == "iSS"
+      end
+
+      it "allows Turkic as an extra option (and applies Turkic semantics)" do
+        a = "iS"
+        a.swapcase!(:lithuanian, :turkic)
+        a.should == "İs"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { a = "aiS"; a.swapcase!(:lithuanian, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    it "does not allow the :fold option for upcasing" do
+      lambda { a = "abc"; a.swapcase!(:fold) }.should raise_error(ArgumentError)
+    end
+
+    it "does not allow invalid options" do
+      lambda { a = "abc"; a.swapcase!(:invalid_option) }.should raise_error(ArgumentError)
     end
   end
 
diff --git a/spec/ruby/core/string/upcase_spec.rb b/spec/ruby/core/string/upcase_spec.rb
index 001cf1f495..347f567be9 100644
--- a/spec/ruby/core/string/upcase_spec.rb
+++ b/spec/ruby/core/string/upcase_spec.rb
@@ -23,8 +23,61 @@ describe "String#upcase" do
   end
 
   ruby_version_is '2.4' do
-    it "works for all of Unicode" do
-      "äöü".upcase.should == "ÄÖÜ"
+    describe "full Unicode case mapping" do
+      it "works for all of Unicode with no option" do
+        "äöü".upcase.should == "ÄÖÜ"
+      end
+
+      it "updates string metadata" do
+        upcased = "aßet".upcase
+
+        upcased.should == "ASSET"
+        upcased.size.should == 5
+        upcased.bytesize.should == 5
+        upcased.ascii_only?.should be_true
+      end
+    end
+
+    describe "ASCII-only case mapping" do
+      it "does not upcase non-ASCII characters" do
+        "aßet".upcase(:ascii).should == "AßET"
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Turkic languages" do
+      it "upcases ASCII characters according to Turkic semantics" do
+        "i".upcase(:turkic).should == "İ"
+      end
+
+      it "allows Lithuanian as an extra option" do
+        "i".upcase(:turkic, :lithuanian).should == "İ"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { "i".upcase(:turkic, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    describe "full Unicode case mapping adapted for Lithuanian" do
+      it "currently works the same as full Unicode case mapping" do
+        "iß".upcase(:lithuanian).should == "ISS"
+      end
+
+      it "allows Turkic as an extra option (and applies Turkic semantics)" do
+        "iß".upcase(:lithuanian, :turkic).should == "İSS"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { "iß".upcase(:lithuanian, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    it "does not allow the :fold option for upcasing" do
+      lambda { "abc".upcase(:fold) }.should raise_error(ArgumentError)
+    end
+
+    it "does not allow invalid options" do
+      lambda { "abc".upcase(:invalid_option) }.should raise_error(ArgumentError)
     end
   end
 
@@ -46,12 +99,75 @@ describe "String#upcase!" do
     a.should == "HELLO"
   end
 
-
   ruby_version_is '2.4' do
-    it "modifies self in place for all of Unicode" do
-      a = "äöü"
-      a.upcase!.should equal(a)
-      a.should == "ÄÖÜ"
+    describe "full Unicode case mapping" do
+      it "modifies self in place for all of Unicode with no option" do
+        a = "äöü"
+        a.upcase!
+        a.should == "ÄÖÜ"
+      end
+
+      it "updates string metadata for self" do
+        upcased = "aßet"
+        upcased.upcase!
+
+        upcased.should == "ASSET"
+        upcased.size.should == 5
+        upcased.bytesize.should == 5
+        upcased.ascii_only?.should be_true
+      end
+    end
+
+    describe "modifies self in place for ASCII-only case mapping" do
+      it "does not upcase non-ASCII characters" do
+        a = "aßet"
+        a.upcase!(:ascii)
+        a.should == "AßET"
+      end
+    end
+
+    describe "modifies self in place for full Unicode case mapping adapted for Turkic languages" do
+      it "upcases ASCII characters according to Turkic semantics" do
+        a = "i"
+        a.upcase!(:turkic)
+        a.should == "İ"
+      end
+
+      it "allows Lithuanian as an extra option" do
+        a = "i"
+        a.upcase!(:turkic, :lithuanian)
+        a.should == "İ"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { a = "i"; a.upcase!(:turkic, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    describe "modifies self in place for full Unicode case mapping adapted for Lithuanian" do
+      it "currently works the same as full Unicode case mapping" do
+        a = "iß"
+        a.upcase!(:lithuanian)
+        a.should == "ISS"
+      end
+
+      it "allows Turkic as an extra option (and applies Turkic semantics)" do
+        a = "iß"
+        a.upcase!(:lithuanian, :turkic)
+        a.should == "İSS"
+      end
+
+      it "does not allow any other additional option" do
+        lambda { a = "iß"; a.upcase!(:lithuanian, :ascii) }.should raise_error(ArgumentError)
+      end
+    end
+
+    it "does not allow the :fold option for upcasing" do
+      lambda { a = "abc"; a.upcase!(:fold) }.should raise_error(ArgumentError)
+    end
+
+    it "does not allow invalid options" do
+      lambda { a = "abc"; a.upcase!(:invalid_option) }.should raise_error(ArgumentError)
     end
   end