Adapt specs so Strings are always created with a #bytesize multiple of the encoding minimum length

* Only UTF-16 and UTF-32 have a encoding minimum length > 1,
  so UTF-16 strings must have #bytesize multiple of 2,
  and UTF-32 strings must have #bytesize multiple of 4.

Author: eregon

core/encoding/compatible_spec.rb

@@ -54,7 +54,7 @@
 
     it "returns nil if the second's Encoding is not ASCII compatible" do
       a = "abc".force_encoding("UTF-8")
-      b = "123".force_encoding("UTF-16LE")
+      b = "1234".force_encoding("UTF-16LE")
       Encoding.compatible?(a, b).should be_nil
     end
   end

core/encoding/converter/putback_spec.rb

@@ -34,14 +34,23 @@
     @ec.putback.should == ""
   end
 
+  it "returns the problematic bytes for UTF-16LE" do
+    ec = Encoding::Converter.new("utf-16le", "iso-8859-1")
+    src = "\x00\xd8\x61\x00"
+    dst = ""
+    ec.primitive_convert(src, dst).should == :invalid_byte_sequence
+    ec.primitive_errinfo.should == [:invalid_byte_sequence, "UTF-16LE", "UTF-8", "\x00\xD8", "a\x00"]
+    ec.putback.should == "a\x00".force_encoding("utf-16le")
+    ec.putback.should == ""
+  end
+
   it "accepts an integer argument corresponding to the number of bytes to be put back" do
     ec = Encoding::Converter.new("utf-16le", "iso-8859-1")
     src = "\x00\xd8\x61\x00"
     dst = ""
     ec.primitive_convert(src, dst).should == :invalid_byte_sequence
     ec.primitive_errinfo.should == [:invalid_byte_sequence, "UTF-16LE", "UTF-8", "\x00\xD8", "a\x00"]
-    ec.putback(1).should == "\x00".force_encoding("utf-16le")
-    ec.putback.should == "a".force_encoding("utf-16le")
+    ec.putback(2).should == "a\x00".force_encoding("utf-16le")
     ec.putback.should == ""
   end
 end

core/string/force_encoding_spec.rb

@@ -60,7 +60,7 @@
   end
 
   it "does not transcode self" do
-    str = "\u{8612}"
+    str = "é"
     str.dup.force_encoding('utf-16le').should_not == str.encode('utf-16le')
   end
 

core/string/scrub_spec.rb

@@ -56,7 +56,9 @@
   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
 

core/string/shared/eql.rb

@@ -21,7 +21,7 @@
   end
 
   it "considers encoding compatibility" do
-    "hello".force_encoding("utf-8").send(@method, "hello".force_encoding("utf-32le")).should be_false
+    "abcd".force_encoding("utf-8").send(@method, "abcd".force_encoding("utf-32le")).should be_false
   end
 
   it "ignores subclass differences" do

core/string/valid_encoding_spec.rb

@@ -43,10 +43,10 @@
     str.force_encoding('KOI8-R').valid_encoding?.should be_true
     str.force_encoding('KOI8-U').valid_encoding?.should be_true
     str.force_encoding('Shift_JIS').valid_encoding?.should be_false
-    str.force_encoding('UTF-16BE').valid_encoding?.should be_false
-    str.force_encoding('UTF-16LE').valid_encoding?.should be_false
-    str.force_encoding('UTF-32BE').valid_encoding?.should be_false
-    str.force_encoding('UTF-32LE').valid_encoding?.should be_false
+    "\xD8\x00".force_encoding('UTF-16BE').valid_encoding?.should be_false
+    "\x00\xD8".force_encoding('UTF-16LE').valid_encoding?.should be_false
+    "\x04\x03\x02\x01".force_encoding('UTF-32BE').valid_encoding?.should be_false
+    "\x01\x02\x03\x04".force_encoding('UTF-32LE').valid_encoding?.should be_false
     str.force_encoding('Windows-1251').valid_encoding?.should be_true
     str.force_encoding('IBM437').valid_encoding?.should be_true
     str.force_encoding('IBM737').valid_encoding?.should be_true