Adapt spec tests to not use UTF-32BE or UTF-16BE strings with trailing bytes.

PullRequest: truffleruby/4417

Author: djoooooe

doc/user/compatibility.md

@@ -123,7 +123,7 @@ A possible workaround could be to use natively-allocated strings, but it would b
 
 ### Strings in UTF-16 and UTF-32 encoding
 
-TruffleRuby does not support UTF-16 strings with an odd number of bytes (in native endianness). Similarly, with UTF-32 it needs to be a multiple of 4. This is necessary for optimizations, compression, invariants, etc.
+TruffleRuby does not support UTF-16 strings with an odd number of bytes. Similarly, with UTF-32 it needs to be a multiple of 4. This is necessary for optimizations, compression, invariants, etc.
 
 ### Threads detect interrupts at different points
 

spec/ruby/core/dir/shared/glob.rb

@@ -12,7 +12,7 @@
   end
 
   it "raises an Encoding::CompatibilityError if the argument encoding is not compatible with US-ASCII" do
-    pattern = "file*".dup.force_encoding Encoding::UTF_16BE
+    pattern = "files*".dup.force_encoding Encoding::UTF_16BE
     -> { Dir.send(@method, pattern) }.should raise_error(Encoding::CompatibilityError)
   end
 

spec/ruby/core/encoding/compatible_spec.rb

@@ -159,7 +159,7 @@
   platform_is_not :windows do
     it "raises an Encoding::CompatibilityError if the external encoding is not compatible" do
       Encoding.default_external = Encoding::UTF_16BE
-      -> { File.expand_path("./a") }.should raise_error(Encoding::CompatibilityError)
+      -> { File.expand_path("./ab") }.should raise_error(Encoding::CompatibilityError)
     end
   end
 

spec/ruby/core/file/expand_path_spec.rb

@@ -62,11 +62,11 @@
     @io.rewind
     @io.set_encoding(Encoding::ASCII_8BIT)
 
-    File.binwrite(@name, "\xFE\xFFabc")
+    File.binwrite(@name, "\xFE\xFFabcd")
 
     @io.set_encoding_by_bom.should == Encoding::UTF_16BE
     @io.external_encoding.should == Encoding::UTF_16BE
-    @io.read.b.should == "abc".b
+    @io.read.b.should == "abcd".b
   end
 
   it "returns the result encoding if found BOM UTF_32LE sequence" do
@@ -94,11 +94,11 @@
     @io.rewind
     @io.set_encoding(Encoding::ASCII_8BIT)
 
-    File.binwrite(@name, "\x00\x00\xFE\xFFabc")
+    File.binwrite(@name, "\x00\x00\xFE\xFFabcd")
 
     @io.set_encoding_by_bom.should == Encoding::UTF_32BE
     @io.external_encoding.should == Encoding::UTF_32BE
-    @io.read.b.should == "abc".b
+    @io.read.b.should == "abcd".b
   end
 
   it "returns nil if io is empty" do

spec/ruby/core/io/set_encoding_by_bom_spec.rb

@@ -94,7 +94,7 @@
   end
 
   it "does not transcode the given string when the external encoding is set and the string encoding is BINARY" do
-    str = "été".b
+    str = "été_".b
     io = StringIO.new.set_encoding(Encoding::UTF_16BE)
     io.external_encoding.should == Encoding::UTF_16BE
 

spec/ruby/library/stringio/shared/write.rb

@@ -55,7 +55,7 @@
   it "checks that we are starting to match at the head of a character" do
     "\xC3\xA9".send(@method).should_not.end_with?("\xA9")
     "\xe3\x81\x82".send(@method).should_not.end_with?("\x82")
-    "ab".dup.force_encoding("UTF-16BE").send(@method).should_not.end_with?(
-        "b".dup.force_encoding("UTF-16BE"))
+    "\xd8\x00\xdc\x00".dup.force_encoding("UTF-16BE").send(@method).should_not.end_with?(
+        "\xdc\x00".dup.force_encoding("UTF-16BE"))
   end
 end

spec/ruby/shared/string/end_with.rb

@@ -19,3 +19,4 @@ slow:File.expand_path raises an Encoding::CompatibilityError if the external enc
 slow:File.expand_path does not modify the string argument
 slow:File.expand_path does not modify a HOME string argument
 slow:File.expand_path returns a String when passed a String subclass
+fails:File.expand_path raises an Encoding::CompatibilityError if the external encoding is not compatible

spec/tags/core/file/expand_path_tags.txt

@@ -6,3 +6,4 @@ fails:C-API Encoding function rb_enc_get_index returns -1 for an object without
 fails:C-API Encoding function rb_enc_set_index raises an ArgumentError for a non-encoding capable object
 fails:C-API Encoding function ENCODING_SET raises an ArgumentError for a non-encoding capable object
 slow:C-API Encoding function rb_define_dummy_encoding raises EncodingError if too many encodings
+fails:C-API Encoding function rb_enc_strlen handles broken strings combines valid character and invalid character counts in UTF-16