[GR-17457] Fix StringIO#ungetbyte behaviour and add specs.

PullRequest: truffleruby/2913

CHANGELOG.md

@@ -41,6 +41,7 @@ Bug fixes:
 * Fix handling of incompatible types in `Float#<=>` (#2432, @chrisseaton).
 * Fix issue with escaping curly braces for `Dir.glob` (#2425).
 * Fix `base64` decoding issue with missing output (#2435).
+* Fix `StringIO#ungetbyte` to treat a byte as a byte, not a code point (#2436). 
 
 Compatibility:
 

lib/truffle/stringio.rb

@@ -599,20 +599,20 @@ def ungetbyte(bytes)
     return unless bytes
 
     if bytes.kind_of? Integer
-      bytes = '' << bytes
+      bytes = ''.b << (bytes & 0xff)
     else
-      bytes = StringValue(bytes)
+      bytes = StringValue(bytes).b
       return if bytes.bytesize == 0
     end
 
     d = @__data__
     pos = d.pos
-    string = d.string
+    string = d.string.b
 
-    enc = string.encoding
+    enc = d.string.encoding
 
     if d.pos == 0
-      d.string = "#{bytes}#{string}"
+      d.string = bytes << string
     else
       size = bytes.bytesize
       a = string.byteslice(0, pos - size) if size < pos

spec/ruby/library/stringio/ungetbyte_spec.rb

@@ -1,6 +1,42 @@
+# frozen_string_literal: false
 require_relative '../../spec_helper'
 require 'stringio'
 
 describe "StringIO#ungetbyte" do
-  it "needs to be reviewed for spec completeness"
+  it "ungets a single byte from a string starting with a single byte character" do
+    str = 'This is a simple string.'
+    io = StringIO.new("#{str}")
+    c = io.getc
+    c.should == 'T'
+    io.ungetbyte(83)
+    io.string.should == 'Shis is a simple string.'
+  end
+
+  it "ungets a single byte from a string in the middle of a multibyte characte" do
+    str = "\u01a9"
+    io = StringIO.new(str)
+    b = io.getbyte
+    b.should == 0xc6 # First byte of UTF-8 encoding of \u01a9
+    io.ungetbyte(0xce) # First byte of UTF-8 encoding of \u03a9
+    io.string.should == "\u03a9"
+  end
+
+  it "constrains the value of a numeric argument to a single byte" do
+    str = 'This is a simple string.'
+    io = StringIO.new("#{str}")
+    c = io.getc
+    c.should == 'T'
+    io.ungetbyte(83 | 0xff00)
+    io.string.should == 'Shis is a simple string.'
+  end
+
+  it "ungets the bytes of a string if given a string as an arugment" do
+    str = "\u01a9"
+    io = StringIO.new(str)
+    b = io.getbyte
+    b.should == 0xc6 # First byte of UTF-8 encoding of \u01a9
+    io.ungetbyte("\u01a9")
+    io.string.bytes.should == [198, 169, 169]
+  end
+
 end