[GR-17457] Fix for unpack Z null terminator consumption.

PullRequest: truffleruby/3351

Author: aardvark179

CHANGELOG.md

@@ -13,6 +13,7 @@ Bug fixes:
 * Fix `String#[Regexp, Integer]` when the capture group exists but is not matched (@eregon).
 * Fix `File.open` mode string parsing when binary option is the third character (@bjfish).
 * Fix `rb_scan_args_kw` macro to avoid shadowing variables (#2649, @aardvark179).
+* Fix `String#unpack("Z")` to not advance after the null byte, like CRuby (#2659, @aardvark179).
 
 Compatibility:
 

spec/ruby/core/string/unpack/z_spec.rb

@@ -20,4 +20,9 @@
       ["\x00a\x00 bc \x00",  ["", "c"]]
     ].should be_computed_by(:unpack, "Z5Z")
   end
+
+  it "does not advance past the null byte when given a 'Z' format specifier" do
+    "a\x00\x0f".unpack('Zxc').should == ['a', 15]
+    "a\x00\x0f".unpack('Zcc').should == ['a', 0, 15]
+  end
 end

src/main/java/org/truffleruby/core/format/read/bytes/ReadBinaryStringNode.java

@@ -81,10 +81,6 @@ protected RubyString read(VirtualFrame frame, byte[] source,
             while (start + length < getSourceLength(frame) && length < count && source[start + length] != 0) {
                 length++;
             }
-
-            if (start + length < getSourceLength(frame) && source[start + length] == 0) {
-                length++;
-            }
         } else {
             length = count;