[GR-17457] Handle ImmutableString in String#scrub

PullRequest: truffleruby/2309

Author: bjfish

ci.jsonnet

@@ -44,7 +44,7 @@ local part_definitions = {
       },
 
       packages+: {
-        'mx': '5.279.2',
+        'mx': '5.280.5',
         "pip:ninja_syntax": "==1.7.2",  # Required by NFI and mx
       },
 

spec/ruby/core/string/scrub_spec.rb

@@ -39,6 +39,13 @@
     "abc\u3042#{x81}".scrub("*").should == "abc\u3042*"
   end
 
+  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')
+    (-(utf16_str)).scrub("*".encode('UTF-16LE')).should == "abc*".encode('UTF-16LE')
+  end
+
   it "replaces an incomplete character at the end with a single replacement" do
     xE3x80 = [0xE3, 0x80].pack('CC').force_encoding 'utf-8'
     xE3x80.scrub("*").should == "*"

src/main/java/org/truffleruby/core/string/StringNodes.java

@@ -1803,9 +1803,13 @@ public abstract static class ScrubNode extends PrimitiveArrayArgumentsNode {
                 .create();
         @Child private RopeNodes.BytesNode bytesNode = RopeNodes.BytesNode.create();
 
-        @Specialization(guards = { "isBrokenCodeRange(string.rope, codeRangeNode)", "isAsciiCompatible(string)" })
-        protected RubyString scrubAsciiCompat(RubyString string, RubyProc block) {
-            final Rope rope = string.rope;
+        @Specialization(
+                guards = {
+                        "isBrokenCodeRange(rope, codeRangeNode)",
+                        "isAsciiCompatible(rope)" })
+        protected RubyString scrubAsciiCompat(Object string, RubyProc block,
+                @CachedLibrary(limit = "2") RubyStringLibrary strings,
+                @Bind("strings.getRope(string)") Rope rope) {
             final Encoding enc = rope.getEncoding();
             final CodeRange cr = codeRangeNode.execute(rope);
             Rope buf = RopeConstants.EMPTY_ASCII_8BIT_ROPE;
@@ -1874,10 +1878,14 @@ protected RubyString scrubAsciiCompat(RubyString string, RubyProc block) {
             return makeStringNode.fromRope(buf);
         }
 
-        @Specialization(guards = { "isBrokenCodeRange(string.rope, codeRangeNode)", "!isAsciiCompatible(string)" })
-        protected RubyString scrubAsciiIncompatible(RubyString string, RubyProc block,
+        @Specialization(
+                guards = {
+                        "isBrokenCodeRange(rope, codeRangeNode)",
+                        "!isAsciiCompatible(rope)" })
+        protected RubyString scrubAsciiIncompatible(Object string, RubyProc block,
+                @CachedLibrary(limit = "2") RubyStringLibrary strings,
+                @Bind("strings.getRope(string)") Rope rope,
                 @Cached RopeNodes.CalculateCharacterLengthNode calculateCharacterLengthNode) {
-            final Rope rope = string.rope;
             final Encoding enc = rope.getEncoding();
             final CodeRange cr = codeRangeNode.execute(rope);
             Rope buf = RopeConstants.EMPTY_ASCII_8BIT_ROPE;