TRegex: update CaseFoldData for JavaFlavor SDK version 25

Author: djoooooe

regex/src/com.oracle.truffle.regex/src/com/oracle/truffle/regex/flavor/java/JavaFlavor.java

@@ -42,6 +42,7 @@
 
 import com.oracle.truffle.api.CompilerDirectives;
 import com.oracle.truffle.regex.RegexLanguage;
+import com.oracle.truffle.regex.RegexOptions;
 import com.oracle.truffle.regex.RegexSource;
 import com.oracle.truffle.regex.tregex.buffer.CompilationBuffer;
 import com.oracle.truffle.regex.tregex.parser.CaseFoldData;
@@ -85,12 +86,16 @@ public RegexValidator createValidator(RegexLanguage language, RegexSource source
 
     @Override
     public EqualsIgnoreCasePredicate getEqualsIgnoreCasePredicate(RegexAST ast) {
-        return (a, b, altMode) -> getCaseFoldingAlgorithm(altMode).getEqualsPredicate().test(a, b);
+        return (a, b, altMode) -> getCaseFoldingAlgorithm(ast.getOptions(), altMode).getEqualsPredicate().test(a, b);
     }
 
-    public static CaseFoldData.CaseFoldUnfoldAlgorithm getCaseFoldingAlgorithm(boolean isUnicodeCase) {
+    public static CaseFoldData.CaseFoldUnfoldAlgorithm getCaseFoldingAlgorithm(RegexOptions options, boolean isUnicodeCase) {
         if (isUnicodeCase) {
-            return CaseFoldData.CaseFoldUnfoldAlgorithm.JavaUnicode;
+            if (options.getJavaJDKVersion() <= 24) {
+                return CaseFoldData.CaseFoldUnfoldAlgorithm.JavaUnicode15;
+            } else {
+                return CaseFoldData.CaseFoldUnfoldAlgorithm.JavaUnicode16;
+            }
         } else {
             return CaseFoldData.CaseFoldUnfoldAlgorithm.Ascii;
         }

regex/src/com.oracle.truffle.regex/src/com/oracle/truffle/regex/flavor/java/JavaRegexLexer.java

@@ -288,9 +288,7 @@ protected boolean featureEnabledClassSetExpressions() {
 
     @Override
     protected void caseFoldUnfold(CodePointSetAccumulator charClass) {
-        CaseFoldData.CaseFoldUnfoldAlgorithm caseFolding = (getLocalFlags().isUnicodeCase() || getLocalFlags().isUnicodeCharacterClass())
-                        ? CaseFoldData.CaseFoldUnfoldAlgorithm.JavaUnicode
-                        : CaseFoldData.CaseFoldUnfoldAlgorithm.Ascii;
+        CaseFoldData.CaseFoldUnfoldAlgorithm caseFolding = JavaFlavor.getCaseFoldingAlgorithm(source.getOptions(), getLocalFlags().isUnicodeCase() || getLocalFlags().isUnicodeCharacterClass());
         CaseFoldData.applyCaseFoldUnfold(charClass, compilationBuffer.getCodePointSetAccumulator1(), caseFolding);
     }
 

regex/src/com.oracle.truffle.regex/src/com/oracle/truffle/regex/flavor/java/JavaRegexParser.java

@@ -207,7 +207,8 @@ public RegexAST parse() {
                     break;
                 case classSet:
                     astBuilder.addClassSet((Token.ClassSet) token,
-                                    getFlags().isCaseInsensitive() ? JavaFlavor.getCaseFoldingAlgorithm(getFlags().isUnicodeCase() || getFlags().isUnicodeCharacterClass()) : null);
+                                    getFlags().isCaseInsensitive() ? JavaFlavor.getCaseFoldingAlgorithm(lexer.source.getOptions(), getFlags().isUnicodeCase() || getFlags().isUnicodeCharacterClass())
+                                                    : null);
                     break;
                 case literalString:
                     literalString((Token.LiteralString) token);

regex/src/com.oracle.truffle.regex/src/com/oracle/truffle/regex/tregex/parser/CaseFoldData.java

@@ -65,7 +65,8 @@ public enum CaseFoldUnfoldAlgorithm {
         Ascii,
         ECMAScriptNonUnicode,
         ECMAScriptUnicode,
-        JavaUnicode,
+        JavaUnicode15,
+        JavaUnicode16,
         OracleDBSimple,
         PythonUnicode;
 
@@ -93,12 +94,13 @@ private static CaseFoldEquivalenceTable getTable(CaseFoldUnfoldAlgorithm algorit
                 return UNICODE_16_0_0_EQ_SIMPLE;
             case Ascii:
                 return ASCII;
-            case JavaUnicode:
-                // Currently supported JDK versions for the Java flavor are 21, 22 and 23, where 21
-                // uses Unicode version 15.0.0 and the other versions use Unicode 15.1.0. There are
-                // no differences in the case folding table between those two Unicode versions, so
-                // we can use the same table on all supported JDK versions for now.
+            case JavaUnicode15:
+                // JDK version 21 uses Unicode version 15.0.0, and JDK versions 22-24 use Unicode
+                // 15.1.0. There are no differences in the case folding table between those two
+                // Unicode versions, so we can use the same table for both.
                 return UNICODE_15_0_0_JAVA;
+            case JavaUnicode16:
+                return UNICODE_16_0_0_JAVA;
             case PythonUnicode:
                 return UNICODE_15_1_0_PY;
             default:
@@ -1162,6 +1164,13 @@ public void appendRangesTo(RangesBuffer buffer, int startIndex, int endIndex) {
     private static final CaseFoldEquivalenceTable UNICODE_15_0_0_JAVA = new CaseFoldEquivalenceTable(UNICODE_15_1_0_PY, new CodePointSet[]{
     }, new int[]{
     });
+    private static final CaseFoldEquivalenceTable UNICODE_16_0_0_JAVA = new CaseFoldEquivalenceTable(UNICODE_16_0_0_EQ_SIMPLE, new CodePointSet[]{
+                    rangeSet(0x000049, 0x000049, 0x000069, 0x000069, 0x000130, 0x000131),
+    }, new int[]{
+                    0x000049, 0x000049, DIRECT_MAPPING, 0,
+                    0x000069, 0x000069, DIRECT_MAPPING, 0,
+                    0x000130, 0x000131, DIRECT_MAPPING, 0,
+    });
     private static final CaseFoldTable UNICODE_15_1_0_CF_FULL = new CaseFoldTable(null, new int[]{
                     0x000041, 0x00005a, INTEGER_OFFSET, 32,
                     0x0000b5, 0x0000b5, INTEGER_OFFSET, 775,