TRegex: minor cleanups and more assertions

oracle · May 31, 2024 · a19a609 · a19a609
1 parent f5b8bd7
commit a19a609
Show file tree

Hide file tree

Showing 3 changed files with 17 additions and 12 deletions.
diff --git a/...e.truffle.regex/src/com/oracle/truffle/regex/tregex/parser/MultiCharacterCaseFolding.java b/...e.truffle.regex/src/com/oracle/truffle/regex/tregex/parser/MultiCharacterCaseFolding.java
@@ -259,7 +259,7 @@ public static void caseClosure(CaseFoldData.CaseFoldAlgorithm algorithm, CodePoi
                     CodePointSet allowedCodePoints, boolean transitiveEquivalence) {
         tmp.clear();
         caseFoldCharClass(algorithm, charClass, (from, to) -> {
-            if ((transitiveEquivalence || hasNoCaseFolding(algorithm, to[0]))) {
+            if (transitiveEquivalence || hasNoCaseFolding(algorithm, to[0])) {
                 if (to.length == 1) {
                     // Add the case-folded version to the character class...
                     if (filter.test(from, to[0])) {

diff --git a/...e.regex/src/com/oracle/truffle/regex/tregex/parser/flavors/OracleDBCharClassTrieNode.java b/...e.regex/src/com/oracle/truffle/regex/tregex/parser/flavors/OracleDBCharClassTrieNode.java
@@ -178,11 +178,12 @@ public void setEndOfString() {
     }
 
     public void clear() {
+        assert codepoints == null && !isEndOfString : "clear can only be called on the root node";
         children.clear();
     }
 
     public void generateAST(RegexASTBuilder astBuilder, boolean negate) {
-        if (needsGroupWrapper(negate)) {
+        if (rootNeedsGroupWrapper(negate)) {
             astBuilder.pushGroup();
         }
         if (negate) {
@@ -217,7 +218,7 @@ public void generateAST(RegexASTBuilder astBuilder, boolean negate) {
                 }
             }
         }
-        if (needsGroupWrapper(negate)) {
+        if (rootNeedsGroupWrapper(negate)) {
             astBuilder.popGroup();
         }
     }
@@ -265,4 +266,8 @@ private void generateASTInner(RegexASTBuilder astBuilder, boolean negate) {
     private boolean needsGroupWrapper(boolean negate) {
         return children.size() > 1 || !negate && isEndOfString() && !children.isEmpty();
     }
+
+    private boolean rootNeedsGroupWrapper(boolean negate) {
+        return needsGroupWrapper(negate) || negate && children.size() == 1 && !children.get(0).isEndOfString;
+    }
 }
diff --git a/...truffle.regex/src/com/oracle/truffle/regex/tregex/parser/flavors/OracleDBRegexParser.java b/...truffle.regex/src/com/oracle/truffle/regex/tregex/parser/flavors/OracleDBRegexParser.java
@@ -253,16 +253,16 @@ private void addCCAtomIgnoreCase(ClassSetContents contents) {
     private void addCCAtomMultiCharExpansion(ClassSetContents contents, CaseFoldAlgorithm algorithm) {
         caseClosure(algorithm, contents.getCodePointSet());
         addCCAtomCodePointSet(charClassTmpCaseClosure.toCodePointSet());
+        assert contents.isCharacter() || contents.isPosixCollationElement() || contents.isPosixCollationEquivalenceClass();
+        assert contents.getCodePointSet().matchesSingleChar();
         // No transitive closure
-        for (Range range : contents.getCodePointSet()) {
-            CaseFoldData.getTable(algorithm).caseFold(range, (codepoint, caseFolded) -> {
-                if (caseFolded.length > 1) {
-                    CodePointSet encodingRange = Encodings.UTF_8.getFullSet();
-                    CompilationBuffer compilationBuffer = lexer.getCompilationBuffer();
-                    MultiCharacterCaseFolding.caseFoldUnfoldString(algorithm, caseFolded, encodingRange, false, false, null, curCharClass, compilationBuffer);
-                }
-            });
-        }
+        CaseFoldData.getTable(algorithm).caseFold(contents.getCodePointSet().iterator().next(), (codepoint, caseFolded) -> {
+            if (caseFolded.length > 1) {
+                CodePointSet encodingRange = Encodings.UTF_8.getFullSet();
+                CompilationBuffer compilationBuffer = lexer.getCompilationBuffer();
+                MultiCharacterCaseFolding.caseFoldUnfoldString(algorithm, caseFolded, encodingRange, false, false, null, curCharClass, compilationBuffer);
+            }
+        });
     }
 
     private void caseClosure(CaseFoldAlgorithm algorithm, CodePointSet codePointSet) {