[GR-51523] [GR-51674] TRegex bug fixes.

PullRequest: graal/16783

Author: jirkamarsik

regex/src/com.oracle.truffle.regex.test/src/com/oracle/truffle/regex/tregex/test/JsTests.java

@@ -268,4 +268,11 @@ public void gr48586() {
     public void gr50807() {
         test("(?<=%b{1,4}?)foo", "", "%bbbbfoo", 0, true, 5, 8);
     }
+
+    @Test
+    public void gr51523() {
+        test("(?:^|\\.?)([A-Z])", "g", "desktopBrowser", 0, true, 7, 8, 7, 8);
+        test("(?:^|\\.?)([A-Z])", "g", "locationChanged", 0, true, 8, 9, 8, 9);
+        test("(?:^|\\.?)([A-Z]|(?<=[a-z])\\d(?=\\d+))", "g", "helloWorld", 0, true, 5, 6, 5, 6);
+    }
 }

regex/src/com.oracle.truffle.regex.test/src/com/oracle/truffle/regex/tregex/test/OracleDBTests.java

@@ -229,6 +229,12 @@ public void testSpecialGroups() {
         }
     }
 
+    @Test
+    public void testNestedQuantifier() {
+        test("(a*|b)*", "", "aaaaaabaaaaaaaaaaaaaabb", 0, true, 0, 6, 6, 6);
+        test("a((b?)*)*", "", "ab", 0, true, 0, 2, 2, 2, 2, 2);
+    }
+
     @Test
     public void generatedTests() {
         /* GENERATED CODE BEGIN - KEEP THIS MARKER FOR AUTOMATIC UPDATES */

regex/src/com.oracle.truffle.regex/src/com/oracle/truffle/regex/tregex/nfa/ASTStepVisitor.java

@@ -144,6 +144,7 @@ public ASTStep step(NFAState expandState) {
 
     @Override
     protected void visit(RegexASTNode target) {
+        assert noPredicatesInGuards(getQuantifierGuardsOnPath());
         ASTSuccessor successor = new ASTSuccessor();
         ASTTransition transition = new ASTTransition(ast.getLanguage());
         transition.setGroupBoundaries(getGroupBoundaries());
@@ -183,6 +184,19 @@ protected void visit(RegexASTNode target) {
         stepCur.addSuccessor(successor);
     }
 
+    private static boolean noPredicatesInGuards(QuantifierGuard[] quantifierGuards) {
+        // Normalization should remove any exitZeroWidth, escapeZeroWidth, checkGroupMatched and
+        // checkGroupNotMatched guards. The effect of updateCG guards is implemented using
+        // getGroupBoundaries and enterZeroWidth guards have no effect when exitZeroWidth and
+        // escapeZeroWidth are removed already. Other guards shouldn't be used when building a DFA.
+        for (QuantifierGuard guard : quantifierGuards) {
+            if (guard.getKind() != QuantifierGuard.Kind.updateCG && guard.getKind() != QuantifierGuard.Kind.enterZeroWidth) {
+                return false;
+            }
+        }
+        return true;
+    }
+
     @Override
     protected void enterLookAhead(LookAheadAssertion assertion) {
         TBitSet currentMatchedConditionGroups = getCurrentMatchedConditionGroups();

regex/src/com.oracle.truffle.regex/src/com/oracle/truffle/regex/tregex/nfa/QuantifierGuard.java

@@ -44,6 +44,7 @@
 import com.oracle.truffle.api.CompilerDirectives.TruffleBoundary;
 import com.oracle.truffle.regex.tregex.parser.Token.Quantifier;
 import com.oracle.truffle.regex.tregex.parser.ast.ConditionalBackReferenceGroup;
+import com.oracle.truffle.regex.tregex.parser.flavors.RegexFlavor;
 
 import java.util.Objects;
 
@@ -85,15 +86,17 @@ public enum Kind {
         enterZeroWidth,
         /**
          * Transition is leaving a quantified expression that may match the empty string. Check if
-         * the current index is greater than the saved index. In the case of Ruby, also check if any
-         * capture groups were modified.
+         * the current index is greater than the saved index. In the case of flavors in which
+         * {@link RegexFlavor#emptyChecksMonitorCaptureGroups()}, also check if any capture groups
+         * were modified.
          */
         exitZeroWidth,
         /**
          * Transition is leaving a quantified expression that may match the empty string and it is
-         * about to continue to what follows the loop. This is only possible in Ruby and only when
-         * the last iteration of the quantiifed expression fails the empty check (the check for the
-         * index and the state of capture groups tested by {@link #exitZeroWidth}).
+         * about to continue to what follows the loop. This is possible in flavors in which
+         * {@link RegexFlavor#failingEmptyChecksDontBacktrack()} and only when the last iteration of
+         * the quantified expression fails the empty check (the check for the index and the state of
+         * capture groups tested by {@link #exitZeroWidth}).
          */
         escapeZeroWidth,
         /**
@@ -111,8 +114,9 @@ public enum Kind {
         exitEmptyMatch,
         /**
          * Transition is passing a capture group boundary. We need this information in order to
-         * implement the empty check test in {@link #exitZeroWidth}, which, in the case of Ruby,
-         * also needs to monitor the state of capture groups in between {@link #enterZeroWidth} and
+         * implement the empty check test in {@link #exitZeroWidth}, which, in the case of flavors
+         * in which {@link RegexFlavor#emptyChecksMonitorCaptureGroups()}, where we need to monitor
+         * the state of capture groups in between {@link #enterZeroWidth} and
          * {@link #exitZeroWidth}.
          */
         updateCG,