Fix the calculation of enclosedCaptureGroups when wrapping groups

zakkak · Feb 16, 2024 · c4bdc53 · c4bdc53
1 parent bf6c001
commit c4bdc53
Show file tree

Hide file tree

Showing 4 changed files with 22 additions and 6 deletions.
diff --git a/...egex/src/com/oracle/truffle/regex/tregex/nodes/nfa/TRegexBacktrackingNFAExecutorNode.java b/...egex/src/com/oracle/truffle/regex/tregex/nodes/nfa/TRegexBacktrackingNFAExecutorNode.java
@@ -131,8 +131,8 @@ public TRegexBacktrackingNFAExecutorNode(RegexAST ast, PureNFA nfa, int numberOf
             QuantifiableTerm quantifiable = zeroWidthQuantifiables.get(i);
             if (quantifiable.isGroup()) {
                 Group group = quantifiable.asGroup();
-                this.zeroWidthTermEnclosedCGLow[i] = group.getEnclosedCaptureGroupsLow();
-                offset += 2 * (group.getEnclosedCaptureGroupsHigh() - group.getEnclosedCaptureGroupsLow());
+                this.zeroWidthTermEnclosedCGLow[i] = group.getCaptureGroupsLow();
+                offset += 2 * (group.getCaptureGroupsHigh() - group.getCaptureGroupsLow());
             }
             this.zeroWidthQuantifierCGOffsets[i + 1] = offset;
         }

diff --git a/.../com.oracle.truffle.regex/src/com/oracle/truffle/regex/tregex/parser/RegexASTBuilder.java b/.../com.oracle.truffle.regex/src/com/oracle/truffle/regex/tregex/parser/RegexASTBuilder.java
@@ -783,8 +783,8 @@ private void setQuantifier(QuantifiableTerm term, Token.Quantifier quantifier) {
     private Group wrapTermInGroup(Term term) {
         Group wrapperGroup = ast.createGroup();
         if (term.isGroup()) {
-            wrapperGroup.setEnclosedCaptureGroupsLow(term.asGroup().getEnclosedCaptureGroupsLow());
-            wrapperGroup.setEnclosedCaptureGroupsHigh(term.asGroup().getEnclosedCaptureGroupsHigh());
+            wrapperGroup.setEnclosedCaptureGroupsLow(term.asGroup().getCaptureGroupsLow());
+            wrapperGroup.setEnclosedCaptureGroupsHigh(term.asGroup().getCaptureGroupsHigh());
         } else if (term.isAtomicGroup()) {
             wrapperGroup.setEnclosedCaptureGroupsLow(term.asAtomicGroup().getEnclosedCaptureGroupsLow());
             wrapperGroup.setEnclosedCaptureGroupsHigh(term.asAtomicGroup().getEnclosedCaptureGroupsHigh());

diff --git a/...racle.truffle.regex/src/com/oracle/truffle/regex/tregex/parser/RegexASTPostProcessor.java b/...racle.truffle.regex/src/com/oracle/truffle/regex/tregex/parser/RegexASTPostProcessor.java
@@ -258,8 +258,8 @@ private void createOptional(QuantifiableTerm term, Token.Quantifier quantifier,
                 curGroup.setExpandedQuantifier(unroll);
                 curGroup.setQuantifier(quantifier);
                 if (term.isGroup()) {
-                    curGroup.setEnclosedCaptureGroupsLow(term.asGroup().getEnclosedCaptureGroupsLow());
-                    curGroup.setEnclosedCaptureGroupsHigh(term.asGroup().getEnclosedCaptureGroupsHigh());
+                    curGroup.setEnclosedCaptureGroupsLow(term.asGroup().getCaptureGroupsLow());
+                    curGroup.setEnclosedCaptureGroupsHigh(term.asGroup().getCaptureGroupsHigh());
                 }
                 if (quantifier.isGreedy()) {
                     createOptionalBranch(term, quantifier, copy, unroll, recurse, emptyGuard);

diff --git a/regex/src/com.oracle.truffle.regex/src/com/oracle/truffle/regex/tregex/parser/ast/Group.java b/regex/src/com.oracle.truffle.regex/src/com/oracle/truffle/regex/tregex/parser/ast/Group.java
@@ -227,6 +227,15 @@ public int getEnclosedCaptureGroupsLow() {
         return enclosedCaptureGroupsLow;
     }
 
+    /**
+     * Gets the (inclusive) lower bound of the range of capture groups in this term. In contrast to
+     * {@link #getEnclosedCaptureGroupsLow()}, this range contains the group itself if it is a
+     * capturing group.
+     */
+    public int getCaptureGroupsLow() {
+        return isCapturing() ? getGroupNumber() : enclosedCaptureGroupsLow;
+    }
+
     /**
      * Sets the (inclusive) lower bound of the range of capture groups contained within this group.
      */
@@ -242,6 +251,13 @@ public int getEnclosedCaptureGroupsHigh() {
         return enclosedCaptureGroupsHigh;
     }
 
+    /**
+     * Gets the (exclusive) upper bound of the range of capture groups in this term.
+     */
+    public int getCaptureGroupsHigh() {
+        return enclosedCaptureGroupsHigh;
+    }
+
     /**
      * Sets the (exclusive) upper bound of the range of capture groups contained within this group.
      */