Fix local initialization in generated regex conditional backreference…

…s in loops (#73164) The variable is later read if there's any backtracking and/or the conditional backreference is in a loop, but it was only being initialized if there was backtracking.
dotnet · Aug 1, 2022 · 68bd58e · 68bd58e
1 parent 0e2d041
commit 68bd58e
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 6 deletions.
diff --git a/src/libraries/System.Text.RegularExpressions/gen/RegexGenerator.Emitter.cs b/src/libraries/System.Text.RegularExpressions/gen/RegexGenerator.Emitter.cs
@@ -1799,7 +1799,7 @@ void EmitBackreferenceConditional(RegexNode node)
                 writer.WriteLine();
                 TransferSliceStaticPosToPos(); // make sure sliceStaticPos is 0 after each branch
                 string postYesDoneLabel = doneLabel;
-                if (postYesDoneLabel != originalDoneLabel)
+                if (postYesDoneLabel != originalDoneLabel || isInLoop)
                 {
                     writer.WriteLine($"{resumeAt} = 0;");
                 }
@@ -1821,7 +1821,7 @@ void EmitBackreferenceConditional(RegexNode node)
                     writer.WriteLine();
                     TransferSliceStaticPosToPos(); // make sure sliceStaticPos is 0 after each branch
                     postNoDoneLabel = doneLabel;
-                    if (postNoDoneLabel != originalDoneLabel)
+                    if (postNoDoneLabel != originalDoneLabel || isInLoop)
                     {
                         writer.WriteLine($"{resumeAt} = 1;");
                     }
@@ -1831,7 +1831,7 @@ void EmitBackreferenceConditional(RegexNode node)
                     // There's only a yes branch.  If it's going to cause us to output a backtracking
                     // label but code may not end up taking the yes branch path, we need to emit a resumeAt
                     // that will cause the backtracking to immediately pass through this node.
-                    if (postYesDoneLabel != originalDoneLabel)
+                    if (postYesDoneLabel != originalDoneLabel || isInLoop)
                     {
                         writer.WriteLine($"{resumeAt} = 2;");
                     }

diff --git a/...raries/System.Text.RegularExpressions/src/System/Text/RegularExpressions/RegexCompiler.cs b/...raries/System.Text.RegularExpressions/src/System/Text/RegularExpressions/RegexCompiler.cs
@@ -1884,7 +1884,7 @@ void EmitBackreferenceConditional(RegexNode node)
                 EmitNode(yesBranch);
                 TransferSliceStaticPosToPos();
                 Label postYesDoneLabel = doneLabel;
-                if (!isAtomic && postYesDoneLabel != originalDoneLabel)
+                if ((!isAtomic && postYesDoneLabel != originalDoneLabel) || isInLoop)
                 {
                     // resumeAt = 0;
                     Ldc(0);
@@ -1907,7 +1907,7 @@ void EmitBackreferenceConditional(RegexNode node)
                     EmitNode(noBranch);
                     TransferSliceStaticPosToPos(); // make sure sliceStaticPos is 0 after each branch
                     postNoDoneLabel = doneLabel;
-                    if (!isAtomic && postNoDoneLabel != originalDoneLabel)
+                    if ((!isAtomic && postNoDoneLabel != originalDoneLabel) || isInLoop)
                     {
                         // resumeAt = 1;
                         Ldc(1);
@@ -1919,7 +1919,7 @@ void EmitBackreferenceConditional(RegexNode node)
                     // There's only a yes branch.  If it's going to cause us to output a backtracking
                     // label but code may not end up taking the yes branch path, we need to emit a resumeAt
                     // that will cause the backtracking to immediately pass through this node.
-                    if (!isAtomic && postYesDoneLabel != originalDoneLabel)
+                    if ((!isAtomic && postYesDoneLabel != originalDoneLabel) || isInLoop)
                     {
                         // resumeAt = 2;
                         Ldc(2);

diff --git a/src/libraries/System.Text.RegularExpressions/tests/FunctionalTests/Regex.Match.Tests.cs b/src/libraries/System.Text.RegularExpressions/tests/FunctionalTests/Regex.Match.Tests.cs
@@ -619,6 +619,7 @@ public static IEnumerable<object[]> Match_MemberData()
                 yield return (@"(...)(?(1)\w*|\s*)[a1 ]", "----       ", RegexOptions.None, 0, 11, true, "--- ");
                 yield return (@"(...)(?(1)\w*|\s*)[a1 ]", "zabcaaaaaaa", RegexOptions.RightToLeft, 0, 11, true, "aaaa");
                 yield return (@"(...)(?(1)\w*|\s*)[a1 ]", "----       ", RegexOptions.RightToLeft, 0, 11, true, "---       ");
+                yield return (@"(aaa)(?(1)aaa|b?)*", "aaaaaa", RegexOptions.None, 0, 6, true, "aaaaaa");
             }
 
             // Character Class Subtraction