Coalesce adjacent equivalent anchors (#118083)

Sometimes you see patterns where folks have put the same anchor multiple times in a row, e.g. `\b\b`. The subsequent anchors are nops and can just be removed.

Author: stephentoub

src/libraries/System.Text.RegularExpressions/src/System/Text/RegularExpressions/RegexNode.cs

@@ -1832,6 +1832,16 @@ static bool CanCombineCounts(int nodeMin, int nodeMax, int nextMin, int nextMax)
                             currentNode.MakeRep(RegexNodeKind.Oneloop, 2, 2);
                             next++;
                             continue;
+
+                        // Coalescing identical anchors (e.g. \b\b). These don't need to become loops, as they collapse to a single anchor.
+                        case RegexNodeKind.Beginning or RegexNodeKind.Start or
+                             RegexNodeKind.End or RegexNodeKind.EndZ or
+                             RegexNodeKind.Bol or RegexNodeKind.Eol or
+                             RegexNodeKind.Boundary or RegexNodeKind.NonBoundary or
+                             RegexNodeKind.ECMABoundary or RegexNodeKind.NonECMABoundary
+                             when nextNode.Kind == currentNode.Kind:
+                            next++;
+                            continue;
                     }
                 }
 

src/libraries/System.Text.RegularExpressions/tests/UnitTests/RegexReductionTests.cs

@@ -376,6 +376,16 @@ public class RegexReductionTests
         [InlineData("(abc?)*?d", "(ab(?>c?))*?d")]
         [InlineData("(ab*c)*d", "(?>(a(?>b*)c)*)d")]
         [InlineData("(aba)?d", "(?>(aba)?)d")]
+        // Anchors
+        [InlineData(@"\b\b", @"\b")]
+        [InlineData(@"\b\b\b\b\b", @"\b")]
+        [InlineData(@"\B\B", @"\B")]
+        [InlineData(@"^^", @"^")]
+        [InlineData(@"$", @"$")]
+        [InlineData(@"\Z\Z", @"\Z")]
+        [InlineData(@"\z\z", @"\z")]
+        [InlineData(@"\G\G", @"\G")]
+        [InlineData(@"\A\A", @"\A")]
         // Nothing handling
         [InlineData(@"\wabc(?!)def", "(?!)")]
         [InlineData(@"\wabc(?!)def|ghi(?!)", "(?!)")]
@@ -532,6 +542,10 @@ public void PatternsReduceIdentically(string actual, string expected)
         [InlineData("a*(?(xyz)bcd)", "(?>a*)(?(xyz)bcd)")]
         // Different prefixes on alternation branches
         [InlineData("^abcd|$abce", "^abcd|^abce")]
+        // Anchors
+        [InlineData(@"\b\B", "\b")]
+        [InlineData(@"^$", "^")]
+        [InlineData(@"^$", "$")]
         public void PatternsReduceDifferently(string actual, string expected)
         {
             // NOTE: RegexNode.ToString is only compiled into debug builds, so DEBUG is currently set on the unit tests project.