First cut of look up table for speeding up Go()

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

@@ -14,6 +14,7 @@
 // Strings and sets are indices into a string table.
 
 using System.Collections;
+using System.Collections.Generic;
 using System.Diagnostics;
 using System.Diagnostics.CodeAnalysis;
 
@@ -106,11 +107,13 @@ internal sealed class RegexCode
         public readonly int LeadingAnchor;                                              // the leading anchor, if one exists (RegexPrefixAnalyzer.Bol, etc)
         public readonly bool RightToLeft;                                               // true if right to left
 
+        public readonly Dictionary<char, HashSet<int>> FirstLetterToStringTableIndices;
+
         public RegexCode(RegexTree tree, int[] codes, string[] strings, int trackcount,
                          Hashtable? caps, int capsize,
                          RegexBoyerMoore? boyerMoorePrefix,
                          (string CharClass, bool CaseInsensitive)[]? leadingCharClasses,
-                         int leadingAnchor, bool rightToLeft)
+                         int leadingAnchor, bool rightToLeft, Dictionary<char, HashSet<int>> firstLetterToStringTableIndices)
         {
             Debug.Assert(boyerMoorePrefix is null || leadingCharClasses is null);
 
@@ -125,6 +128,7 @@ public RegexCode(RegexTree tree, int[] codes, string[] strings, int trackcount,
             LeadingCharClasses = leadingCharClasses;
             LeadingAnchor = leadingAnchor;
             RightToLeft = rightToLeft;
+            FirstLetterToStringTableIndices = firstLetterToStringTableIndices;
         }
 
         public static bool OpcodeBacktracks(int Op)

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

@@ -1,6 +1,7 @@
 // Licensed to the .NET Foundation under one or more agreements.
 // The .NET Foundation licenses this file to you under the MIT license.
 
+using System.Collections.Generic;
 using System.Diagnostics;
 using System.Diagnostics.CodeAnalysis;
 using System.Globalization;
@@ -1058,7 +1059,17 @@ protected override void Go()
                         continue;
 
                     case RegexCode.Multi:
-                        if (!MatchString(_code.Strings[Operand(0)]))
+                        int stringTableIndex = Operand(0);
+                        char textChar = runtext![runtextpos];
+                        if (_code.FirstLetterToStringTableIndices.TryGetValue(textChar, out HashSet<int>? stringTableIndices))
+                        {
+                            if (!stringTableIndices.Contains(stringTableIndex))
+                            {
+                                // We are trying a pattern that doesn't start with the right char, so there's no way we can match.
+                                break;
+                            }
+                        }
+                        if (!MatchString(_code.Strings[stringTableIndex]))
                         {
                             break;
                         }

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

@@ -172,14 +172,25 @@ public RegexCode RegexCodeFromRegexTree(RegexTree tree)
             int leadingAnchor = RegexPrefixAnalyzer.FindLeadingAnchor(tree);
 
             // Convert the string table into an ordered string array.
+            var firstLetterToStringTableIndices = new Dictionary<char, HashSet<int>>();
             var strings = new string[_stringTable.Count];
             foreach (KeyValuePair<string, int> stringEntry in _stringTable)
             {
+                if (firstLetterToStringTableIndices.TryGetValue(stringEntry.Key[0], out HashSet<int>? indices))
+                {
+                    indices.Add(stringEntry.Value);
+                }
+                else
+                {
+                    firstLetterToStringTableIndices.Add(stringEntry.Key[0], new HashSet<int>() { stringEntry.Value });
+                }
                 strings[stringEntry.Value] = stringEntry.Key;
             }
 
+
+
             // Return all that in a RegexCode object.
-            return new RegexCode(tree, emitted, strings, _trackCount, _caps, capsize, boyerMoorePrefix, leadingCharClasses, leadingAnchor, rtl);
+            return new RegexCode(tree, emitted, strings, _trackCount, _caps, capsize, boyerMoorePrefix, leadingCharClasses, leadingAnchor, rtl, firstLetterToStringTableIndices);
         }
 
         /// <summary>

src/libraries/System.Text.RegularExpressions/tests/Regex.Match.Tests.cs

@@ -1175,5 +1175,18 @@ public void Synchronized()
 
             AssertExtensions.Throws<ArgumentNullException>("inner", () => System.Text.RegularExpressions.Match.Synchronized(null));
         }
+
+        [Fact]
+        public void HowManyAlternationsAreChecked()
+        {
+            // We can statically determine if it's impossible for an alternation branch N + 1 to match after we've gotten to a certain place in matching branch N, e.g. given the alternation "abc|def" we know that once we match the 'a', there's no point in even considering the second branch. We should be able to utilize that knowledge to avoid unnecessarily checking branches when a previous one fails to match.
+
+            Debugger.Launch();
+            var regex = new Regex("(abc|def)xyz");
+            var match = regex.Match("abqabqabqabqabqabqabqabqabqabqabqabqabqabqabqabqabqabqdefxyz");
+            Assert.True(match.Success);
+            Assert.Equal("defxyz", match.Value);
+            Console.WriteLine("BH");
+        }
     }
 }