ToTitleCase Perf improvements (#1442)

src/Benchmarks/TransformersBenchmarks.cs

@@ -0,0 +1,62 @@
+[MemoryDiagnoser(false)]
+public class TransformersBenchmarks
+{
+    // hard-coded seed ensures the same random strings are generated each time.
+    const int RAND_SEED = 17432;
+
+    static readonly char[] _alphabet =
+        Enumerable
+            .Repeat((int) ' ', 12)
+            .Concat(Enumerable.Range('a', 'z' - 'a'))
+            .Concat(Enumerable.Range('A', 'Z' - 'A'))
+            .Concat(Enumerable.Range('0', '9' - '0'))
+            .Concat(new int[]
+            {
+                '.',
+                ',',
+                '(',
+                ')',
+                '!',
+                '$'
+            })
+            .Select(x => (char) x)
+            .ToArray();
+
+    Random random = new(RAND_SEED);
+    string input = null!;
+
+    [Params(10, 100, 1000)]
+    public int StringLen;
+
+    [GlobalSetup]
+    public void GlobalSetup()
+    {
+        var chars = new char[StringLen];
+        for (var i = 0; i < StringLen; i++)
+        {
+            chars[i] = _alphabet[random.Next(0, _alphabet.Length)];
+        }
+
+        input = new(chars);
+    }
+
+    [Benchmark]
+    public string AllTransforms() =>
+        input.Transform(To.LowerCase, To.UpperCase, To.SentenceCase, To.TitleCase);
+
+    [Benchmark]
+    public string LowerCase() =>
+        input.Transform(To.LowerCase);
+
+    [Benchmark]
+    public string UpperCase() =>
+        input.Transform(To.UpperCase);
+
+    [Benchmark]
+    public string SentenceCase() =>
+        input.Transform(To.SentenceCase);
+
+    [Benchmark]
+    public string TitleCase() =>
+        input.Transform(To.TitleCase);
+}

src/Humanizer/Transformer/ToTitleCase.cs

@@ -1,54 +1,59 @@
-namespace Humanizer
+namespace Humanizer;
+
+class ToTitleCase : ICulturedStringTransformer
 {
-    class ToTitleCase : ICulturedStringTransformer
-    {
-        public string Transform(string input) =>
-            Transform(input, null);
+    public string Transform(string input) =>
+        Transform(input, null);
 
-        public string Transform(string input, CultureInfo? culture)
-        {
-            culture ??= CultureInfo.CurrentCulture;
+    static Regex regex = new(@"(\w|[^\u0000-\u007F])+'?\w*", RegexOptions.Compiled);
 
-            var result = input;
-            var matches = Regex.Matches(input, @"(\w|[^\u0000-\u007F])+'?\w*");
-            var firstWord = true;
-            foreach (Match word in matches)
+    public string Transform(string input, CultureInfo? culture)
+    {
+        culture ??= CultureInfo.CurrentCulture;
+        var matches = regex.Matches(input);
+        var builder = new StringBuilder(input);
+        var textInfo = culture.TextInfo;
+        foreach (Match word in matches)
+        {
+            var value = word.Value;
+            if (AllCapitals(value) || lookups.Contains(value))
             {
-                if (!AllCapitals(word.Value))
-                {
-                    result = ReplaceWithTitleCase(word, result, culture, firstWord);
-                }
-                firstWord = false;
+                continue;
             }
 
-            return result;
+            builder[word.Index] = textInfo.ToUpper(value[0]);
+            Overwrite(builder, word.Index + 1, textInfo.ToLower(value[1..]));
         }
 
-        static bool AllCapitals(string input) =>
-            input.All(char.IsUpper);
-
-        static string ReplaceWithTitleCase(Match word, string source, CultureInfo culture, bool firstWord)
-        {
-            var articles = new List<string> { "a", "an", "the" };
-            var conjunctions = new List<string> { "and", "as", "but", "if", "nor", "or", "so", "yet" };
-            var prepositions = new List<string> { "as", "at", "by", "for", "in", "of", "off", "on", "to", "up", "via" };
-
-            var wordToConvert = word.Value;
-            string replacement;
+        return builder.ToString();
+    }
 
-            if (firstWord ||
-                (!articles.Contains(wordToConvert) &&
-                !conjunctions.Contains(wordToConvert) &&
-                !prepositions.Contains(wordToConvert)))
-            {
-                replacement = culture.TextInfo.ToUpper(wordToConvert[0]) + culture.TextInfo.ToLower(wordToConvert.Remove(0, 1));
+    static void Overwrite(StringBuilder builder, int index, string replacement) =>
+        builder
+            .Remove(index, replacement.Length)
+            .Insert(index, replacement);
 
-            }
-            else
+    static bool AllCapitals(string input)
+    {
+        foreach (var ch in input)
+        {
+            if (!char.IsUpper(ch))
             {
-                replacement = culture.TextInfo.ToLower(wordToConvert);
+                return false;
             }
-            return source.Substring(0, word.Index) + replacement + source.Substring(word.Index + word.Length);
         }
+
+        return true;
+    }
+
+    static FrozenSet<string> lookups;
+
+    static ToTitleCase()
+    {
+        var articles = new List<string> { "a", "an", "the" };
+        var conjunctions = new List<string> { "and", "as", "but", "if", "nor", "or", "so", "yet" };
+        var prepositions = new List<string> { "as", "at", "by", "for", "in", "of", "off", "on", "to", "up", "via" };
+
+        lookups = articles.Concat(conjunctions).Concat(prepositions).ToFrozenSet();
     }
-}
+}