在C＃中转义字符串中字符的最快方法

Question

I'd like to find the fastest way to replace all reserved characters in a string with their escaped version.

There are two naive ways that come to my mind spontaneously (note that the set of reserved characters is just an example):

A: Using a lookup dictionary and String.Replace

private Dictionary<string, string> _someEscapeTokens = new Dictionary<string, string>()
{
    {"\t", @"\t"},
    {"\n", @"\n"},
    {"\r", @"\r"}
};

public string GetEscapedStringByNaiveLookUp(string s)
{
    foreach (KeyValuePair<string, string> escapeToken in _someEscapeTokens.Where(kvp => s.Contains(kvp.Key)))
    {
        s = s.Replace(escapeToken.Key, escapeToken.Value);
    }
    return s;
}

B: Traversing each character in the string

public string GetEscapedStringByTraversingCharArray(string s)
{
    char[] chars = s.ToCharArray();
    StringBuilder sb = new StringBuilder();
    for (int i = 0; i < chars.Length; i++)
    {
        switch (chars[i])
        {
            case '\t':
                sb.Append(@"\t"); break;
            case '\n':
                sb.Append(@"\n"); break;
            case '\r':
                sb.Append(@"\r"); break;
            default:
                sb.Append(chars[i]); break;
         }
    }
    return sb.ToString();
}

As I've already tested, Version B outperforms the first one easily.

Note: I already considered Regex.Escape but since the character set doesn't match mine it doesn't fit.

However are there other ways you would appraoch this problem (with performance in mind)?

Update: Testing

I've done some more testing and would like to share the results. See below for the code.

Testing was done on two different systems targeting the .Net Framework 4.0 . Anyway the results are pretty much the same:

Char Array (short string) average time: 38600 ns
Foreach (short string) average time: 26680 ns
Char Array (long string) average time: 48,1 ms
Foreach (long string) average time: 64,2 ms
Char Array (escaping only) average time: 13,6 ms
Foreach (escaping only) average time: 17,3 ms

Which leads me to the conclusion that the foreach version seems to be slightly faster for short strings but somehow "falls of" for longer strings. However we're talking about really small differences here.

Testing code:

private static void Main(string[] args)
{
    //around 700 characters
    string shortString = new StackTrace().ToString();
    string longString;
    string pureEscape;
    //loading from a file with 1000000 words http://loremipsum.de/
    using (StreamReader sr = new StreamReader(@"C:\users\ekrueger\desktop\LoremIpsum.txt"))
    {
        longString = sr.ReadToEnd();
    }
    //text file containing only escapable characters (length ~1000000)
    using (StreamReader sr = new StreamReader(@"C:\users\ekrueger\desktop\PureEscape.txt"))
    {
        pureEscape = sr.ReadToEnd();
    }
    List<double> timesCharArrayShortString = new List<double>();
    List<double> timesForeachShortString = new List<double>();
    List<long> timesCharArrayLongString = new List<long>();
    List<long> timesForeachLongString = new List<long>();
    List<long> timesCharArrayPureEscape = new List<long>();
    List<long> timesForeachPureEscape = new List<long>();
    Stopwatch sw = new Stopwatch();

    for (int i = 0; i < 10; i++)
    {
        sw.Restart();
        GetEscapedStringByTraversingCharArray(shortString);
        sw.Stop();
        timesCharArrayShortString.Add(sw.Elapsed.TotalMilliseconds * 1000000);
    }

    for (int i = 0; i < 10; i++)
    {
        sw.Restart();
        GetEscapedStringForeach(shortString);
        sw.Stop();
        timesForeachShortString.Add(sw.Elapsed.TotalMilliseconds * 1000000);
    }

    for (int i = 0; i < 10; i++)
    {
        sw.Restart();
        GetEscapedStringByTraversingCharArray(longString);
        sw.Stop();
        timesCharArrayLongString.Add(sw.ElapsedMilliseconds);
    }

    for (int i = 0; i < 10; i++)
    {
        sw.Restart();
        GetEscapedStringForeach(longString);
        sw.Stop();
        timesForeachLongString.Add(sw.ElapsedMilliseconds);
    }

    for (int i = 0; i < 10; i++)
    {
        sw.Restart();
        GetEscapedStringByTraversingCharArray(pureEscape);
        sw.Stop();
        timesCharArrayPureEscape.Add(sw.ElapsedMilliseconds);
    }

    for (int i = 0; i < 10; i++)
    {
        sw.Restart();
        GetEscapedStringForeach(pureEscape);
        sw.Stop();
        timesForeachPureEscape.Add(sw.ElapsedMilliseconds);
    }

    Console.WriteLine("Char Array (short string) average time: {0} ns", timesCharArrayShortString.Average());
    Console.WriteLine("Foreach (short string) average time: {0} ns", timesForeachShortString.Average());
    Console.WriteLine("Char Array (long string) average time: {0} ms", timesCharArrayLongString.Average());
    Console.WriteLine("Foreach (long string) average time: {0} ms", timesForeachLongString.Average());
    Console.WriteLine("Char Array (escaping only) average time: {0} ms", timesCharArrayPureEscape.Average());
    Console.WriteLine("Foreach (escaping only) average time: {0} ms", timesForeachPureEscape.Average());

    Console.Read();
}

private static string GetEscapedStringByTraversingCharArray(string s)
{
    if (String.IsNullOrEmpty(s))
        return s;

    char[] chars = s.ToCharArray();
    StringBuilder sb = new StringBuilder(s.Length);
    for (int i = 0; i < chars.Length; i++)
    {
        switch (chars[i])
        {
            case '\t':
                sb.Append(@"\t"); break;
            case '\n':
                sb.Append(@"\n"); break;
            case '\r':
                sb.Append(@"\r"); break;
            case '\f':
                sb.Append(@"\f"); break;
            default:
                sb.Append(chars[i]); break;
        }
    }
    return sb.ToString();
}

public static string GetEscapedStringForeach(string s)
{
    if (String.IsNullOrEmpty(s))
        return s;

    StringBuilder sb = new StringBuilder(s.Length);
    foreach (Char ch in s)
    {
        switch (ch)
        {
            case '\t':
                sb.Append(@"\t"); break;
            case '\n':
                sb.Append(@"\n"); break;
            case '\r':
                sb.Append(@"\r"); break;
            default:
                sb.Append(ch); break;
        }
    }
    return sb.ToString();
}

Answer 1

It makes sense that the first option is slower since you're creating a lot of string objects with:

s = s.Replace(escapeToken.Key, escapeToken.Value);

In the second method, there's no need to create a char[] because string has an indexer too. Perhaps the only thing you can do to improve performance is initialize the StringBuilder with a capacity, so it doesn't need to resize. You can still use the Dictionary in the second method.

Answer 2

You have no need to convert string into Char[] , so the solution can be slightly improved:

public string GetEscapedStringByTraversingCharArray(string s) {
  // do not forget about null...
  if (String.IsNullOrEmpty(s))
    return s;

  // we can be sure, that it requires at least s.Length symbols, let's allocate then
  // in case we want to trade memory for speed we can even put
  //   StringBuilder sb = new StringBuilder(2 * s.Length);
  // for the worst case when all symbols should be escaped
  StringBuilder sb = new StringBuilder(s.Length);

  foreach(Char ch in s) {
    switch (ch) {
      case '\t':
        sb.Append(@"\t"); 

        break;
      case '\n':
        sb.Append(@"\n"); 

        break;
      case '\r':
        sb.Append(@"\r"); 

        break;
      default:
        sb.Append(ch); 

        break;
    }
  }

  return sb.ToString();
}