RegEx vs字符串操作函数：什么是更好的

Question

If I have to find let's say a word in a sentence, i can think of two approaches

1. Using string.IndexOf
2. Using Regex

Which one is better in terms of performance or best practice

Answer 1

If it's fairly straightforward to do something without regex, it's almost always cheaper that way. String.IndexOf (or String.Contains ) is definitely an example of this.

Answer 2

It depends on your exact requirements. If you truly need to find a word in a sentence (not a substring), then I believe that could be expressed more concisely and more explicitly using a well-named regex pattern than using IndexOf plus all the extra logic to make sure you're actually getting a complete single word.

On the other hand, if you're simply looking for a substring, then IndexOf is far superior in terms of performance and readability.

Answer 3

This is by no means the most scientific way of measuring things but here is a bit of source code that indicates (under very specific constraints) regex is about 4 times slower then indexof.

class Program
{
private const string Sentence = "The quick brown fox jumps over the lazy dog";
private const string Word = "jumps";

static void Main(string[] args)
{
    var indexTimes = new List<long>();
    var regexTimes = new List<long>();
    var timer = new Stopwatch();

    for (int i = 0; i < 1000; i++)
    {
        timer.Reset();
        timer.Start();
        Sentence.IndexOf(Word);
        timer.Stop();
        indexTimes.Add(timer.ElapsedTicks);
    }

    Console.WriteLine(indexTimes.Average());

    for (int i = 0; i < 1000; i++)
    {
        timer.Reset();
        timer.Start();
        Regex.Match(Sentence, Word);
        timer.Stop();
        regexTimes.Add(timer.ElapsedTicks);
    }

    Console.WriteLine(regexTimes.Average());

    Console.ReadLine();
}
}

Answer 4

In terms of best practices, string.IndexOf is probably a little more obvious to someone reading the code. People's brains tend to close up as soon as they see a regular expression, so something straight-forward like IndexOf would keep their brains open.

As for performance, that's dependent on a lot of things and can only be properly answered through benchmarking of specific code.