Is Regex instance thread safe for matches in C#

2019-04-18 00:24发布

问题:

I have this regex which I am using in a Parallel.ForEach<string>. Is it safe?

Regex reg = new Regex(SomeRegexStringWith2Groups);
Parallel.ForEach<string>(MyStrings.ToArray(), (str) =>
{
    foreach (Match match in reg.Matches(str)) //is this safe?
        lock (dict) if (!dict.ContainsKey(match.Groups[1].Value))
            dict.Add(match.Groups[1].Value, match.Groups[2].Value);
});

回答1:

Regex objects are read-only, and therefore are thread safe. It's their returns, the Match objects that could potentially cause problems. MSDN confirms this:

The Regex class itself is thread safe and immutable (read-only). That is, Regex objects can be created on any thread and shared between threads; matching methods can be called from any thread and never alter any global state.

However, result objects (Match and MatchCollection) returned by Regex should be used on a single thread ..

I'd be concerned about how your Match collection is being generated in a way that might be concurrent, which could cause the collection to act kinda weird. Some Match implementations use delayed evaluation, which could cause some crazy behavior in that foreach loop. I would probably collect all the Matches and then evaluate them later, both to be safe and to get consistent performance.