.NET REGEX Matching matches empty strings

问题

I have this

pattern:

[0-9]*\.?[0-9]*

Target:

X=113.3413475 Y=18.2054775

And i want to match the numbers. It matches find in testing software like http://regexpal.com/ and Regex Coach.

But in Dot net and http://derekslager.com/blog/posts/2007/09/a-better-dotnet-regular-expression-tester.ashx

I get:

Found 11 matches:

1.
2.
3.
4.
5.
6.  113.3413475
7.
8.
9.
10. 18.2054775
11.

String literals for use in programs:

C#
    @"[0-9]*[\.]?[0-9]*"

Any one have any idea why i'm getting all these empty matches.

Thanks and Regards, Kevin

回答1:

Yes, that will match empty string. Look at it:

[0-9]* - zero or more digits
\.?    - an optional period
[0-9]* - zero or more digits

Everything's optional, so an empty string matches.

It sounds like you always want there to be digits somewhere, for example:

[0-9]+\.[0-9]*|\.[0-9]+|[0-9]+

(The order here matters, as you want it to take the most possible.)

That works for me:

using System;
using System.Text.RegularExpressions;

class Test
{
    static void Main(string[] args)
    {
        string x = "X=113.3413475 Y=18.2054775";
        Regex regex = new Regex(@"[0-9]+\.[0-9]*|\.[0-9]+|[0-9]+");
        var matches = regex.Matches(x);
        foreach (Match match in matches)
        {
            Console.WriteLine(match);
        }
    }
}

Output:

113.3413475
18.2054775

There may well be better ways of doing it, admittedly :)

回答2:

Another alternative is to keep your original regex, and just assert it must have a number in it (maybe after a dot):

[0-9]*\.?[0-9]*

Goes to:

(?=\.?[0-9])[0-9]*\.?[0-9]*

回答3:

Try this one:

[0-9]+(\.[0-9]+)?

It's slightly different that Jon Skeet's answer in that it won't match .45, it requires either a number alone (e.g. 8) or a real decimal (e.g. 8.1 or 0.1)

回答4:

The key problem is the *, which means "match zero or more of the preceding characters". The empty string matches zero or more digits, which is why you're getting all those matches.

Change your two *s to +s and you'll get what you want.

回答5:

The problem with this regex is that it is completely optional in all the fields, so an empty string also is matched by it. I would consider adding all the cases. By the regex, I see you want the numbers with or without dot, and with or without a set of decimal digits. You can separate first those that contain only numbers [0-9]+, then those that contain numbers plus only a dot, [0-9]+\. and then join them all with | (or).

The problem with the regex as it is is that it allows cases that are not real numbers, for example, the cases in which the first set of numbers and the last set of numbers are empty (just a dot), so you have to put the valid cases explicitly.

回答6:

Regex pattern = new Regex( @"[0-9]+[\.][0-9]+");

string info = "X=113.3413475 Y=18.2054775";

MatchCollection matches = pattern.Matches(info);

int count = 1;
foreach(Match match in matches)
{
    Console.WriteLine("{0} : {1}", count++, match.Value);
}

//output
//1 : 113.3413475
//2 : 18.2054775

Replace your * with + and remove ? from your period case.

EDIT: from above conversation: @"[0-9]+.[0-9]*|.[0-9]+|[0-9]+", is the better case. catches 123, .123, 123.123 etc

来源：https://stackoverflow.com/questions/7326679/net-regex-matching-matches-empty-strings