[RESOLVED] Regex problem

[rjv_rnjn]

Hi,

I've used the below regex expresssion to strip all the characters from a string value and only return the numbers.

String.Join(Nothing, System.Text.RegularExpressions.Regex.Split(strValue, "[^\d]"))

If the input is: abc12a
Output is: 12
The problem is that if I pass abc12.23a then the output is 1223 while I want is 12.23
Can somebody suggest what should I do?

Thanks.

[Coool]

Well And what if Input is : abc12d34 ???

[rjv_rnjn]

Though interesting input, but that won't happen. The reason is that I'm trying to take price out of a string. So price can be $100,234.55MM etc but never in the format that you've specified. And if by any chance my method gets that and computes incorrectly, that won't be an application issue.

[Bulldog]

How about "[0-9]*\.?[0-9]*" or to be more compact "[\d]*\.?[\d]*"

[bmahler]

how bout if you just do a replace on a-z characters?

System.Text.RegularExpressions.Regex.Replace("abc12.34ac", "[a-zA-Z]*", String.Empty)

[gigemboy]

All you had to do originally is add in the "." to your pattern in your very first post. Since it is a special character, escape it with a slash, so the pattern you needed was just "[^\d\.]"

Dim strValue As String = "abc12.34abc"
MessageBox.Show(String.Join(Nothing, System.Text.RegularExpressions.Regex.Split(strValue, "[^\d\.]")))

[rjv_rnjn]

Thanks gigemboy's. That provoked me to read a more about Regex. I think I'll take some more time to get a full hold on it.
@bmahler's solution won't work on $12.5MM.
Bulldog, you didn't even try to test the solution you proposed. It strips all the numbers from the string that is passed.
Thanks to all anyways who took the pain to help me out.

[Bulldog]

Bulldog, you didn't even try to test the solution you proposed. It strips all the numbers from the string that is passed.

What I was showing you was how to match the number sequence you require thinking that you would use this to do a match. So rather than using a "split" which generates an array of segments to sort through, you could use a "match" and generate a match collection. It was tested in RegExBuddy btw.

You are currently separating out the string sequence you want by removing anything that isnt a "." or a number. This may work today, but for future proofing I would be inclined to match what you "know" to be there, rather than removing the "unknowns" around it. In the future, the database may end up with "Sold,3kilos,at,$125k" or "Note:-Refund,-35$" etc.

Over time you can easily make a regex match more specific, to deal with negative numbers you can add an optional "-" at the start of the expression, you could also deal with extra spaces, different currency symbols, or the use of a trailing "k" for thousand etc. etc. It would be extremely difficult to do this correctly by removing the characters you dont expect.

About half of my code is always consumed by error checking, a good mantra being assume nothing! and expect the unexpected!

Anyway just my $0.02...