Regexp MatchCollection

[drkefir]

I have a problem with my regexp.
The input to match are "telegrams" of various length.
"Fields" within a telegram are separated by ","
If a "," char is part of a field the field must be inclosed by quotes: (field1,"field2a,field2b",field3,...)

The problem with the regexp I'm using is that it does not parse "empty fields" correct.

This telegram will return 6 matches:
"04",N,1,2,3,4
This telegram will return 6 matches:
"04",N,1,2,,4
But this telegram will only return 5 matches:
"04",N,1,,,4

private static MatchCollection doMatch(string telegramString)
{
   string pattern = "(\".+?\"|[^,]+|,,)";
   MatchCollection col = Regex.Matches(telegramString, pattern);
   return col;
}

I have very little experience of regexp, please help.

//Fredrik.

[tlhintoq]

Does it have to be a RegEx?
Since all your items are separated by commas you could use String.Split which will give you a string[] in return. Some of your elements will be empty, but you will get back all your fields in correct order including the empties.

[drkefir]

tlhintoq! I have considerd that approach. I cant just use the telegramString.Split(new char[] { ',' }); out of the box since the fields might include the comma char (should be fairly easy to solve with som ugly search and replace etc. before doing the Split)
But when looking at the design I have base class that implements methods like this one:
protected abstract bool checkFieldCount(MatchCollection col);
And Im not that found of breaking the design at this point. (multiple classes/implementations derives from this).

Convert.ToMatchCollection(string[]) ??????

[tlhintoq]

If your data contains commas... and your field delimiters are commas... I don't see how any system would know where to break...

Unless your data with commas is within quotes. Is it?

[drkefir]

Rtfm ;) ......................................