hi,all,I need to process some strings, that for each string, I need to find some non-letter, non-digit symbols/splitters, and then use the them as separators to divide the string, e.g. I have a string

"Accent Plates by Lenox, Tuxedo", the comma "," is treated as such a symbol, and I need to use it as a separator to divide the string into two parts:

"Accent Plates by Lenox" and "Tuxedo". Now I have 2 questions:

1. I need to identify such symbols, and record them, I know that I can define the splitters using a regex like

      cout << "==> \"" << *i << "\"" << endl;
    }

cheers

daiyue 

Can you please read the posting guidelines in the sticky note at the start of the forum and try to provide responses t as many if the items as possible - in this case knowing the regex variant you are trying to use would help.

[Edit: I see that the normally present sticky note I refer to is missing. What follows is a brief outline. One of the items we need to know is the regex variant you are using as they differ in their syntax and capabilities; what works for one may be a syntax error for another. Also we generally ask for several examples of the input strings and the output you expect that cover a range of situations you will encounter. In line with this, we ask for real examples and not made up ones as these often hide the complexities that the regex will need to deal with and end up wasting everyone's time.]

The way most regex engines work with the "split" operation is to do what you are seeing in question 1) - i.e. it returns the text that is NOT the splitter itself. This of it this way: it uses the string as a "match" operation to locate where to make the splits; it then removes all of the matched parts and uses the remaining character sequences to make up the array of "split" values. (Of course this is not necessarily how any regex engine ACTUALLY works - but it gives the idea).

To return the characters that are being used for the "split" at each point, you need to know which regex variant you are using. Some (and I know the .NET regex variant does this) allow yuo to put capture parentheses around the split pattern and this will tell the regex engine to also return the characters matched.

For example, translating your pattern into the .NET form (and using '\w' instead of '[:alnum:]') you get a pattern of

\s*[^\w\s]+\s*

which splits the test string you gave into the 2 parts:

Accent Plates by Lenox
Tuxedo

However, adding in parentheses:

(\s*[^\w\s]+\s*)

gives the 3 parts:

Accent Plates by Lenox
,
Tuxedo

(there are spaces after the "," but they don't show up too well!).

Depending on the regex variant you are using, there may be ways of achieving the same effect.

Susan