On Sat, Sep 26, 2009 at 5:04 PM, Ramon F Herrera wrote:

I have been able to parse stuff more complicated than this, but now I am stuck with something seemingly simpler.

The expression being parsed is the common sequence:

 1,2-5,7,8-11

Question 1: Notice my approach. I first match the whole expression, with "regex_match", to make sure that it is valid (that works great). Next, I use "regex_iterator" to break down the parts. Is that good practice? Am I being inefficient/redundant?

Question 2: My code below only extracts "range terms" ("x-y"), for some reason I cannot extract "number terms".

As a workaround, I can always feed my data like this:

1-1,2-5,7-7,8-11

but, after a lot of tries, would love to learn how to do this properly.

TIA,

-Ramon

-----------------------------------------------------

#include <iostream> #include using namespace std;

bool term_callback(const boost::match_resultsstd::string::const_iterator& what) {    for (unsigned int i = 0; i < what.size(); i++) {        cout << "what[" << i << "]: " <<  what[i].str() << endl;        cout << "---------" << endl;    }    return true; }

int main(int argc, char *argv[]) {    const char hyphen      = '-';    const char left_paren  = '(';    const char right_paren = ')';    const char bar         = '|';    const char comma       = ',';    const char star        = '*';

   const string number    = "[0-9]+";    const string range     = number + hyphen + number;    const string term      = left_paren + number + bar + range + right_paren;    const string sequence  = term + bar + left_paren + term + comma + right_paren + star + term;

   boost::regex expression(sequence);    boost::regex piece(range);    boost::cmatch matches;

   char argument[1024];    strcpy(argument, argv[1]);

   if (!boost::regex_match(argument, matches, expression)) {        cerr << "There is no match" << endl;        return 1;    }

   string text = argument;

   boost::sregex_iterator m1(text.begin(), text.end(), piece);    boost::sregex_iterator m2;    for_each(m1, m2, &term_callback);

   return 0; }

Do note, if you are wanting to do something with your numbers, like convert them to numbers and do some operations on them, there is a much easier way to do this if you use Boost.Spirit2.1 instead of Boost.Regex. Your problem is more of a parsing problem then a matching problem, and regex is nice for matching, and Spirit2.1 is better for parsing. If you are interested then I or someone else could whip up some code that does the same thing in Spirit2.1, but will run a whole lot faster and be a lot easier to use.