[boost] Still confused with regex++

From: John Maddock (John_Maddock_at_[hidden])
Date: 2001-06-12 06:26:05

>I want to use a regular expression that can pick two separate fields 
>in the above message. For example I wish to search for messages that 
>include "tx_seq_num = 3" AND "ack_seq_num = 2" but I am not 
>interested to see messages that include only one of the two.

That's not something that regex's do, or at least I've never seen one that
does that.

>A way I thought was to use an ORed expression 
>i.e. "tx_seq_num=3|ack_seq_num = 2" (which works fine btw) and then 
>check [what] to see what was matched, if both fields were matched 
>then I want to see the message else discard it.
>
>But I am not very sure if in that case the algorithm after finding 
>the first match tries to look for the second or returns true with the 
>first match (possibly for optimization issues).

It returns the first thing that matches, in your example case that would be
the "tx_seq_num = 3" field.

One way to do what you want would be to use regex_grep, and keep a track or
how many unique matches were found - in fact if each field occures only
once regex_grep does this for you as it returns the total number of matches
found:

//
// this regex assumes that each field starts a new line (speeds up
seaches):
//
boost::regex re("^(?:tx_seq_num|ack_seq_num)\\s*=\\s*\\d+");

//
// this predicate stops the grep once we've found N matches
// (optimisation: there's no point in continuing once we've found as many
as we want),
// we could also store the matches if that was desired:
//
struct counting_predicate
{
        unsigned target;
        unsigned count;
        counting_predicate(int i) : target(i), count(0) {}
        bool operator()(const
boost::match_results<std::string::const_iterator>&)
        { return ++count < target; }
};

bool check_message(const std::string& msg)
{
        counting_predicate pred(2);
        return (regex_grep(pred, msg, re) > 1);
}

- John Maddock
http://ourworld.compuserve.com/homepages/john_maddock/