boost_1_45_0/libs/xpressive/doc/named_captures.qbk

[/
 / Copyright (c) 2009 Eric Niebler
 /
 / Distributed under the Boost Software License, Version 1.0. (See accompanying
 / file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
 /]

[section Named Captures]

[h2 Overview]

For complicated regular expressions, dealing with numbered captures can be a
pain. Counting left parentheses to figure out which capture to reference is
no fun. Less fun is the fact that merely editing a regular expression could
cause a capture to be assigned a new number, invaliding code that refers back
to it by the old number.

Other regular expression engines solve this problem with a feature called
/named captures/. This feature allows you to assign a name to a capture, and
to refer back to the capture by name rather by number. Xpressive also supports
named captures, both in dynamic and in static regexes.

[h2 Dynamic Named Captures]

For dynamic regular expressions, xpressive follows the lead of other popular
regex engines with the syntax of named captures. You can create a named capture
with `"(?P<xxx>...)"` and refer back to that capture with `"(?P=xxx)"`. Here,
for instance, is a regular expression that creates a named capture and refers
back to it:

    // Create a named capture called "char" that matches a single
    // character and refer back to that capture by name.
    sregex rx = sregex::compile("(?P<char>.)(?P=char)");

The effect of the above regular expression is to find the first doubled
character.

Once you have executed a match or search operation using a regex with named
captures, you can access the named capture through the _match_results_ object
using the capture's name.

    std::string str("tweet");
    sregex rx = sregex::compile("(?P<char>.)(?P=char)");
    smatch what;
    if(regex_search(str, what, rx))
    {
        std::cout << "char = " << what["char"] << std::endl;
    }

The above code displays:

[pre
char = e
]

You can also refer back to a named capture from within a substitution string.
The syntax for that is `"\\g<xxx>"`. Below is some code that demonstrates how
to use named captures when doing string substitution.

    std::string str("tweet");
    sregex rx = sregex::compile("(?P<char>.)(?P=char)");
    str = regex_replace(str, rx, "**\\g<char>**", regex_constants::format_perl);
    std::cout << str << std::endl;

Notice that you have to specify `format_perl` when using named captures. Only
the perl syntax recognizes the `"\\g<xxx>"` syntax. The above code displays:

[pre
tw\*\*e\*\*t
]

[h2 Static Named Captures]

If you're using static regular expressions, creating and using named
captures is even easier. You can use the _mark_tag_ type to create
a variable that you can use like [globalref boost::xpressive::s1 `s1`],
[globalref boost::xpressive::s1 `s2`] and friends, but with a name
that is more meaningful. Below is how the above example would look 
using static regexes:

    mark_tag char_(1); // char_ is now a synonym for s1
    sregex rx = (char_= _) >> char_;

After a match operation, you can use the `mark_tag` to index into the
_match_results_ to access the named capture:

    std::string str("tweet");
    mark_tag char_(1);
    sregex rx = (char_= _) >> char_;
    smatch what;
    if(regex_search(str, what, rx))
    {
        std::cout << what[char_] << std::endl;
    }

The above code displays:

[pre
char = e
]

When doing string substitutions with _regex_replace_, you can use named
captures to create /format expressions/ as below:

    std::string str("tweet");
    mark_tag char_(1);
    sregex rx = (char_= _) >> char_;
    str = regex_replace(str, rx, "**" + char_ + "**");
    std::cout << str << std::endl;

The above code displays:

[pre
tw\*\*e\*\*t
]

[note You need to include [^<boost/xpressive/regex_actions.hpp>] to
use format expressions.]

[endsect]