[/ 
  Copyright 2006-2007 John Maddock.
  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:captures Understanding Marked Sub-Expressions and Captures]

Captures are the iterator ranges that are "captured" by marked 
sub-expressions as a regular expression gets matched.  Each marked 
sub-expression can result in more than one capture, if it is matched 
more than once.  This document explains how captures and marked 
sub-expressions in Boost.Regex are represented and accessed.

[h4 Marked sub-expressions]

Every time a Perl regular expression contains a parenthesis group `()`, it 
spits out an extra field, known as a marked sub-expression, 
for example the expression:

[pre (\w+)\W+(\w+)]

Has two marked sub-expressions (known as $1 and $2 respectively), in 
addition the complete match is known as $&, everything before the 
first match as $\`, and everything after the match as $'.  So 
if the above expression is searched for within `"@abc def--"`, then we obtain:

[table
[[Sub-expression][Text found]]
[[$\`]["@"]]
[[$&]["abc def"]]
[[$1]["abc"]]
[[$2]["def"]]
[[$']["--"]]
]

In Boost.Regex all these are accessible via the [match_results] class that 
gets filled in when calling one of the regular expression matching algorithms 
([regex_search], [regex_match], or [regex_iterator]).  So given:

   boost::match_results<IteratorType> m;

The Perl and Boost.Regex equivalents are as follows:

[table 
[[Perl][Boost.Regex]]
[[$\`][`m.prefix()`]]
[[$&][`m[0]`]]
[[$n][`m[n]`]]
[[$\'][`m.suffix()`]]
]

In Boost.Regex each sub-expression match is represented by a [sub_match] object, 
this is basically just a pair of iterators denoting the start and end 
position of the sub-expression match, but there are some additional 
operators provided so that objects of type [sub_match] behave a lot like a 
`std::basic_string`: for example they are implicitly convertible to a 
`basic_string`, they can be compared to a string, added to a string, or 
streamed out to an output stream.

[h4 Unmatched Sub-Expressions]

When a regular expression match is found there is no need for all of the 
marked sub-expressions to have participated in the match, for example the expression:

[pre (abc)|(def)]

can match either $1 or $2, but never both at the same time.  In Boost.Regex 
you can determine which sub-expressions matched by accessing the 
`sub_match::matched` data member.

[h4 Repeated Captures]

When a marked sub-expression is repeated, then the sub-expression gets 
"captured" multiple times, however normally only the final capture is available, 
for example if

[pre (?:(\w+)\W+)+]

is matched against

[pre one fine day]

Then $1 will contain the string "day", and all the previous captures will have 
been forgotten.

However, Boost.Regex has an experimental feature that allows all the capture 
information to be retained - this is accessed either via the 
`match_results::captures` member function or the `sub_match::captures` member 
function.  These functions return a container that contains a sequence of all 
the captures obtained during the regular expression matching.  The following 
example program shows how this information may be used:

   #include <boost/regex.hpp>
   #include <iostream>

   void print_captures(const std::string& regx, const std::string& text)
   {
      boost::regex e(regx);
      boost::smatch what;
      std::cout << "Expression:  \"" << regx << "\"\n";
      std::cout << "Text:        \"" << text << "\"\n";
      if(boost::regex_match(text, what, e, boost::match_extra))
      {
         unsigned i, j;
         std::cout << "** Match found **\n   Sub-Expressions:\n";
         for(i = 0; i < what.size(); ++i)
            std::cout << "      $" << i << " = \"" << what[i] << "\"\n";
         std::cout << "   Captures:\n";
         for(i = 0; i < what.size(); ++i)
         {
            std::cout << "      $" << i << " = {";
            for(j = 0; j < what.captures(i).size(); ++j)
            {
               if(j)
                  std::cout << ", ";
               else
                  std::cout << " ";
               std::cout << "\"" << what.captures(i)[j] << "\"";
            }
            std::cout << " }\n";
         }
      }
      else
      {
         std::cout << "** No Match found **\n";
      }
   }

   int main(int , char* [])
   {
      print_captures("(([[:lower:]]+)|([[:upper:]]+))+", "aBBcccDDDDDeeeeeeee");
      print_captures("(.*)bar|(.*)bah", "abcbar");
      print_captures("(.*)bar|(.*)bah", "abcbah");
      print_captures("^(?:(\\w+)|(?>\\W+))*$", 
         "now is the time for all good men to come to the aid of the party");
      return 0;
   }

Which produces the following output:

[pre 
Expression:  "(([[:lower:\]\]+)|([[:upper:\]\]+))+"
Text:        "aBBcccDDDDDeeeeeeee"
'''**''' Match found '''**'''
   Sub-Expressions:
      $0 = "aBBcccDDDDDeeeeeeee"
      $1 = "eeeeeeee"
      $2 = "eeeeeeee"
      $3 = "DDDDD"
   Captures:
      $0 = { "aBBcccDDDDDeeeeeeee" }
      $1 = { "a", "BB", "ccc", "DDDDD", "eeeeeeee" }
      $2 = { "a", "ccc", "eeeeeeee" }
      $3 = { "BB", "DDDDD" }
Expression:  "(.'''*''')bar|(.'''*''')bah"
Text:        "abcbar"
'''**''' Match found '''**'''
   Sub-Expressions:
      $0 = "abcbar"
      $1 = "abc"
      $2 = ""
   Captures:
      $0 = { "abcbar" }
      $1 = { "abc" }
      $2 = { }
Expression:  "(.'''*''')bar|(.'''*''')bah"
Text:        "abcbah"
'''**''' Match found '''**'''
   Sub-Expressions:
      $0 = "abcbah"
      $1 = ""
      $2 = "abc"
   Captures:
      $0 = { "abcbah" }
      $1 = { }
      $2 = { "abc" }
Expression:  "^(?:(\w+)|(?>\W+))'''*$'''"
Text:        "now is the time for all good men to come to the aid of the party"
'''**''' Match found '''**'''
   Sub-Expressions:
      $0 = "now is the time for all good men to come to the aid of the party"
      $1 = "party"
   Captures:
      $0 = { "now is the time for all good men to come to the aid of the party" }
      $1 = { "now", "is", "the", "time", "for", "all", "good", "men", "to", 
         "come", "to", "the", "aid", "of", "the", "party" }
]

Unfortunately enabling this feature has an impact on performance 
(even if you don't use it), and a much bigger impact if you do use it, 
therefore to use this feature you need to:

* Define BOOST_REGEX_MATCH_EXTRA for all translation units including the library source (the best way to do this is to uncomment this define in boost/regex/user.hpp and then rebuild everything.
* Pass the match_extra flag to the particular algorithms where you actually need the captures information (regex_search, regex_match, or regex_iterator).
    
[endsect]