[Boost-users] seeking a better way of regex replacing!

From: llwaeva_at_[hidden]
Date: 2006-07-30 05:24:05

Hi there,
  boost:regex is powerful in string replacing. I am working with a
complicate case which need a faster method in replacing substring. Here
is my problem.

I have a big string contains some pattern %K{xxx}, where xxx is a
filename which may vary from place to place. e.g.

BIG_STRING = 
...
...
%K{abc.txt}
...
...
...
...
%K{foo.ini}
...
...

OK, what complicated is that: I must retrieve the filename from %K{xxx}, load
the content of the file and then replace %K{xxx} with the content. I
search the code with a loop

 string::const_iterator start, end; 
 string filename, content;
 boost::regex re("%K\\{(.*)\\}");
 boost::match_results<std::string::const_iterator> what; 
 boost::match_flag_type flags = boost::match_default; 
 start = BIG_STRING.begin();
 end = BIG_STRING.end();

 while ( regex_search(start, end, what, re, flags) )
 {
    if  (what[1].first!=what[1].second)
    {
       filename = string(what[1].first, what[1].second);
    }
    content =   LoadFile(filename);
    // transform the const_iterator to iterator
    string::iterator sbegin(BIG_STRING.begin()), send(BIG_STRING.begin());       
    std::advance(sbegin, distance<string::const_iterator>(sbegin, what[0].first));
    std::advance(send, distance<string::const_iterator>(send, what[0].second));

    // replace the content
    BIG_STRING.replace( sbegin, send, content  );

    // reset the current position of the source string to search the next pattern
    start = what[1].second; 

    // I copy the code from the document of boost, don't know what the following code for
    flags |= boost::match_prev_avail; 
    flags |= boost::match_not_bob; 
 } 

The second approach is to retrieve the filename and load the content seperately, build the
substitute format and replace all the patterns with replace_all_regex, e.g.

string& escape_format( string& input )
{
  replace_all( input, "$", "$$" );
  replace_all( input, "\\", "\\\\" );
  replace_all( input, ":", "\\:" );
  replace_all( input, "?", "\\?" );
  replace_all( input, "(", "\\(" );
  replace_all( input, ")", "\\)" );
  return input;
}

 stringstream reformat;
 string regex_text;
 int n=0;
 while ( regex_search(start, end, what, re, flags) )
 {
    if  (what[1].first!=what[1].second)
    {
       filename = string(what[1].first, what[1].second);
    }
    content =   LoadFile(filename);
    regex_text += string(what[0].first, what[0].second);
    reformat << "(?" << ++n << " " << escape_format( content ) << ")";

    // reset the current position of the source string to search the next pattern
    start = what[1].second; 

    // I copy the code from the document of boost, don't know what the following code for
    flags |= boost::match_prev_avail; 
    flags |= boost::match_not_bob; 
 } 

// replace  the whole string at last
replace_all_regex( BIG_STRING, boost::regex(regex_text), reformat.str(), format_all );

But this approach cause exception!!! 

Any better idea for my problem?

Thanks in advance.