Question

I am trying to write some code that changes a string like “/path/file.extension” to another specified extension. I am trying to use boost::xpressive to do so. But, I am having problems. It appears that a garbage character appears in the output:

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

using namespace boost::xpressive;
using namespace std;


int main()
{
    std::string str( "xml.xml.xml.xml");
    sregex date = sregex::compile( "(\\.*)(\\.xml)$");

    std::string format( "\1.zipxml");


    std::string str2 = regex_replace( str, date, format );
    std::cout << "str  = " << str << "\n";
    std::cout << "str2 = " << str2 << "\n";

    return 0;
}

Now compile and run it:

[bitdiot@kantpute foodir]$ g++ badregex.cpp
[bitdiot@kantpute foodir]$ ./a.out > output
[bitdiot@kantpute foodir]$ less output
[bitdiot@kantpute foodir]$ cat -vte output
str  = xml.xml.xml.xml$
str2 = xml.xml.xml^A.zipxml$

In the above example, I redirect output to a file, and use cat to print out the non-printable character. Notice the ctrl-A in the str2.

Anyways, am I using boost libraries incorrectly? Is this a boost bug? Is there another regular expression I can use that can allow me to string replace the “.tail” with some other string? (It’s fix in my example.)

thanks.

Answer 1

At least as I’m reading things, the culprit is right here: std::string format( "\1.zipxml");.

You forgot to escape the backslash, so \1 is giving you a control-A. You almost certainly want \\1.

Alternatively (if your compiler is new enough) you could use a raw string instead, so it would be something like: R"(\1.zipxml)", and you wouldn’t have to escape your backslashes. I probably wouldn’t bother to mention this, except for the fact that if you’re writing REs in C++ strings, raw strings are pretty much your new best friend (IMO, anyway).