C program, that prints its executable file name

Question

Suppose source code file name is test.cpp. When it is compiled, it produce test.exe file. When I execute it, it should identify its file name that is test.exe and print it.

I'm able to get list of all the files and directories present in the current directory using following code:

DIR * directory;
struct dirent * direct;
direct = readdir(directory);

But, how can I identify the associated file name, in this case which is "test.exe"?

Answer 1

In your main function, argv[0] is the name of the executable from the command line

#include <stdio.h>
int main(int argc, char ** argv)
{
    printf("%s", argv[0]);
    return 0;
}

Live Demo

This prints the command name, which is the directory relative to the current working directory, plus the executable name (if available, which is not guaranteed) To get the current working directory, use getcwd() standard library C function.

Extracting the file name from the command path in argv[0] is platform specific : unix use slashes '/', windows allows mixed uses of slash / and backslash \, and any other platform could use any other path separator. Extracting the file name from a path requires a cross-platform library, such as Qt or Boost. On POSIX environments, basename could be used.

Answer 2

#include <stdio.h>

int main(int argc, char *argv[])
{
    printf("%s\n", argv[0]);
    return 0;
}

Note that your program can be launched as:

/home/user/./app

In this case you can get the name by using strrchr:

#include <stdio.h>
#include <string.h>

int main(int argc, char *argv[])
{
    char *appname;

    appname = strrchr(argv[0], '/'); /* '\\' on Windows */
    printf("%s\n", appname ? ++appname : argv[0]);
    return 0;
}

Answer 3

You know the name of the executable when you're building it; the simplest solution is to embed it in the program, using a -D or /D option to define a macro on the command line.

Other than that, the generic answer is that it isn't possible:

According to the standard

• argv[0] should contain the name which was used to invoke the program (whatever that means). Which is nice, but 1) it isn't even implementable under Unix, and 2) under most systems, there are all sorts of aliasing which means that the name used to invoke the program bears no relationship to the name of the executable.

Under Windows

• There's a system function GetModuleFileName which can be used to obtain the path of the executable. Once you've got the path, the last element of the path is the name of your executable.

Under Unix

• It's fundamentally impossible. When starting a new process, Unix takes separate arguments for the path to the executable and for what ends up in argv[0], so they can potentially have no relationship to one another. It all depends on who starts your process. bash will put the full path to the executable in the environment variable "_", so you can use getenv to get it. But that only works if your program was started by bash. On most Unices, you can also find it in the /proc filesystem, if you know your way around there; but the organization of this varies from one Unix to the next. Note too that because of hard links, you're executable may not have just one name.

The real question is why you want to do this. What is the problem you are trying to solve?

Answer 4

On Linux specifically:

you might use proc(5) and the /proc/self/exe symlink, so use readlink(2) on that. Once you did that, you could use basename(3) or realpath(3) on the obtained symlink.

However, be aware that a program might not always have a file path. It could have several file paths (for example /bin/rbash is a symlink to /bin/bash, and the shell process behave differently when invoked as rbash or as bash). Sometimes a given file (actually an inode, see inode(7)) has several hardlinks to it. In weird cases there is none. And it could happen that a program is execve(2)-ed and then removed with unlink(2) (perhaps from another process scheduled to run before yours) etc..

^{(BTW, your question is OS specific; readdir(3) is POSIX and there are some operating systems not even having directories, and readdir is not mentioned by the C11 standard; check by reading n1570)}

I experimented with the following (pathological) ./selfremove program (in my /home/basile/tmp/ directory) which removes its own binary:

 // file selfremove.c
 #include <stdio.h>
 #include <stdlib.h>
 #include <unistd.h>
 #include <string.h>

 int
 main (int argc, char **argv)
 {
   char selfpath[128];
   memset (selfpath, 0, sizeof (selfpath));
   if (readlink ("/proc/self/exe", selfpath, sizeof (selfpath) - 1) < 0)
     {
       perror ("first readlink");
       exit (EXIT_FAILURE);
     };
   printf ("initial /proc/self/exe -> %s\n", selfpath);
   if (unlink (argv[0]))
     {
       fprintf (stderr, "unlink %s: %m\n", argv[0]);
       exit (EXIT_FAILURE);
     };
   printf ("%s unlinked\n", argv[0]);
   if (readlink ("/proc/self/exe", selfpath, sizeof (selfpath) - 1) < 0)
     {
       perror ("second readlink");
       exit (EXIT_FAILURE);
     };
   printf ("final /proc/self/exe -> %s\n", selfpath);
   return 0;
 }    

It works, and the kernel is constructing a * (deleted) symlink (since a perverse coder might rename an executable as selfremove (deleted), so that kernel added suffix is only an indication ....):

 % ./selfremove 
 initial /proc/self/exe -> /home/basile/tmp/selfremove
 ./selfremove unlinked
 final /proc/self/exe -> /home/basile/tmp/selfremove (deleted)

So even with /proc/self/exe you cannot always trust the result.

If you assume that your program has been execve(2)-ed by some shell (or similar program doing execvp(3)) -and that is not always the case- then the PATH variable might have been used (and searched from your main's argv[0] if it has no /). You might use getenv(3) as getenv("PATH") to get it from your environment (see environ(7) for more). That $PATH is generally set and used, but they are pathological cases too.

So in general there is no way to reliably print its own executable (as my pathological selfremove.c demonstrates). In most cases you could find it (e.g. by readlink of /proc/self/exe, or by searching in $PATH using argv[0]).

Answer 5

In Linux systems, I think you can also use basename(char *path); from libgen.h, try $ man basename:

Print NAME with any leading directory components removed. If specified, also remove a trailing SUFFIX.

I tried it as:

// filename.c
#include<stdio.h>
#include<stdlib.h>
#include<libgen.h>
int main(int argc, char* argv[]){
    char* exe_name = basename(argv[0]);
    printf(" Executable Name: %s", exe_name);
    printf("\n");
    return EXIT_SUCCESS;
}

Observations,compiled and tested it as:

taxspanner@:~$ ls filename.c
filename.c
taxspanner@:~$ gcc -std=gnu99 -Wall -pedantic filename.c -o filename
taxspanner@:~$ ls filename*
filename  filename.c
taxspanner@:~$ 

Now run it in current directory:

taxspanner@:~$ ./filename 
 Executable Name: filename

Using absolute path:

taxspanner@:~$ pwd
/home/taxspanner
taxspanner@:~$ /home/taxspanner/filename 
 Executable Name: filename

Relative path:

taxspanner@:~$ cd study/divide-5/
taxspanner@:~/study/divide-5$ pwd
/home/taxspanner/study/divide-5
taxspanner@:~/study/divide-5$ ls ../../filename*
../../filename  ../../filename.c
taxspanner@:~/study/divide-5$ ../../filename
 Executable Name: filename
taxspanner@:~/study/divide-5$ 

one more try with executable name with suffix:

taxspanner@:~$ gcc -std=gnu99 -Wall -pedantic filename.c -o filename.out
taxspanner@:~$ ./filename.out 
 Executable Name: filename.out
taxspanner@:~$ cd study/divide-5/
taxspanner@:~/study/divide-5$ ../../filename.out 
 Executable Name: filename.out

Be careful when using it: Both dirname() and basename() may modify the contents of path, so it may be desirable to pass a copy when calling one of these functions.

Give it a try!!