Search code examples
cbashgccmallocstrtok

Parse $PATH variable and save the directory names into an array of strings


I want to parse the $PATH variable of Linux, and then save the directory names that are getting separated with ':' into an array of strings.

I know it's a simple task but I am stuck and any help would be nice.

My code so far is something like this but something ain't right.

char **array;
char *path_string;
char *path_var = getenv("PATH");
int size_of_path_var = strlen(path_var);

path_string = strtok(path_var, ":");
while (path_string != NULL) {
    ss = strlen(path_string)
    array[i] = (char *)malloc(ss + 1);
    array[i] = path_string; //this is actually all i want to do for every path
    i++;
    path_string = strtok(NULL, ":");
}

Solution

  • 2 main things wrong with your code, pretty much summarized by the comments:

    • you strtok a public buffer (returned by getenv)
    • you don't know how many variables will be in the buffer so you don't allocate the array of arrays at all!

    Let me propose a working implementation not using strtok, and thus allowing to detect empty path (and replace it by . as Jonathan hinted). Compiles without any warnings using gcc -Wall -Wwrite-strings:

    #include <stdio.h>
    #include <string.h>
    #include <stdlib.h>
    
    int main()
    {
        const char **array;
        const char *orig_path_var = getenv("PATH");
        char *path_var = strdup(orig_path_var ? orig_path_var : ""); // just in case PATH is NULL, very unlikely
        const char *the_dot = ".";
        int j;
        int len=strlen(path_var);
        int nb_colons=0;
        char pathsep = ':';
        int current_colon = 0;
    
        // first count how many paths we have, and "split" almost like strtok would do
        for (j=0;j<len;j++)
        {
            if (path_var[j]==pathsep)
            {
                nb_colons++;
                path_var[j] = '\0';
            }       
        }
    
        // allocate the array of strings
        array=malloc((nb_colons+1) * sizeof(*array));
    
        array[0] = path_var;  // first path
    
        // rest of paths
        for (j=0;j<len;j++)
        {
            if (path_var[j]=='\0')
            {
                current_colon++;
                array[current_colon] = path_var+j+1;
                if (array[current_colon][0]=='\0')
                {
                    // special case: add dot if path is empty
                    array[current_colon] = the_dot;
                }
    
            }
        }
    
        for (j=0;j<nb_colons+1;j++)
        {
            printf("Path %d: <%s>\n",j,array[j]);
        }
    
        return(0);
    }
    

    Details of the operations:

    • make a copy of the env string to avoid butchering it
    • count the colons (to make it work with windows, just replace with ;) and tokenize
    • allocate the array according to number of colons + 1 (1 more token than number of separators!)
    • second pass to go through the string again and fill it with parts of the tokenized string (no need to allocate again, the original string is already allocated)
    • special case: empty path: replace by .. Could display a warning to tell the user that this is not safe.
    • print the result