Getting full file path during recursive file walk C - c

Question

I am doing a recursive walk through directories to make changes to files. My change file function needs the full path of the file to be able to do stuff. However, what my program is doing right now is just getting the name of the current file or folder but not the full path.
My approach is that I would make a string and keeps appending names to it until I get the full path. However, because I'm doing recursion, I'm having troubles passing the string around to append more strings to it.
This is my code:
#include <stdio.h>
#include <stdlib.h>
#include <regex.h>
#include <string.h>
#include <dirent.h>
#include <unistd.h>
#include <sys/types.h>
#include <errno.h>
void recursiveWalk(const char *pathName, char *fullPath, int level) {
DIR *dir;
struct dirent *entry;
if (!(dir = opendir(pathName))) {
fprintf(stderr, "Could not open directory\n");
return;
}
if (!(entry = readdir(dir))) {
fprintf(stderr, "Could not read directory\n");
return;
}
do {
if (entry->d_type == DT_DIR) { // found subdirectory
char path[1024];
int len = snprintf(path, sizeof(path)-1, "%s/%s", pathName, entry->d_name); // get depth
path[len] = 0;
// skip hidden paths
if (strcmp(entry->d_name, ".") == 0 || strcmp(entry->d_name, "..") == 0) {
continue;
}
fprintf(stdout, "%*s[%s]\n", level*2, "", entry->d_name);
// Append fullPath to entry->d_name here
recursiveWalk(path, fullPath, level + 1);
}
else { // files
fprintf(stdout, "%*s- %s\n", level*2, "", entry->d_name);
//changeFile(fullPath);
}
} while (entry = readdir(dir));
closedir(dir);
}
int main(int argn, char *argv[]) {
int level = 0;
recursiveWalk(".", "", level);
return 0;
}

Answer 1

Well there are a number of little problems in your code.
you never use nor change fullPath in recursiveWalk
your formats are weird : you use level*2 to limit the number of characters printed from an empty string
you compute the actual path only when you have found a directory, while you say you need it to change a file.
you add path[len] = 0 after a snprintf when snprintf guarantees that but buffer is null terminated
But apart from that, you correctly pass the path to the analyzed dir append to the path passed in initial call, but in pathName variable, and computed as path.
So a possible fix for your code would be :
fix the formats for printf
remove the unused fullPath parameter from recursiveWalk
allways compute path and use it in the file branch
comment out the unnecessary path[len] = '\0'
I also replaced while (entry = readdir(dir)); with while ((entry = readdir(dir))); to explicitely tell the compiler that I want to set entry and then test its value - and remove the warning
Possible code:
#include <stdio.h>
#include <stdlib.h>
#include <regex.h>
#include <string.h>
#include <dirent.h>
#include <unistd.h>
#include <sys/types.h>
#include <errno.h>
void recursiveWalk(const char *pathName, int level) {
DIR *dir;
struct dirent *entry;
if (!(dir = opendir(pathName))) {
fprintf(stderr, "Could not open directory\n");
return;
}
if (!(entry = readdir(dir))) {
fprintf(stderr, "Could not read directory\n");
return;
}
do {
char path[1024];
int len = snprintf(path, sizeof(path)-1, "%s/%s", pathName, entry->d_name); // get depth
// path[len] = 0;
if (entry->d_type == DT_DIR) { // found subdirectory
// skip hidden paths
if (strcmp(entry->d_name, ".") == 0 || strcmp(entry->d_name, "..") == 0) {
continue;
}
fprintf(stdout, "%s [%s] (%d)\n", pathName, entry->d_name, level);
// Append fullPath to entry->d_name here
recursiveWalk(path, level + 1);
}
else { // files
fprintf(stdout, "%s (%d)\n", path, level);
//changeFile(fullPath);
}
} while ((entry = readdir(dir)));
closedir(dir);
}
int main(int argn, char *argv[]) {
int level = 0;
recursiveWalk(".", level);
return 0;
}

Answer 2

Recursion is a succinct way to express things (especially walking directories), but actually you should normally avoid it in practice. If the directory tree is deep enough, it will crash your software.
Using a queue eliminates the need for recursion, and is generally an efficient way to traverse.
I'm including the code I use to process directory trees in a project...
static int on_dir(const char* const dir, struct duplicate** dp) {
bool r = opts.recurse;
DIR* d = opendir(dir);
if (!d)
return - 1;
struct dirent* de;
while ((de = readdir(d))) {
struct stat s;
size_t bs = strlen(dir) + strlen(de->d_name) + 2;
char b[bs];
const char* const a = strjoin(b, dir, de->d_name, '/');
if (lstat(a, &s)) {
print_error("unable to stat %s", d);
continue;
}
if (S_ISREG(s.st_mode))
if (on_file(a, &s, dp))
print_error("unable to process file %s/%s", dir, de->d_name);
}
if (!r) {
if (closedir(d))
on_fatal("unable to close directory %s", dir);
return 0;
}
rewinddir(d);
while ((de = readdir(d))) {
struct stat ds;
size_t bs = strlen(dir) + strlen(de->d_name) + 2;
char b[bs];
const char* const d = strjoin(b, dir, de->d_name, '/');
if (lstat(d, &ds)) {
print_error("unable to stat %s", d);
continue;
}
if (S_ISDIR(ds.st_mode)) {
const char* const dot = ".";
const char* const dotdot = "..";
if (!strcmp(dot, de->d_name) || !strcmp(dotdot, de->d_name))
continue;
struct path* p = path_create(strcpy(fmalloc(bs), d));
queue_add(&paths, &p->queue);
}
}
if (closedir(d))
print_error("unable to close directory %s", dir);
return 0;
}
and the code for strjoin
static inline char* strjoin(char* restrict const d, const char* restrict const a, const char* restrict const b, const char c) {
size_t na = strlen(a);
size_t nb = strlen(b);
memcpy(d, a, na);
d[na] = c;
memcpy(d + na + 1, b, nb);
d[na + nb + 1] = '\0';
return d;
}
I'm hoping this helps. Please feel free to use any of the code you find in the git repository.