A little script to write an index of all files in the current directory

file-indexer.c

#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <unistd.h>
#include <fcntl.h>
#include <dirent.h>
#include <string.h>
#include <errno.h>
#include <stdlib.h>


void printdirent(struct dirent *entry, int parent_fd, char *parent_name);
void collect(char *dirname, int parent_fd, char *parent_name);


void printdirent(struct dirent *entry, int parent_fd, char *parent_name)
{
	// Get entryname
	char *name = entry->d_name;

	// Get some stats such as size and modification time
	struct stat statbuf;
	if (fstatat(parent_fd, name, &statbuf, AT_SYMLINK_NOFOLLOW) == -1) {
		fprintf(stderr, "lstat failed for %s: %s\n", name, strerror(errno));
		return;
	}

	off_t size = statbuf.st_size;
	struct timespec modiftime = statbuf.st_mtim;

	unsigned mode = statbuf.st_mode;
	char *type = S_ISREG(mode) ? "REG" : S_ISDIR(mode) ? "DIR" : S_ISLNK(mode) ? "LNK" : "UNK";

	// Print all stats
	printf("size=%ld\1mtime=%lu\1type=%s\1path=%s\1name=%s\n",
		size, modiftime.tv_sec, type, parent_name, name);
}


void collect(char *dirname, int parent_fd, char *parent_name)
{
	// skip Trash directory
	if (strcmp(dirname, ".Trash-1000") == 0) return;

	// Open FD for directory
	int dir_fd = openat(parent_fd, dirname, O_DIRECTORY | O_RDONLY);
	if (dir_fd == -1) {
		fprintf(stderr, "open failed for %s%s: %s\n", parent_name, dirname, strerror(errno));
		return;
	}

	// Get DIR struct for directory
	DIR *dir = fdopendir(dir_fd);
	if (dir == NULL) {
		fprintf(stderr, "opendir failed for %s%s: %s\n", parent_name, dirname, strerror(errno));
		return;
	}


	// Current directory string
	// TODO: support paths longer than 1024 chars?
	char *dirstr = malloc(1024 * sizeof(*dirstr));
	strncpy(dirstr, parent_name, 1024);
	strncat(dirstr, dirname, 1024);
	strncat(dirstr, "/", 1024);

	// Loop over all entries in directory
	struct dirent *stats;
	while ((stats = readdir(dir)) != NULL) {
		// Ignore . and .. but do track dotfiles
		if (strcmp(stats->d_name, ".") == 0 || strcmp(stats->d_name, "..") == 0) continue;

		// Print stats for current entry
		printdirent(stats, dir_fd, dirstr);

		// If directory: recurse
		if (stats->d_type == DT_DIR) {
			collect(stats->d_name, dir_fd, dirstr);
		}
	}


	// Clean up all resources
	free(dirstr);
	closedir(dir);
	close(dir_fd);
}




int main(int argc, char *argv[])
{
	// Get directory name from first argument or default to .
	char *dirname = ".";
	if (argc > 1) {
		dirname = argv[1];
	}

	collect(dirname, AT_FDCWD, "");
}

Suggestions left for the reader:

• use \x1F (unit separator) instead of \x01 (start of heading) as separator
• don't do strncat on the original pointer several times, but keep track where the previous one ended and continue there
• allow paths over 1024 characters long
• don't duplicate the full path for every file in the index but only print it at the directory line
• benchmark it against du -a --time
• RIIR 🦀 🚀