examples/find_identical_files.c

   1 #include "../include/filelist.h"
   2 #include "../include/filelib.h"
   3 #include "../include/logger.h"
   4 #include <string.h>
   5 #include <stdio.h>
   6
   7 static int stringptr_endswithchar(stringptr* s, int ch) {
   8         if(stringptr_getsize(s) == 0) return 0;
   9         return stringptr_get(s)[stringptr_getsize(s)-1] == ch;
  10 }
  11
  12 static int compare_files(stringptr* fn1, stringptr* fn2) {
  13         FILE *f1 = fopen(stringptr_get(fn1), "r");
  14         FILE *f2 = fopen(stringptr_get(fn2), "r");
  15         int ret = 1;
  16         size_t n;
  17         while(1) {
  18                 unsigned char buf1[16*1024];
  19                 unsigned char buf2[16*1024];
  20                 n = fread(buf1, 1, sizeof buf1, f1);
  21                 if(n == 0) { ret = 0; break; }
  22                 if(n != fread(buf2, 1, sizeof buf2, f2)) break;
  23                 if(memcmp(buf1, buf2, n)) break;
  24         }
  25         fclose(f1);
  26         fclose(f2);
  27         return ret;
  28 }
  29
  30 static int mode = 0; /* 0: identical, 1: not identical */
  31
  32 static int process_path(stringptr* dir1, stringptr* dir2) {
  33         filelist fl1, fl2;
  34         int ret;
  35         ret = filelist_search(&fl1, dir1, SPL("*"), FLF_EXCLUDE_PATH|FLF_INCLUDE_HIDDEN);
  36         if(ret) return ret;
  37         ret = filelist_search(&fl2, dir2, SPL("*"), FLF_EXCLUDE_PATH|FLF_INCLUDE_HIDDEN);
  38         if(ret) return ret;
  39         size_t i, j;
  40         for(i=0; i < stringptrlist_getsize(fl1.files); ++i) {
  41                 stringptr *f1 = stringptrlist_get(fl1.files, i);
  42                 if(!stringptrlist_contains(fl2.files, f1)) continue;
  43                 stringptr* full1 = stringptr_concat(dir1, stringptr_endswithchar(dir1, '/') ? SPL("") : SPL("/"),  f1, NULL);
  44                 stringptr* full2 = stringptr_concat(dir2, stringptr_endswithchar(dir1, '/') ? SPL("") : SPL("/"),  f1, NULL);
  45                 if(stringptr_endswithchar(f1, '/')) {
  46                         ret = process_path(full1, full2);
  47                         if(ret) return ret;
  48                         goto next;
  49                 }
  50                 if(getfilesize(stringptr_get(full1)) != getfilesize(stringptr_get(full2)))
  51                         if(mode) goto print;
  52                         else goto next;
  53                 if(compare_files(full1, full2) == mode) {
  54                 print:
  55                         log_puts(1, full1);
  56                         log_putln(1);
  57                 }
  58         next:
  59                 stringptr_free(full1);
  60                 stringptr_free(full2);
  61         }
  62         filelist_free(&fl1);
  63         filelist_free(&fl2);
  64 }
  65
  66 static int usage() {
  67         log_puts(2, SPL(
  68                 "usage: PROG MODE DIR1 DIR2\n"
  69                 "MODE: i - identical or n - not identical\n"
  70                 "finds all (not) identical files in DIR1/2 and its subdirs.\n"
  71                 "prints filenames of DIR1 that match MODE.\n"
  72                 "files that exist only in DIR1 or DIR2 are excluded.\n"
  73         ));
  74         return 1;
  75 }
  76
  77 int main(int argc, char** argv) {
  78         if(argc != 4) return usage();
  79         mode = argv[1][0] == 'i' ? 0 : 1;
  80         SPDECLAREC(dir1, argv[2]);
  81         SPDECLAREC(dir2, argv[3]);
  82         int ret = process_path(dir1, dir2);
  83         return ret;
  84 }