X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=finddup.c;h=adf138ebc1530ff5e2e560a095d34c686f1d55dc;hb=834f5d0970d97555f204eea4897204db0511e86e;hp=cfef6be5c005a2f4203185c2b7cfe31e710ebdab;hpb=8f5d2ea79d687c9353e4ab803a8069c22275f1db;p=finddup.git

diff --git a/finddup.c b/finddup.c
index cfef6be..adf138e 100644
--- a/finddup.c
+++ b/finddup.c
@@ -23,12 +23,13 @@
  *
  */
 
-#define VERSION_NUMBER "0.5"
+#define VERSION_NUMBER "0.6"
 
 #define _BSD_SOURCE
 
 #include <sys/types.h>
 #include <sys/stat.h>
+#include <sys/param.h>
 #include <dirent.h>
 #include <stdlib.h>
 #include <stdio.h>
@@ -52,15 +53,18 @@ int ignore_dotfiles = 0; /* 1 means ignore files and directories
 int show_realpaths = 0; /* 1 means ignore files and directories
                             starting with a dot */
 
-int show_progress = 1; /* 1 means show a progress bar when we are in a
+int show_progress = 0; /* 1 means show a progress bar when we are in a
                           tty */
 
 int show_hits = 1; /* 1 means to show the files from dir2
                       corresponding to the ones from dir1 */
 
-int show_groups = 0; /* 1 means to show the group IDs when printing
+int show_groups = 1; /* 1 means to show the group IDs when printing
                         file names */
 
+int ignore_same_inode = 0; /* 1 means that comparison between two file
+                              with same inode will always be false */
+
 /********************************************************************/
 
 /* malloc with error checking.  */
@@ -105,7 +109,7 @@ struct file_with_size {
   size_t size;
   ino_t inode;
   struct file_with_size *next;
-  int id;
+  int group_id;
 };
 
 void file_list_delete(struct file_with_size *head) {
@@ -169,14 +173,14 @@ int same_content(struct file_with_size *f1, struct file_with_size *f2) {
 
     if(fd1 < 0) {
       fprintf(stderr,
-              "Can not open `%s' error: %s\n",
+              "Can not open \"%s\" error: %s\n",
               f1->filename,
               strerror(errno));
     }
 
     if(fd2 < 0) {
       fprintf(stderr,
-              "Can not open `%s' error: %s\n",
+              "Can not open \"%s\" error: %s\n",
               f2->filename,
               strerror(errno));
     }
@@ -185,6 +189,9 @@ int same_content(struct file_with_size *f1, struct file_with_size *f2) {
 }
 
 int same_files(struct file_with_size *f1, struct file_with_size *f2) {
+  if(ignore_same_inode && f1->inode == f2->inode) {
+    return 0;
+  }
   return f1->size == f2->size && same_content(f1, f2);
 }
 
@@ -196,7 +203,7 @@ struct file_with_size *scan_directory(struct file_with_size *tail,
   struct dirent *dir_e;
   struct stat sb;
   struct file_with_size *tmp;
-  char subname[PATH_MAX];
+  char subname[PATH_MAX + 1];
 
   if(lstat(name, &sb) != 0) {
     fprintf(stderr, "Can not stat \"%s\": %s\n", name, strerror(errno));
@@ -224,7 +231,7 @@ struct file_with_size *scan_directory(struct file_with_size *tail,
       tmp->filename = strdup(name);
       tmp->size = sb.st_size;
       tmp->inode = sb.st_ino;
-      tmp->id = -1;
+      tmp->group_id = -1;
       tail = tmp;
     }
   }
@@ -233,62 +240,133 @@ struct file_with_size *scan_directory(struct file_with_size *tail,
 }
 
 void print_file(struct file_with_size *node) {
+  char tmp[PATH_MAX + 1];
   if(show_realpaths) {
     if(show_groups) {
-      printf("%d %s\n", node->id, realpath(node->filename, 0));
+      realpath(node->filename, tmp);
+      printf("%d %s\n", node->group_id, tmp);
     } else {
-      printf("%s\n", realpath(node->filename, 0));
+      realpath(node->filename, tmp);
+      printf("%s\n", tmp);
     }
   } else {
     if(show_groups) {
-      printf("%d %s\n", node->id, node->filename);
+      printf("%d %s\n", node->group_id, node->filename);
     } else {
       printf("%s\n", node->filename);
     }
   }
 }
 
-void start(const char *dirname1, const char *dirname2) {
-  struct file_with_size *list1, *list2;
-  struct file_with_size *node1, *node2;
-  struct stat sb1, sb2;
-  int not_in, found, same_dir;
+int compare_nodes(const void *x1, const void *x2) {
+  const struct file_with_size **f1, **f2;
+
+  f1 = (const struct file_with_size **) x1;
+  f2 = (const struct file_with_size **) x2;
 
-  if(strncmp(dirname2, "not:", 4) == 0) {
-    not_in = 1;
-    dirname2 += 4;
+  if((*f1)->group_id < (*f2)->group_id) {
+    return -1;
+  } else if((*f1)->group_id > (*f2)->group_id) {
+    return 1;
   } else {
-    not_in = 0;
+    return 0;
   }
+}
 
-  if(lstat(dirname1, &sb1) != 0) {
-    fprintf(stderr, "Can not stat \"%s\": %s\n", dirname1, strerror(errno));
-    exit(EXIT_FAILURE);
+
+void print_result(struct file_with_size *list1, struct file_with_size *list2) {
+  struct file_with_size *node1, *node2;
+  struct file_with_size **nodes;
+  int nb, n;
+
+  nb = 0;
+  for(node1 = list1; node1; node1 = node1->next) {
+    if(node1->group_id >= 0) { nb++; }
   }
 
-  if(lstat(dirname2, &sb2) != 0) {
-    fprintf(stderr, "Can not stat \"%s\": %s\n", dirname2, strerror(errno));
-    exit(EXIT_FAILURE);
+  if(show_hits) {
+    for(node2 = list2; node2; node2 = node2->next) {
+      if(node2->group_id >= 0) { nb++; }
+    }
+  }
+
+  nodes = safe_malloc(nb * sizeof(struct file_with_size *));
+
+  n = 0;
+  for(node1 = list1; node1; node1 = node1->next) {
+    if(node1->group_id >= 0) {
+      nodes[n++] = node1;
+    }
+  }
+
+  if(show_hits) {
+    for(node2 = list2; node2; node2 = node2->next) {
+      if(node2->group_id >= 0) {
+        nodes[n++] = node2;
+      }
+    }
   }
 
-  same_dir = (sb1.st_ino == sb2.st_ino);
+  qsort(nodes, nb, sizeof(struct file_with_size *), compare_nodes);
+
+  for(n = 0; n < nb; n++) {
+    print_file(nodes[n]);
+  }
+
+  free(nodes);
+}
+
+void start(const char *dirname1, const char *dirname2) {
+  struct file_with_size *list1, *list2;
+  struct file_with_size *node1, *node2;
+  int not_in, found;
+  int k, p, pp, l1, n;
+
+  not_in = 0;
+
+  if(show_progress) {
+    fprintf(stderr, "Scanning %s ... ", dirname1);
+  }
 
-  fprintf(stderr, "Scanning %s ... ", dirname1);
   list1 = scan_directory(0, dirname1);
-  if(same_dir) {
-    list2 = list1;
-  } else {
-    fprintf(stderr, " %s ... ", dirname2);
+
+  if(dirname2) {
+    if(strncmp(dirname2, "not:", 4) == 0) {
+      not_in = 1;
+      dirname2 += 4;
+    }
+    if(show_progress) {
+      fprintf(stderr, "%s ... ", dirname2);
+    }
     list2 = scan_directory(0, dirname2);
+  } else {
+    list2 = list1;
   }
-  fprintf(stderr, "done.\n");
+
+  if(show_progress) {
+    fprintf(stderr, "done.\n");
+  }
+
+  k = 0;
+  pp = -1;
+  n = 0;
+  l1 = file_list_length(list1);
 
   if(not_in) {
     for(node1 = list1; node1; node1 = node1->next) {
+      if(show_progress) {
+        p = (100 * n)/l1;
+        if(p > pp) {
+          fprintf(stderr, "%d%%\n", p);
+          pp = p;
+        }
+        n++;
+      }
+
       found = 0;
 
       for(node2 = list2; !found && node2; node2 = node2->next) {
-        if(node1->inode != node2->inode && same_files(node1, node2)) {
+        if(same_files(node1, node2)) {
           found = 1;
         }
       }
@@ -303,33 +381,38 @@ void start(const char *dirname1, const char *dirname2) {
     }
 
   } else {
-    int k = 0;
-
     for(node1 = list1; node1; node1 = node1->next) {
+      if(show_progress) {
+        p = (100 * n)/l1;
+        if(p > pp) {
+          fprintf(stderr, "%d%%\n", p);
+          pp = p;
+        }
+        n++;
+      }
+
       for(node2 = list2; node2; node2 = node2->next) {
-        if(node1->inode != node2->inode && same_files(node1, node2)) {
-          if(node1->id < 0) {
-            if(node2->id >= 0) {
-              node1->id = node2->id;
+        if(same_files(node1, node2)) {
+          if(node1->group_id < 0) {
+            if(node2->group_id >= 0) {
+              node1->group_id = node2->group_id;
             } else {
-              node1->id = k;
+              node1->group_id = k;
               k++;
             }
-            print_file(node1);
           }
-          if(node2->id < 0) {
-            node2->id = node1->id;
-            if(show_hits) {
-              print_file(node2);
-            }
+          if(node2->group_id < 0) {
+            node2->group_id = node1->group_id;
           }
         }
       }
     }
   }
 
+  print_result(list1, list2);
+
   file_list_delete(list1);
-  if(!same_dir) {
+  if(dirname2) {
     file_list_delete(list2);
   }
 }
@@ -342,8 +425,10 @@ void print_help(FILE *out) {
   fprintf(out, "   -h   show this help\n");
   fprintf(out, "   -d   ignore dot files and directories\n");
   fprintf(out, "   -c   do not show which files in DIR2 corresponds to those in DIR1\n");
-  fprintf(out, "   -g   show the file groups\n");
+  fprintf(out, "   -g   do not show the file groups\n");
+  fprintf(out, "   -p   show progress\n");
   fprintf(out, "   -r   show the real file paths\n");
+  fprintf(out, "   -i   consider files with same inode as different\n");
   fprintf(out, "\n");
   fprintf(out, "Report bugs and comments to <francois@fleuret.org>\n");
 }
@@ -359,7 +444,7 @@ int main(int argc, char **argv) {
   setlocale (LC_ALL, "");
 
   while (1) {
-    c = getopt(argc, argv, "hrcgd");
+    c = getopt(argc, argv, "hrcgdp");
     if (c == -1)
       break;
 
@@ -379,8 +464,16 @@ int main(int argc, char **argv) {
       show_realpaths = 1;
       break;
 
+    case 'i':
+      ignore_same_inode = 1;
+      break;
+
     case 'g':
-      show_groups = 1;
+      show_groups = 0;
+      break;
+
+    case 'p':
+      show_progress = 1;
       break;
 
     case 'c':
@@ -392,10 +485,11 @@ int main(int argc, char **argv) {
     }
   }
 
-  if(optind + 1 < argc) {
+  if(optind + 2 == argc) {
     start(argv[optind], argv[optind + 1]);
-  } else if(optind < argc) {
-    start(argv[optind], argv[optind]);
+  } else if(optind + 1 == argc) {
+    ignore_same_inode = 1;
+    start(argv[optind], 0);
   } else {
     print_help(stderr);
     exit(EXIT_FAILURE);