Multiple improvements.

diff --git a/finddup.1 b/finddup.1
index f88f42f..54034b6 100644 (file)
--- a/finddup.1
+++ b/finddup.1
@@ -24,6 +24,15 @@ present in DIR1 and not in DIR2.
 \fB-h\fR
 display help and exit
 .TP
+\fB-d\fR
+ignore files and directories starting with a dot
+.TP
+\fB-c\fR
+do not show which files from DIR2 corresponds to files from DIR1
+.TP
+\fB-g\fR
+show the file group IDs (one group for each content)
+.TP
 \fB-r\fR
 shows the real path of the files
 

diff --git a/finddup.c b/finddup.c
index 8d6e67a..cfef6be 100644 (file)
--- a/finddup.c
+++ b/finddup.c
@@ -55,6 +55,12 @@ int show_realpaths = 0; /* 1 means ignore files and directories
 int show_progress = 1; /* 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
+                        file names */
+
 /********************************************************************/
 
 /* malloc with error checking.  */
@@ -99,7 +105,7 @@ struct file_with_size {
   size_t size;
   ino_t inode;
   struct file_with_size *next;
-  int error;
+  int id;
 };
 
 void file_list_delete(struct file_with_size *head) {
@@ -123,10 +129,6 @@ int file_list_length(struct file_with_size *head) {
 
 /**********************************************************************/
 
-int same_size(struct file_with_size *f1, struct file_with_size *f2) {
-  return f1->size == f2->size;
-}
-
 int same_content(struct file_with_size *f1, struct file_with_size *f2) {
   int fd1, fd2, s1, s2;
   char buffer1[BUFFER_SIZE], buffer2[BUFFER_SIZE];
@@ -164,14 +166,26 @@ int same_content(struct file_with_size *f1, struct file_with_size *f2) {
       }
     }
   } else {
-    if(fd1 >= 0) { close(fd1); }
-    if(fd2 >= 0) { close(fd2); }
-    return 0;
+
+    if(fd1 < 0) {
+      fprintf(stderr,
+              "Can not open `%s' error: %s\n",
+              f1->filename,
+              strerror(errno));
+    }
+
+    if(fd2 < 0) {
+      fprintf(stderr,
+              "Can not open `%s' error: %s\n",
+              f2->filename,
+              strerror(errno));
+    }
+    exit(EXIT_FAILURE);
   }
 }
 
 int same_files(struct file_with_size *f1, struct file_with_size *f2) {
-  return same_size(f1, f2) && same_content(f1, f2);
+  return f1->size == f2->size && same_content(f1, f2);
 }
 
 /**********************************************************************/
@@ -180,16 +194,16 @@ struct file_with_size *scan_directory(struct file_with_size *tail,
                                       const char *name) {
   DIR *dir;
   struct dirent *dir_e;
-  struct stat dummy;
+  struct stat sb;
   struct file_with_size *tmp;
   char subname[PATH_MAX];
 
-  if(lstat(name, &dummy) != 0) {
+  if(lstat(name, &sb) != 0) {
     fprintf(stderr, "Can not stat \"%s\": %s\n", name, strerror(errno));
     exit(EXIT_FAILURE);
   }
 
-  if(S_ISLNK(dummy.st_mode)) {
+  if(S_ISLNK(sb.st_mode)) {
     return tail;
   }
 
@@ -204,12 +218,13 @@ struct file_with_size *scan_directory(struct file_with_size *tail,
     }
     closedir(dir);
   } else {
-    if(S_ISREG(dummy.st_mode)) {
+    if(S_ISREG(sb.st_mode)) {
       tmp = safe_malloc(sizeof(struct file_with_size));
       tmp->next = tail;
       tmp->filename = strdup(name);
-      tmp->size = dummy.st_size;
-      tmp->inode = dummy.st_ino;
+      tmp->size = sb.st_size;
+      tmp->inode = sb.st_ino;
+      tmp->id = -1;
       tail = tmp;
     }
   }
@@ -217,10 +232,27 @@ struct file_with_size *scan_directory(struct file_with_size *tail,
   return tail;
 }
 
+void print_file(struct file_with_size *node) {
+  if(show_realpaths) {
+    if(show_groups) {
+      printf("%d %s\n", node->id, realpath(node->filename, 0));
+    } else {
+      printf("%s\n", realpath(node->filename, 0));
+    }
+  } else {
+    if(show_groups) {
+      printf("%d %s\n", node->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;
-  int not_in, found;
+  struct stat sb1, sb2;
+  int not_in, found, same_dir;
 
   if(strncmp(dirname2, "not:", 4) == 0) {
     not_in = 1;
@@ -229,8 +261,27 @@ void start(const char *dirname1, const char *dirname2) {
     not_in = 0;
   }
 
+  if(lstat(dirname1, &sb1) != 0) {
+    fprintf(stderr, "Can not stat \"%s\": %s\n", dirname1, strerror(errno));
+    exit(EXIT_FAILURE);
+  }
+
+  if(lstat(dirname2, &sb2) != 0) {
+    fprintf(stderr, "Can not stat \"%s\": %s\n", dirname2, strerror(errno));
+    exit(EXIT_FAILURE);
+  }
+
+  same_dir = (sb1.st_ino == sb2.st_ino);
+
+  fprintf(stderr, "Scanning %s ... ", dirname1);
   list1 = scan_directory(0, dirname1);
-  list2 = scan_directory(0, dirname2);
+  if(same_dir) {
+    list2 = list1;
+  } else {
+    fprintf(stderr, " %s ... ", dirname2);
+    list2 = scan_directory(0, dirname2);
+  }
+  fprintf(stderr, "done.\n");
 
   if(not_in) {
     for(node1 = list1; node1; node1 = node1->next) {
@@ -252,16 +303,25 @@ void start(const char *dirname1, const char *dirname2) {
     }
 
   } else {
+    int k = 0;
 
     for(node1 = list1; node1; node1 = node1->next) {
       for(node2 = list2; node2; node2 = node2->next) {
         if(node1->inode != node2->inode && same_files(node1, node2)) {
-          if(show_realpaths) {
-            printf("%s %s\n",
-                   realpath(node1->filename, 0),
-                   realpath(node2->filename, 0));
-          } else {
-            printf("%s %s\n", node1->filename, node2->filename);
+          if(node1->id < 0) {
+            if(node2->id >= 0) {
+              node1->id = node2->id;
+            } else {
+              node1->id = k;
+              k++;
+            }
+            print_file(node1);
+          }
+          if(node2->id < 0) {
+            node2->id = node1->id;
+            if(show_hits) {
+              print_file(node2);
+            }
           }
         }
       }
@@ -269,7 +329,9 @@ void start(const char *dirname1, const char *dirname2) {
   }
 
   file_list_delete(list1);
-  file_list_delete(list2);
+  if(!same_dir) {
+    file_list_delete(list2);
+  }
 }
 
 void print_help(FILE *out) {
@@ -277,8 +339,11 @@ void print_help(FILE *out) {
   fprintf(out, "Version %s (%s)\n", VERSION_NUMBER, UNAME);
   fprintf(out, "Without DIR2, lists duplicated files found in DIR1. With DIR2, lists files common to both directories. With the not: prefix, lists files found in DIR1 which do not exist in DIR2.\n");
   fprintf(out, "\n");
-  fprintf(out, "   -h   show this help.\n");
-  fprintf(out, "   -r   show the real file paths.\n");
+  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, "   -r   show the real file paths\n");
   fprintf(out, "\n");
   fprintf(out, "Report bugs and comments to <francois@fleuret.org>\n");
 }
@@ -294,7 +359,7 @@ int main(int argc, char **argv) {
   setlocale (LC_ALL, "");
 
   while (1) {
-    c = getopt(argc, argv, "hr");
+    c = getopt(argc, argv, "hrcgd");
     if (c == -1)
       break;
 
@@ -306,10 +371,22 @@ int main(int argc, char **argv) {
 
       break;
 
+    case 'd':
+      ignore_dotfiles = 1;
+      break;
+
     case 'r':
       show_realpaths = 1;
       break;
 
+    case 'g':
+      show_groups = 1;
+      break;
+
+    case 'c':
+      show_hits = 0;
+      break;
+
     default:
       exit(EXIT_FAILURE);
     }