X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=finddup.c;h=adf138ebc1530ff5e2e560a095d34c686f1d55dc;hb=834f5d0970d97555f204eea4897204db0511e86e;hp=983168b6feea2f0d17ff2e5720cf54421304c67b;hpb=ef337e5e5c15540fe4f4f367a9bd4599b4559634;p=finddup.git

diff --git a/finddup.c b/finddup.c
index 983168b..adf138e 100644
--- a/finddup.c
+++ b/finddup.c
@@ -29,6 +29,7 @@
 
 #include <sys/types.h>
 #include <sys/stat.h>
+#include <sys/param.h>
 #include <dirent.h>
 #include <stdlib.h>
 #include <stdio.h>
@@ -61,6 +62,9 @@ int show_hits = 1; /* 1 means to show the files from dir2
 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.  */
@@ -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));
@@ -233,7 +240,7 @@ struct file_with_size *scan_directory(struct file_with_size *tail,
 }
 
 void print_file(struct file_with_size *node) {
-  char tmp[PATH_MAX];
+  char tmp[PATH_MAX + 1];
   if(show_realpaths) {
     if(show_groups) {
       realpath(node->filename, tmp);
@@ -312,28 +319,10 @@ void print_result(struct file_with_size *list1, struct file_with_size *list2) {
 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 not_in, found;
   int k, p, pp, l1, n;
 
-  if(strncmp(dirname2, "not:", 4) == 0) {
-    not_in = 1;
-    dirname2 += 4;
-  } else {
-    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);
+  not_in = 0;
 
   if(show_progress) {
     fprintf(stderr, "Scanning %s ... ", dirname1);
@@ -341,13 +330,17 @@ void start(const char *dirname1, const char *dirname2) {
 
   list1 = scan_directory(0, dirname1);
 
-  if(same_dir) {
-    list2 = list1;
-  } else {
+  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;
   }
 
   if(show_progress) {
@@ -373,7 +366,7 @@ void start(const char *dirname1, const char *dirname2) {
       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;
         }
       }
@@ -399,7 +392,7 @@ void start(const char *dirname1, const char *dirname2) {
       }
 
       for(node2 = list2; node2; node2 = node2->next) {
-        if(node1->inode != node2->inode && same_files(node1, node2)) {
+        if(same_files(node1, node2)) {
           if(node1->group_id < 0) {
             if(node2->group_id >= 0) {
               node1->group_id = node2->group_id;
@@ -419,7 +412,7 @@ void start(const char *dirname1, const char *dirname2) {
   print_result(list1, list2);
 
   file_list_delete(list1);
-  if(!same_dir) {
+  if(dirname2) {
     file_list_delete(list2);
   }
 }
@@ -435,6 +428,7 @@ void print_help(FILE *out) {
   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");
 }
@@ -470,6 +464,10 @@ int main(int argc, char **argv) {
       show_realpaths = 1;
       break;
 
+    case 'i':
+      ignore_same_inode = 1;
+      break;
+
     case 'g':
       show_groups = 0;
       break;
@@ -487,10 +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);