X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?p=finddup.git;a=blobdiff_plain;f=finddup.c;h=1ea54cec256ed09f7b8d0d4dd9143a82756f145c;hp=aa16dac99582b2c1365c06b97f9c9552044df26b;hb=0d200c3b81ec3672daea4cffd15551e4e3788d9d;hpb=0c962f1c0fabf85606069ab40c7b943a28c61abd

diff --git a/finddup.c b/finddup.c
index aa16dac..1ea54ce 100644
--- a/finddup.c
+++ b/finddup.c
@@ -23,7 +23,7 @@
  *
  */
 
-#define VERSION_NUMBER "0.8"
+#define VERSION_NUMBER "0.9"
 
 #define _BSD_SOURCE
 
@@ -40,6 +40,9 @@
 #include <locale.h>
 #include <getopt.h>
 #include <fcntl.h>
+#ifdef WITH_MD5
+#include <openssl/md5.h>
+#endif
 
 /* 1M really helps compared to 64k */
 #define READ_BUFFER_SIZE (1024 * 1024)
@@ -65,12 +68,17 @@ 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 */
+int same_inodes_are_different = 0; /* 1 means that comparison between
+                                      two file with same inode will
+                                      always be false */
 
 int tty_width = -1; /* Positive value means what width to use to show
                        the progress bar */
 
+#ifdef WITH_MD5
+int use_md5 = 0; /* 1 means we keep an MD5 signature for each file */
+#endif
+
 /********************************************************************/
 
 /* malloc with error checking.  */
@@ -102,6 +110,10 @@ struct file_node {
   ino_t inode;
   int group_id; /* one per identical file content */
   int dir_id; /* 1 for DIR1, and 2 for DIR2 */
+#ifdef WITH_MD5
+  int md5_computed;
+  unsigned char md5[MD5_DIGEST_LENGTH];
+#endif
 };
 
 void file_list_delete(struct file_node *head) {
@@ -129,6 +141,25 @@ int same_content(struct file_node *f1, struct file_node *f2,
                  char *buffer1, char *buffer2) {
   int fd1, fd2, s1, s2;
 
+#ifdef WITH_MD5
+  MD5_CTX c1, c2;
+
+  if(use_md5) {
+    if(f1->md5_computed && f2->md5_computed) {
+      if(!memcmp(f1->md5, f2->md5, MD5_DIGEST_LENGTH)) {
+        return 0;
+      }
+    } else {
+      if(!f1->md5_computed) {
+        MD5_Init(&c1);
+      }
+      if(!f2->md5_computed) {
+        MD5_Init(&c2);
+      }
+    }
+  }
+#endif
+
   fd1 = open(f1->name, O_RDONLY);
   fd2 = open(f2->name, O_RDONLY);
 
@@ -147,6 +178,18 @@ int same_content(struct file_node *f1, struct file_node *f2,
         if(s1 == 0) {
           close(fd1);
           close(fd2);
+#ifdef WITH_MD5
+          if(use_md5) {
+            if(!f1->md5_computed) {
+              MD5_Final(f1->md5, &c1);
+              f1->md5_computed = 1;
+            }
+            if(!f2->md5_computed) {
+              MD5_Final(f2->md5, &c2);
+              f2->md5_computed = 1;
+            }
+          }
+#endif
           return 1;
         } else {
           if(memcmp(buffer1, buffer2, s1)) {
@@ -154,6 +197,16 @@ int same_content(struct file_node *f1, struct file_node *f2,
             close(fd2);
             return 0;
           }
+#ifdef WITH_MD5
+          if(use_md5) {
+            if(!f1->md5_computed) {
+              MD5_Update(&c1, buffer1, s1);
+            }
+            if(!f2->md5_computed) {
+              MD5_Update(&c2, buffer2, s2);
+            }
+          }
+#endif
         }
       } else {
         fprintf(stderr,
@@ -183,9 +236,10 @@ int same_content(struct file_node *f1, struct file_node *f2,
 
 int same_files(struct file_node *f1, struct file_node *f2,
                char *buffer1, char *buffer2) {
-  if(ignore_same_inode && f1->inode == f2->inode) {
+  if(same_inodes_are_different && f1->inode == f2->inode) {
     return 0;
   }
+
   return f1->size == f2->size && same_content(f1, f2, buffer1, buffer2);
 }
 
@@ -228,6 +282,9 @@ struct file_node *scan_directory(struct file_node *tail, const char *name) {
           tmp->inode = sb.st_ino;
           tmp->group_id = -1;
           tmp->dir_id = -1;
+#ifdef WITH_MD5
+          tmp->md5_computed = 0;
+#endif
           tail = tmp;
         }
       }
@@ -479,6 +536,9 @@ void print_help(FILE *out) {
   fprintf(out, "   -r, --real-paths           show the real file paths\n");
   fprintf(out, "   -i, --same-inodes-are-different\n");
   fprintf(out, "                              consider files with same inode as different\n");
+#ifdef WITH_MD5
+  fprintf(out, "   -m, --md5                  use MD5 hashing\n");
+#endif
   fprintf(out, "\n");
   fprintf(out, "Report bugs and comments to <francois@fleuret.org>.\n");
 }
@@ -494,6 +554,7 @@ static struct option long_options[] = {
   { "ignore-dots", no_argument, 0, 'd' },
   { "ignore-empty", no_argument, 0, '0' },
   { "show-progress", no_argument, 0, 'p' },
+  { "md5", no_argument, 0, 'm' },
   { 0, 0, 0, 0 }
 };
 
@@ -502,7 +563,7 @@ int main(int argc, char **argv) {
 
   setlocale (LC_ALL, "");
 
-  while ((c = getopt_long(argc, argv, "hircgd0p",
+  while ((c = getopt_long(argc, argv, "hircgd0pm",
                           long_options, NULL)) != -1) {
     switch (c) {
 
@@ -525,7 +586,7 @@ int main(int argc, char **argv) {
       break;
 
     case 'i':
-      ignore_same_inode = 1;
+      same_inodes_are_different = 1;
       break;
 
     case 'g':
@@ -540,6 +601,16 @@ int main(int argc, char **argv) {
       show_hits = 0;
       break;
 
+    case 'm':
+#ifdef WITH_MD5
+      use_md5 = 1;
+#else
+      fprintf(stderr,
+              "finddup has not be compiled with MD5 hashing.\n");
+      exit(EXIT_FAILURE);
+#endif
+      break;
+
     default:
       exit(EXIT_FAILURE);
     }
@@ -548,7 +619,7 @@ int main(int argc, char **argv) {
   if(optind + 2 == argc) {
     start(argv[optind], argv[optind + 1]);
   } else if(optind + 1 == argc) {
-    ignore_same_inode = 1;
+    same_inodes_are_different = 1;
     start(argv[optind], 0);
   } else {
     print_help(stderr);