fdupes is a utility to find duplicate files in a directory hierarchy.

submitted by Antti Harri
"sounds like you should import it" pvalchev@

sysutils/fdupes/Makefile

@@ -0,0 +1,28 @@
+# $OpenBSD: Makefile,v 1.1.1.1 2008/06/20 06:03:43 ckuethe Exp $
+
+COMMENT=	identify or delete duplicates
+
+DISTNAME=	fdupes-1.40
+CATEGORIES=	sysutils
+
+MAINTAINER=	Antti Harri <iku@openbsd.fi>
+
+HOMEPAGE=	http://netdial.caribe.net/~adrian2/fdupes.html
+MASTER_SITES=	${HOMEPAGE:fdupes.html=programs/}
+
+PERMIT_PACKAGE_CDROM=	Yes
+PERMIT_PACKAGE_FTP=	Yes
+PERMIT_DISTFILES_CDROM=	Yes
+PERMIT_DISTFILES_FTP=	Yes
+
+WANTLIB=	c
+
+ALL_TARGET=
+
+do-install:
+	${INSTALL_PROGRAM} ${WRKSRC}/fdupes ${PREFIX}/bin/
+	${INSTALL_MAN} ${WRKSRC}/fdupes.1 ${PREFIX}/man/man1/
+	${INSTALL_DATA_DIR} ${PREFIX}/share/doc/fdupes
+	${INSTALL_DATA} ${WRKSRC}/README ${PREFIX}/share/doc/fdupes/
+
+.include <bsd.port.mk>

sysutils/fdupes/distinfo

@@ -0,0 +1,5 @@
+MD5 (fdupes-1.40.tar.gz) = Ed6atEZgibasu2KBazCxiQ==
+RMD160 (fdupes-1.40.tar.gz) = Cr0vDf7jjLMzwQaTY6ZfirN+GN0=
+SHA1 (fdupes-1.40.tar.gz) = 4bzpvfUNe/cA3aPrij0hixgbOTE=
+SHA256 (fdupes-1.40.tar.gz) = 4e2RxEv0oI3pqSywGcIEIIdP3u1JpW6PJwrX/Ca73ec=
+SIZE (fdupes-1.40.tar.gz) = 16026

sysutils/fdupes/patches/patch-Makefile

@@ -0,0 +1,53 @@
+$OpenBSD: patch-Makefile,v 1.1.1.1 2008/06/20 06:03:43 ckuethe Exp $
+--- Makefile.orig	Thu Mar 15 04:49:11 2001
++++ Makefile	Fri May 30 13:50:55 2008
+@@ -20,22 +20,45 @@ VERSION = "1.40"
+ # built in MD5 message digest routines) uncomment the following
+ # line (try this if you're having trouble with built in code).
+ #
+-#EXTERNAL_MD5 = -DEXTERNAL_MD5=\"md5sum\"
++#EXTERNAL_MD5 = 1
+ 
+ #
++# Uncomment to enable system built-in MD5 functions
++SYSTEM_MD5 = 1
++
++#
+ # This version of fdupes can use a red-black tree structure to
+ # store file information. This is disabled by default, as it
+ # hasn't been optimized or verified correct. If you wish to
+ # enable this untested option, uncomment the following line.
+ #
+-#EXPERIMENTAL_RBTREE = -DEXPERIMENTAL_RBTREE
++#EXPERIMENTAL_RBTREE = 1
+ 
+ #####################################################################
+ # no need to modify anything beyond this point                      #
+ #####################################################################
+ 
+-fdupes: fdupes.c md5/md5.c	
+-	gcc fdupes.c md5/md5.c -Wall -o fdupes -DVERSION=\"$(VERSION)\" $(EXTERNAL_MD5) $(EXPERIMENTAL_RBTREE)
++CFLAGS = -Wall -DVERSION=\"$(VERSION)\"
++LDFLAGS =
++SRC_FILES = fdupes.c
++
++.ifdef SYSTEM_MD5
++CFLAGS += -DHAVE_MD5
++#LDFLAGS += -lssl -lcrypto
++.else
++SRC_FILES += md5/md5.c
++.endif
++
++.ifdef EXTERNAL_MD5
++CFLAGS += -DEXTERNAL_MD5=\"md5sum\"
++.endif
++
++.ifdef EXPERIMENTAL_RBTREE
++CFLAGS += -DEXPERIMENTAL_RBTREE
++.endif
++
++fdupes: $(SRC_FILES)
++	gcc $(CFLAGS) $(SRC_FILES) $(LDFLAGS) -o fdupes
+ 
+ install: fdupes
+ 	cp fdupes $(INSTALLDIR)

sysutils/fdupes/patches/patch-fdupes_c

@@ -0,0 +1,178 @@
+$OpenBSD: patch-fdupes_c,v 1.1.1.1 2008/06/20 06:03:43 ckuethe Exp $
+--- fdupes.c.orig	Thu Mar 15 04:16:09 2001
++++ fdupes.c	Fri May 30 13:49:42 2008
+@@ -34,6 +34,11 @@
+ #include "md5/md5.h"
+ #endif
+ 
++#ifdef HAVE_MD5
++#include <sys/types.h>
++#include <md5.h>
++#endif
++
+ #define ISFLAG(a,b) ((a & b) == b)
+ #define SETFLAG(a,b) (a |= b)
+ 
+@@ -94,11 +99,13 @@ void escapefilename(char *escape_list, char **filename
+ {
+   int x;
+   int tx;
++  size_t l;
+   char *tmp;
+   char *filename;
+ 
+   filename = *filename_ptr;
+ 
++  l = strlen(filename) * 2 + 1;
+   tmp = (char*) malloc(strlen(filename) * 2 + 1);
+   if (tmp == NULL) {
+     errormsg("out of memory!\n");
+@@ -118,7 +125,7 @@ void escapefilename(char *escape_list, char **filename
+       errormsg("out of memory!\n");
+       exit(1);
+     }
+-    strcpy(*filename_ptr, tmp);
++    strlcpy(*filename_ptr, tmp, l);
+   }
+ }
+ 
+@@ -145,6 +152,7 @@ int grokdir(char *dir, file_t **filelistp)
+   struct dirent *dirinfo;
+   int lastchar;
+   int filecount = 0;
++  size_t l;
+   struct stat info;
+   struct stat linfo;
+   static int progress = 0;
+@@ -177,7 +185,8 @@ int grokdir(char *dir, file_t **filelistp)
+       newfile->duplicates = NULL;
+       newfile->hasdupes = 0;
+ 
+-      newfile->d_name = (char*)malloc(strlen(dir)+strlen(dirinfo->d_name)+2);
++      l = strlen(dir)+strlen(dirinfo->d_name)+2;
++      newfile->d_name = (char*)malloc(l);
+ 
+       if (!newfile->d_name) {
+ 	errormsg("out of memory!\n");
+@@ -186,11 +195,11 @@ int grokdir(char *dir, file_t **filelistp)
+ 	exit(1);
+       }
+ 
+-      strcpy(newfile->d_name, dir);
++      strlcpy(newfile->d_name, dir, l);
+       lastchar = strlen(dir) - 1;
+       if (lastchar >= 0 && dir[lastchar] != '/')
+-	strcat(newfile->d_name, "/");
+-      strcat(newfile->d_name, dirinfo->d_name);
++	strlcat(newfile->d_name, "/", l);
++      strlcat(newfile->d_name, dirinfo->d_name, l);
+       
+       if (filesize(newfile->d_name) == 0 && ISFLAG(flags, F_EXCLUDEEMPTY)) {
+ 	free(newfile->d_name);
+@@ -241,14 +250,22 @@ char *getcrcsignature(char *filename)
+   int x;
+   off_t fsize;
+   off_t toread;
++#ifdef HAVE_MD5
++  MD5_CTX state;
++#else
+   md5_state_t state;
++#endif
+   md5_byte_t digest[16];  
+   static md5_byte_t chunk[CHUNK_SIZE];
+   static char signature[16*2 + 1]; 
+   char *sigp;
+   FILE *file;
+-   
++
++#ifdef HAVE_MD5
++  MD5Init(&state);
++#else
+   md5_init(&state);
++#endif
+ 
+   fsize = filesize(filename);
+ 
+@@ -264,16 +281,24 @@ char *getcrcsignature(char *filename)
+       errormsg("error reading from file %s\n", filename);
+       return NULL;
+     }
++#ifdef HAVE_MD5
++    MD5Update(&state, chunk, toread);
++#else
+     md5_append(&state, chunk, toread);
++#endif
+     fsize -= toread;
+   }
+ 
++#ifdef HAVE_MD5
++  MD5Final(digest, &state);
++#else
+   md5_finish(&state, digest);
++#endif
+ 
+   sigp = signature;
+ 
+   for (x = 0; x < 16; x++) {
+-    sprintf(sigp, "%02x", digest[x]);
++    snprintf(sigp, 2, "%02x", digest[x]);
+     sigp = strchr(sigp, '\0');
+   }
+ 
+@@ -488,6 +513,7 @@ file_t *checkmatch(filetree_t **root, filetree_t *chec
+ {
+   int cmpresult;
+   char *crcsignature;
++  size_t l;
+   off_t fsize;
+ 
+   /* If inodes are equal one of the files is a hard link, which
+@@ -508,24 +534,26 @@ file_t *checkmatch(filetree_t **root, filetree_t *chec
+       crcsignature = getcrcsignature(checktree->file->d_name);
+       if (crcsignature == NULL) return NULL;
+ 
+-      checktree->file->crcsignature = (char*) malloc(strlen(crcsignature)+1);
++      l = strlen(crcsignature)+1;
++      checktree->file->crcsignature = (char*) malloc(l);
+       if (checktree->file->crcsignature == NULL) {
+ 	errormsg("out of memory\n");
+ 	exit(1);
+       }
+-      strcpy(checktree->file->crcsignature, crcsignature);
++      strlcpy(checktree->file->crcsignature, crcsignature, l);
+     }
+ 
+     if (file->crcsignature == NULL) {
+       crcsignature = getcrcsignature(file->d_name);
+       if (crcsignature == NULL) return NULL;
+ 
+-      file->crcsignature = (char*) malloc(strlen(crcsignature)+1);
++      l = strlen(crcsignature)+1;
++      file->crcsignature = (char*) malloc(l);
+       if (file->crcsignature == NULL) {
+ 	errormsg("out of memory\n");
+ 	exit(1);
+       }
+-      strcpy(file->crcsignature, crcsignature);
++      strlcpy(file->crcsignature, crcsignature, l);
+     }
+ 
+     cmpresult = strcmp(file->crcsignature, checktree->file->crcsignature);
+@@ -588,7 +616,7 @@ void printmatches(file_t *files)
+   while (files != NULL) {
+     if (files->hasdupes) {
+       if (!ISFLAG(flags, F_OMITFIRST)) {
+-	if (ISFLAG(flags, F_SHOWSIZE)) printf("%ld byte%seach:\n", files->size,
++	if (ISFLAG(flags, F_SHOWSIZE)) printf("%lld byte%seach:\n", files->size,
+ 	 (files->size != 1) ? "s " : " ");
+ 	if (ISFLAG(flags, F_DSAMELINE)) escapefilename("\\ ", &files->d_name);
+ 	printf("%s%c", files->d_name, ISFLAG(flags, F_DSAMELINE)?' ':'\n');
+@@ -676,7 +704,7 @@ void autodelete(file_t *files)
+       do {
+ 	printf("Set %d of %d, preserve files [1 - %d, all]", 
+           curgroup, groups, counter);
+-	if (ISFLAG(flags, F_SHOWSIZE)) printf(" (%ld byte%seach)", files->size,
++	if (ISFLAG(flags, F_SHOWSIZE)) printf(" (%lld byte%seach)", files->size,
+ 	  (files->size != 1) ? "s " : " ");
+ 	printf(": ");
+ 	fflush(stdout);

sysutils/fdupes/patches/patch-md5_md5_c

@@ -0,0 +1,19 @@
+$OpenBSD: patch-md5_md5_c,v 1.1.1.1 2008/06/20 06:03:43 ckuethe Exp $
+--- md5/md5.c.orig	Thu May 29 22:20:16 2008
++++ md5/md5.c	Thu May 29 22:20:27 2008
+@@ -38,6 +38,7 @@
+   1999-05-03 lpd Original version.
+  */
+ 
++#include <string.h>
+ #include "md5.h"
+ 
+ #ifdef TEST
+@@ -46,7 +47,6 @@
+  * The test program should print out the same values as given in section
+  * A.5 of RFC 1321, reproduced below.
+  */
+-#include <string.h>
+ main()
+ {
+     static const char *const test[7] = {

sysutils/fdupes/pkg/DESCR

@@ -0,0 +1,2 @@
+Fdupes is a program for identifying or deleting duplicate
+files residing within specified directories.

sysutils/fdupes/pkg/PLIST

@@ -0,0 +1,5 @@
+@comment $OpenBSD: PLIST,v 1.1.1.1 2008/06/20 06:03:43 ckuethe Exp $
+bin/fdupes
+@man man/man1/fdupes.1
+share/doc/fdupes/
+share/doc/fdupes/README