patch to add `minimum file size` option to duff

add_minimum-file-size_option.diff

From d94859fa294d9974563b48195b7430a36c15d651 Mon Sep 17 00:00:00 2001
From: OrangeRaven <apfelsinenhain@gmx.de>
Date: Thu, 19 Dec 2013 01:45:16 +0100
Subject: [PATCH] Add option option for minimum file size (  `-m _size_` ) ...
 don't check files smaller than this.

---
 src/duff.c       | 16 ++++++++++++++--
 src/duffdriver.c |  6 +++++-
 2 files changed, 19 insertions(+), 3 deletions(-)

diff --git a/src/duff.c b/src/duff.c
index ac1a22d..8b28573 100644
--- a/src/duff.c
+++ b/src/duff.c
@@ -139,6 +139,10 @@ int header_uses_digest = 0;
  */
 off_t sample_limit = 0;
 
+/* Specifies the minimal size of files to be compared with the sampling method.
+ */
+off_t min_file_size = 0;
+
 /* These functions are documented below, where they are defined.
  */
 static void version(void);
@@ -161,7 +165,7 @@ static void version(void)
  */
 static void usage(void)
 {
-  printf(_("Usage: %s [-0DHLPaepqrtuz] [-d function] [-f format] [-l size] [file ...]\n"),
+  printf(_("Usage: %s [-0DHLPaepqrtuz] [-d function] [-f format] [-l size] [-m size] [file ...]\n"),
            PACKAGE_NAME);
 
   printf("       %s -h\n", PACKAGE_NAME);
@@ -179,6 +183,7 @@ static void usage(void)
   printf(_("  -f  format for cluster headers\n"));
   printf(_("  -h  show this help\n"));
   printf(_("  -l  the minimum size that activates sampling\n"));
+  printf(_("  -m  minimum size; files smaller than this will be ignored\n"));
   printf(_("  -q  quiet; suppress warnings and error messages\n"));
   printf(_("  -p  physical files; do not report multiple hard links as duplicates\n"));
   printf(_("  -r  search recursively through specified directories\n"));
@@ -208,7 +213,7 @@ int main(int argc, char** argv)
   bindtextdomain(PACKAGE, LOCALEDIR);
   textdomain(PACKAGE);
 
-  while ((ch = getopt(argc, argv, "0DHLPad:ef:hl:pqrtuvz")) != -1)
+  while ((ch = getopt(argc, argv, "0DHLPad:ef:hl:m:pqrtuvz")) != -1)
   {
     switch (ch)
     {
@@ -251,6 +256,13 @@ int main(int argc, char** argv)
         else
           sample_limit = limit;
         break;
+      case 'm':
+        limit = (off_t) strtoull(optarg, &temp, 10);
+        if (temp == optarg || errno == ERANGE || errno == EINVAL)
+          warning(_("Ignoring invalid minimum file size %s"), optarg);
+        else
+          min_file_size = limit;
+        break;
       case 'p':
         physical_flag = 1;
         break;
diff --git a/src/duffdriver.c b/src/duffdriver.c
index a300858..39cd9da 100644
--- a/src/duffdriver.c
+++ b/src/duffdriver.c
@@ -105,6 +105,7 @@ extern int physical_flag;
 extern int excess_flag;
 extern const char* header_format;
 extern int header_uses_digest;
+extern off_t min_file_size;
 
 /* Represents a single physical directory.
  */
@@ -333,7 +334,10 @@ static void process_file(const char* path, struct stat* sb)
     if (ignore_empty_flag)
       return;
   }
-
+  
+  if (sb->st_size < min_file_size)
+    return;
+  
   /* NOTE: Check for duplicate arguments? */
 
   if (physical_flag)
-- 
1.7.11.1