Merged from master repo.

1 files changed, 18 insertions, 11 deletions

diff --git a/dupfiles b/dupfiles
index 3d9e8db..78e4b5f 100755
--- a/dupfiles
+++ b/dupfiles
@@ -46,30 +46,31 @@ class File(object):
 
     def read(self, num_bytes):
         if self.pathname in filepool:
-            f = filepool[self.pathname]
+            fd = filepool[self.pathname]
             filestats.hit_count += 1
         else:
             try:
-                f = file(self.pathname)
-            except IOError, e:
+                fd = os.open(self.pathname, os.O_RDONLY)
+            except OSError, e:
                 if e.errno != errno.EMFILE:
                     raise
                 victim = random.choice(filepool.keys())
-                filepool[victim].close()
+                os.close(filepool[victim])
                 del filepool[victim]
                 filestats.close_count += 1
-                f = file(self.pathname)
-            f.seek(self.offset)
-            filepool[self.pathname] = f
+                fd = os.open(self.pathname, os.O_RDONLY)
+            os.lseek(fd, self.offset, os.SEEK_SET)
+            filepool[self.pathname] = fd
             filestats.open_count += 1
-        data = f.read(num_bytes)
+        data = os.read(fd, num_bytes)
         self.offset += len(data)
         return data
 
     def close(self):
         if self.pathname in filepool:
-            filepool[self.pathname].close()
+            os.close(filepool[self.pathname])
             del filepool[self.pathname]
+            filestats.close_count += 1
 
 
 class DuplicateFileFinder(object):
@@ -179,7 +180,7 @@ class DuplicateFileFinder(object):
         
         done = []
         not_done = []
-        chunk_size = 64 * 1024
+        chunk_size = 4 * 1024
         for pathname, f in group:
             data = f.read(chunk_size)
             if not data:
@@ -223,6 +224,7 @@ def main():
     for dirname in sorted(args):
         dupfinder.collect(dirname)
 
+
     for duplicates in dupfinder.duplicates():
         if opts.make_hardlinks:
             make_hardlinks(duplicates)
@@ -233,4 +235,9 @@ def main():
 
 
 if __name__ == '__main__':
-    main()
+    profname = os.environ.get('DUPFILES_PROFILE')
+    if profname:
+        import cProfile
+        cProfile.run('main()', profname)
+    else:
+        main()