Refactor

2.py

@@ -9,26 +9,20 @@ parser = argparse.ArgumentParser(description='What the dupe!?')
 optional = parser._action_groups.pop() # Edited this line
 required = parser.add_argument_group('required arguments')
 
-optional.add_argument('--size', type=str,
-                    help='Only output files greater than \'size\'. 16, 16K, 16M, 16G, 16T')
+optional.add_argument('--threshold', type=str,
+                    help='Only output files greater than \'size\', e.g. 100M')
 
 required.add_argument('--dir', type=str, nargs='?', required=True, action='append',
-                    help='Directory to scan. Can be used multiple times.')
+                    help='Directory to scan. Can be issued multiple times.')
 
 parser._action_groups.append(optional)
 
 args = parser.parse_args()
 
-max_size1='10M'
-max_size2='50M'
-max_size3='100M'
-max_size4='1G'
-max_size5='5G'
-max_size6='5G'
-sizes = [max_size1, max_size2, max_size3, max_size4, max_size5, max_size6]
+sizes = ['10M', '50M', '100M', '1G', '5G', 'gt5GB']
 
-if args.size:
-    bytes = humanfriendly.parse_size(args.size)
+if args.threshold:
+    bytes = humanfriendly.parse_size(args.threshold)
 else:
     bytes = 0
 
@@ -74,7 +68,11 @@ def hashfile(path, blocksize = 65536):
 
 
 def printResults(dict1):
-    final = {max_size1:[], max_size2:[], max_size3:[], max_size4:[], max_size5:[], max_size6:[]}
+    final = {}
+    for size in sizes:
+        final[size] = []
+    del size
+    print(final)
     if bytes > 0:
         final[bytes] = []
     results = list(filter(lambda x: len(x) > 1, dict1.values()))
@@ -85,24 +83,19 @@ def printResults(dict1):
                 final[bytes].append(result)
 
         else:
-            if file_size >= humanfriendly.parse_size(max_size1) and file_size < humanfriendly.parse_size(max_size2):
-                final[max_size2].append(result)
-                #print('1M-50M: '+result)
-            elif file_size >= humanfriendly.parse_size(max_size2) and file_size < humanfriendly.parse_size(max_size3):
-                final[max_size3].append(result)
-                #print('50M-100M: '+result)
-            elif file_size >= humanfriendly.parse_size(max_size3) and file_size < humanfriendly.parse_size(max_size4):
-                #print('100M-1G: '+result)
-                final[max_size4].append(result)
-            elif file_size >= humanfriendly.parse_size(max_size4) and file_size < humanfriendly.parse_size(max_size5):
-                #print('1G-5G: '+result)
-                final[max_size5].append(result)
-            elif file_size >= humanfriendly.parse_size(max_size5):
-                #print('5G+: '+result)
-                final[max_size6].append(result)
+            #0=10MB 1=50MB 2=100MB 3=1GB 4=5GB
+            if file_size >= humanfriendly.parse_size(sizes[0]) and file_size < humanfriendly.parse_size(sizes[1]):
+                final[sizes[1]].append(result)
+            elif file_size >= humanfriendly.parse_size(sizes[1]) and file_size < humanfriendly.parse_size(sizes[2]):
+                final[sizes[2]].append(result)
+            elif file_size >= humanfriendly.parse_size(sizes[2]) and file_size < humanfriendly.parse_size(sizes[3]):
+                final[sizes[3]].append(result)
+            elif file_size >= humanfriendly.parse_size(sizes[3]) and file_size < humanfriendly.parse_size(sizes[4]):
+                final[sizes[4]].append(result)
+            elif file_size >= humanfriendly.parse_size(sizes[4]):
+                final[sizes[5]].append(result)
             else:
-                #print('<1M: '+result)
-                final[max_size1].append(result)
+                final[sizes[0]].append(result)
     if len(results) > 0:
         print('___________________')
         print('\n\033[1;34m\033[1;34m\u25b6 Duplicates Found\033[0m\n')
@@ -120,7 +113,10 @@ def printResults(dict1):
             for size in sizes:
                 new.append(size)
                 if len(final[size]) > 0:
-                    print("\n\033[1;34m\u25b6 Between %s and %s\033[0m" % (new[-2],size))
+                    if size == 'gt5GB':
+                        print("\n\033[1;34m\u25b6 >= %s\033[0m" % (new[-2]))
+                    else:
+                        print("\n\033[1;34m\u25b6 %s to %s\033[0m" % (new[-2],size))
                     for dupe in final[size]:
                         print('___________________\n')
                         for file in dupe: