EpGbgZ6Y

1#!/usr/bin/env python
2# -*- coding:utf-8 -*-
3# Script generates statistics table for specified direcories sorted by type(by
4# default), overall files size or files count. This script can also sort files
5# into the directories by type using hardlinks(default) or softlinks. Type
6# determination is done by `file` utility.
7# 
8# To get help just run script with wrong parameters.
9
10import os
11import sys
12import subprocess
13import string
14import getopt
15import operator
16
17def usage():
18    print "Usage: fmetric.py [-C|-S] [-h] [-m] [-b [-s]] [-q] DIR [DIR] ...\n\
19    \t-C\tSort by files count\n\
20    \t-S\tSort by overall size\n\
21    \t-h\tShow sizes in human-readable format\n\
22    \t-m\tUse MIME types instead of file descriptions\n\
23    \t-b\tCreate backup directory\n\
24    \t-s\tUse softlinks (only useful with -b option)\n\
25    \t-q\tUse \"quiet\" mode (do not write any error messages, skip on errors)"
26
27
28def file_info(filename, flags):
29    file_binary = "/usr/bin/file"
30    (file_stdout, file_stderr) = subprocess.Popen( \
31           [file_binary, flags, filename], \
32           stdout=subprocess.PIPE,\
33           stderr=subprocess.PIPE).communicate()
34    return file_stdout.split(",")[0].strip('\n')
35
36
37def sizeof_fmt(num, readable):
38    if not readable:
39        return num
40
41    for x in ['bytes','KB','MB','GB','TB']:
42        if num < 1024.0:
43            return "%3.1f %s" % (num, x)
44        num /= 1024.0
45
46
47def main(argv):
48    # Initializing basic variables
49    backup = False
50    readable = False
51    use_symlinks = False
52    quiet_mode = False
53    info_flags = "-b"
54    sort_key = 0
55    info_list = []
56    size_list = []
57    count_list = []
58
59    # Parsing command-line arguments
60    try:
61        opts, args = getopt.getopt(argv, "CShmbsq")
62    except getopt.GetoptError:
63        usage()
64        sys.exit(2)
65
66    for opt, arg in opts:
67        if opt == "-h":
68            readable = True
69
70        if (opt == "-S"): 
71            if (sort_key > 0):
72                usage()
73                sys.exit(2)
74
75            sort_key = 1
76
77        if (opt == "-C"):
78            if (sort_key > 0):
79                usage()
80                sys.exit(2)
81                
82            sort_key = 2
83
84        if (opt == "-b"):
85            backup = True
86
87        if (opt == "-m"):
88            info_flags = "-bi"
89
90        if (opt == "-s"):
91            use_symlinks = True
92
93        if (opt == "-q"):
94            quiet_mode = True
95
96    # Walk through the files and gather stats
97    for dir in args:
98        backup_dir = os.path.join(dir, "backup")
99        if backup and os.path.exists(backup_dir):
100            print "Backup directory \"%s\" exists, exiting" % backup_dir
101            sys.exit(1)
102
103        for root, dirs, files in os.walk(dir):
104            for name in files:
105                filename = os.path.join(root, name)
106
107                # This might cause some errors for moved files, broken links,
108                # etc. Output all errors to stderr, do not stop the script
109                try:
110                    size = os.path.getsize(filename)
111                except OSError, (errno, strerror):
112                    if not quiet_mode:
113                        sys.stderr.write( "Error reading file size for %s: %s\n" % \
114                                (filename, strerror) )
115                    continue
116
117                info = file_info(filename, info_flags)
118
119                # Add stats to the comparison table
120                if info in info_list:
121                    index = info_list.index(info)
122                    size_list[index] += size
123                    count_list[index] += 1
124                else:
125                    info_list.append(info)
126                    size_list.append(size)
127                    count_list.append(1)
128
129                # Create backup hardlinks/symlinks
130                if backup:
131                    backup_path = os.path.join(backup_dir, info)
132                    link_path = os.path.join(backup_path, name)
133                    if not os.path.exists(backup_path):
134                        os.makedirs(backup_path)
135
136                    try:
137                        if  use_symlinks:
138                            os.symlink(filename, link_path)
139                        else:
140                            os.link(filename, link_path)
141
142                    except OSError, (errno, strerror):
143                        if not quiet_mode:
144                            sys.stderr.write( "Error creating link %s: %s\n" % \
145                                    (link_path, strerror) )
146
147
148    info_field_size = len(max(info_list, key=len))
149    
150    # Print header
151    print "%s | %s | %s" % ( string.ljust("Type",info_field_size), \
152            string.ljust("Size", 12), \
153            "Count")
154    print string.ljust("", info_field_size + 3 + 12 + 3 + 12, "=")
155    
156
157    # Print the result
158    for info, size, count in sorted(zip(info_list, size_list, count_list), \
159            key = operator.itemgetter(sort_key), reverse = (sort_key <> 0)):
160        print "%s | %s | %d" % (string.ljust(info, info_field_size), \
161                string.ljust(str(sizeof_fmt(size, readable)), 12), \
162                count )
163                
164
165if __name__ == '__main__':
166    if len(sys.argv) < 2:
167        usage()
168        sys.exit(2)
169
170    main(sys.argv[1:])