blob: 736a441761fd8548aea14eed82efe59969c8e02e [file] [log] [blame]
Georg Brandld3f467a2010-08-02 22:53:22 +00001#! /usr/bin/env python3
Guido van Rossum6a8f7b72003-01-01 14:41:25 +00002
3"""Show file statistics by extension."""
4
5import os
6import sys
7
Florent Xiclunae4a33802010-08-09 12:24:20 +00008
Guido van Rossum6a8f7b72003-01-01 14:41:25 +00009class Stats:
10
11 def __init__(self):
12 self.stats = {}
13
14 def statargs(self, args):
15 for arg in args:
16 if os.path.isdir(arg):
17 self.statdir(arg)
18 elif os.path.isfile(arg):
19 self.statfile(arg)
20 else:
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000021 sys.stderr.write("Can't find %s\n" % arg)
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000022 self.addstats("<???>", "unknown", 1)
23
24 def statdir(self, dir):
25 self.addstats("<dir>", "dirs", 1)
26 try:
27 names = os.listdir(dir)
Andrew Svetlov8b33dd82012-12-24 19:58:48 +020028 except OSError as err:
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000029 sys.stderr.write("Can't list %s: %s\n" % (dir, err))
30 self.addstats("<dir>", "unlistable", 1)
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000031 return
Florent Xiclunae4a33802010-08-09 12:24:20 +000032 for name in sorted(names):
Guido van Rossum21123ab2003-04-09 19:10:46 +000033 if name.startswith(".#"):
Florent Xiclunae4a33802010-08-09 12:24:20 +000034 continue # Skip CVS temp files
Guido van Rossum21123ab2003-04-09 19:10:46 +000035 if name.endswith("~"):
Florent Xiclunae4a33802010-08-09 12:24:20 +000036 continue # Skip Emacs backup files
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000037 full = os.path.join(dir, name)
38 if os.path.islink(full):
39 self.addstats("<lnk>", "links", 1)
40 elif os.path.isdir(full):
41 self.statdir(full)
42 else:
43 self.statfile(full)
44
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000045 def statfile(self, filename):
46 head, ext = os.path.splitext(filename)
47 head, base = os.path.split(filename)
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000048 if ext == base:
Florent Xiclunae4a33802010-08-09 12:24:20 +000049 ext = "" # E.g. .cvsignore is deemed not to have an extension
Guido van Rossum21123ab2003-04-09 19:10:46 +000050 ext = os.path.normcase(ext)
51 if not ext:
52 ext = "<none>"
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000053 self.addstats(ext, "files", 1)
54 try:
Florent Xiclunae4a33802010-08-09 12:24:20 +000055 with open(filename, "rb") as f:
56 data = f.read()
Guido van Rossumb940e112007-01-10 16:19:56 +000057 except IOError as err:
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000058 sys.stderr.write("Can't open %s: %s\n" % (filename, err))
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000059 self.addstats(ext, "unopenable", 1)
60 return
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000061 self.addstats(ext, "bytes", len(data))
Guido van Rossum7c77f752007-09-27 22:39:12 +000062 if b'\0' in data:
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000063 self.addstats(ext, "binary", 1)
64 return
65 if not data:
66 self.addstats(ext, "empty", 1)
Florent Xiclunae4a33802010-08-09 12:24:20 +000067 # self.addstats(ext, "chars", len(data))
Guido van Rossum7c77f752007-09-27 22:39:12 +000068 lines = str(data, "latin-1").splitlines()
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000069 self.addstats(ext, "lines", len(lines))
70 del lines
71 words = data.split()
72 self.addstats(ext, "words", len(words))
73
74 def addstats(self, ext, key, n):
75 d = self.stats.setdefault(ext, {})
76 d[key] = d.get(key, 0) + n
77
78 def report(self):
Guido van Rossum7c77f752007-09-27 22:39:12 +000079 exts = sorted(self.stats)
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000080 # Get the column keys
81 columns = {}
82 for ext in exts:
83 columns.update(self.stats[ext])
Guido van Rossum7c77f752007-09-27 22:39:12 +000084 cols = sorted(columns)
Guido van Rossum21123ab2003-04-09 19:10:46 +000085 colwidth = {}
86 colwidth["ext"] = max([len(ext) for ext in exts])
87 minwidth = 6
88 self.stats["TOTAL"] = {}
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000089 for col in cols:
Guido van Rossum21123ab2003-04-09 19:10:46 +000090 total = 0
91 cw = max(minwidth, len(col))
92 for ext in exts:
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000093 value = self.stats[ext].get(col)
94 if value is None:
Guido van Rossum21123ab2003-04-09 19:10:46 +000095 w = 0
Guido van Rossum6a8f7b72003-01-01 14:41:25 +000096 else:
Guido van Rossum21123ab2003-04-09 19:10:46 +000097 w = len("%d" % value)
98 total += value
99 cw = max(cw, w)
100 cw = max(cw, len(str(total)))
101 colwidth[col] = cw
102 self.stats["TOTAL"][col] = total
103 exts.append("TOTAL")
104 for ext in exts:
105 self.stats[ext]["ext"] = ext
106 cols.insert(0, "ext")
Florent Xiclunae4a33802010-08-09 12:24:20 +0000107
Guido van Rossum21123ab2003-04-09 19:10:46 +0000108 def printheader():
109 for col in cols:
Collin Winter6afaeb72007-08-03 17:06:41 +0000110 print("%*s" % (colwidth[col], col), end=' ')
111 print()
Florent Xiclunae4a33802010-08-09 12:24:20 +0000112
Guido van Rossum21123ab2003-04-09 19:10:46 +0000113 printheader()
114 for ext in exts:
115 for col in cols:
116 value = self.stats[ext].get(col, "")
Collin Winter6afaeb72007-08-03 17:06:41 +0000117 print("%*s" % (colwidth[col], value), end=' ')
118 print()
Florent Xiclunae4a33802010-08-09 12:24:20 +0000119 printheader() # Another header at the bottom
120
Guido van Rossum6a8f7b72003-01-01 14:41:25 +0000121
122def main():
123 args = sys.argv[1:]
124 if not args:
125 args = [os.curdir]
126 s = Stats()
127 s.statargs(args)
128 s.report()
129
Florent Xiclunae4a33802010-08-09 12:24:20 +0000130
Guido van Rossum6a8f7b72003-01-01 14:41:25 +0000131if __name__ == "__main__":
132 main()