Tools/scripts/byext.py

   1 #! /usr/bin/env python
   2
   3 """Show file statistics by extension."""
   4
   5 import os
   6 import sys
   7
   8 class Stats:
   9
  10     def __init__(self):
  11         self.stats = {}
  12
  13     def statargs(self, args):
  14         for arg in args:
  15             if os.path.isdir(arg):
  16                 self.statdir(arg)
  17             elif os.path.isfile(arg):
  18                 self.statfile(arg)
  19             else:
  20                 sys.stderr.write("Can't find %s\n" % arg)
  21                 self.addstats("<???>", "unknown", 1)
  22
  23     def statdir(self, dir):
  24         self.addstats("<dir>", "dirs", 1)
  25         try:
  26             names = os.listdir(dir)
  27         except os.error, err:
  28             sys.stderr.write("Can't list %s: %s\n" % (dir, err))
  29             self.addstats("<dir>", "unlistable", 1)
  30             return
  31         names.sort()
  32         for name in names:
  33             if name.startswith(".#"):
  34                 continue # Skip CVS temp files
  35             if name.endswith("~"):
  36                 continue# Skip Emacs backup files
  37             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
  45     def statfile(self, filename):
  46         head, ext = os.path.splitext(filename)
  47         head, base = os.path.split(filename)
  48         if ext == base:
  49             ext = "" # E.g. .cvsignore is deemed not to have an extension
  50         ext = os.path.normcase(ext)
  51         if not ext:
  52             ext = "<none>"
  53         self.addstats(ext, "files", 1)
  54         try:
  55             f = open(filename, "rb")
  56         except IOError, err:
  57             sys.stderr.write("Can't open %s: %s\n" % (filename, err))
  58             self.addstats(ext, "unopenable", 1)
  59             return
  60         data = f.read()
  61         f.close()
  62         self.addstats(ext, "bytes", len(data))
  63         if '\0' in data:
  64             self.addstats(ext, "binary", 1)
  65             return
  66         if not data:
  67             self.addstats(ext, "empty", 1)
  68         #self.addstats(ext, "chars", len(data))
  69         lines = data.splitlines()
  70         self.addstats(ext, "lines", len(lines))
  71         del lines
  72         words = data.split()
  73         self.addstats(ext, "words", len(words))
  74
  75     def addstats(self, ext, key, n):
  76         d = self.stats.setdefault(ext, {})
  77         d[key] = d.get(key, 0) + n
  78
  79     def report(self):
  80         exts = self.stats.keys()
  81         exts.sort()
  82         # Get the column keys
  83         columns = {}
  84         for ext in exts:
  85             columns.update(self.stats[ext])
  86         cols = columns.keys()
  87         cols.sort()
  88         colwidth = {}
  89         colwidth["ext"] = max([len(ext) for ext in exts])
  90         minwidth = 6
  91         self.stats["TOTAL"] = {}
  92         for col in cols:
  93             total = 0
  94             cw = max(minwidth, len(col))
  95             for ext in exts:
  96                 value = self.stats[ext].get(col)
  97                 if value is None:
  98                     w = 0
  99                 else:
 100                     w = len("%d" % value)
 101                     total += value
 102                 cw = max(cw, w)
 103             cw = max(cw, len(str(total)))
 104             colwidth[col] = cw
 105             self.stats["TOTAL"][col] = total
 106         exts.append("TOTAL")
 107         for ext in exts:
 108             self.stats[ext]["ext"] = ext
 109         cols.insert(0, "ext")
 110         def printheader():
 111             for col in cols:
 112                 print "%*s" % (colwidth[col], col),
 113             print
 114         printheader()
 115         for ext in exts:
 116             for col in cols:
 117                 value = self.stats[ext].get(col, "")
 118                 print "%*s" % (colwidth[col], value),
 119             print
 120         printheader() # Another header at the bottom
 121
 122 def 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
 130 if __name__ == "__main__":
 131     main()