Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 1 | #! /usr/bin/env python |
| 2 | |
| 3 | """Show file statistics by extension.""" |
| 4 | |
Éric Araujo | 0d983b5 | 2012-07-02 17:45:10 -0400 | [diff] [blame] | 5 | from __future__ import print_function |
| 6 | |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 7 | import os |
| 8 | import sys |
| 9 | |
| 10 | class Stats: |
| 11 | |
| 12 | def __init__(self): |
| 13 | self.stats = {} |
| 14 | |
| 15 | def statargs(self, args): |
| 16 | for arg in args: |
| 17 | if os.path.isdir(arg): |
| 18 | self.statdir(arg) |
| 19 | elif os.path.isfile(arg): |
| 20 | self.statfile(arg) |
| 21 | else: |
Neal Norwitz | 7fbb9d1 | 2006-03-24 05:36:33 +0000 | [diff] [blame] | 22 | sys.stderr.write("Can't find %s\n" % arg) |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 23 | self.addstats("<???>", "unknown", 1) |
| 24 | |
| 25 | def statdir(self, dir): |
| 26 | self.addstats("<dir>", "dirs", 1) |
| 27 | try: |
Guido van Rossum | 1ce5d18 | 2011-06-28 10:33:38 -0700 | [diff] [blame] | 28 | names = sorted(os.listdir(dir)) |
| 29 | except os.error as err: |
Neal Norwitz | 7fbb9d1 | 2006-03-24 05:36:33 +0000 | [diff] [blame] | 30 | sys.stderr.write("Can't list %s: %s\n" % (dir, err)) |
| 31 | self.addstats("<dir>", "unlistable", 1) |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 32 | return |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 33 | for name in names: |
Guido van Rossum | 21123ab | 2003-04-09 19:10:46 +0000 | [diff] [blame] | 34 | if name.startswith(".#"): |
| 35 | continue # Skip CVS temp files |
| 36 | if name.endswith("~"): |
| 37 | continue# Skip Emacs backup files |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 38 | full = os.path.join(dir, name) |
| 39 | if os.path.islink(full): |
| 40 | self.addstats("<lnk>", "links", 1) |
| 41 | elif os.path.isdir(full): |
| 42 | self.statdir(full) |
| 43 | else: |
| 44 | self.statfile(full) |
| 45 | |
Neal Norwitz | 7fbb9d1 | 2006-03-24 05:36:33 +0000 | [diff] [blame] | 46 | def statfile(self, filename): |
| 47 | head, ext = os.path.splitext(filename) |
| 48 | head, base = os.path.split(filename) |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 49 | if ext == base: |
Guido van Rossum | 21123ab | 2003-04-09 19:10:46 +0000 | [diff] [blame] | 50 | ext = "" # E.g. .cvsignore is deemed not to have an extension |
| 51 | ext = os.path.normcase(ext) |
| 52 | if not ext: |
| 53 | ext = "<none>" |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 54 | self.addstats(ext, "files", 1) |
| 55 | try: |
Neal Norwitz | 7fbb9d1 | 2006-03-24 05:36:33 +0000 | [diff] [blame] | 56 | f = open(filename, "rb") |
Guido van Rossum | 1ce5d18 | 2011-06-28 10:33:38 -0700 | [diff] [blame] | 57 | except IOError as err: |
Neal Norwitz | 7fbb9d1 | 2006-03-24 05:36:33 +0000 | [diff] [blame] | 58 | sys.stderr.write("Can't open %s: %s\n" % (filename, err)) |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 59 | self.addstats(ext, "unopenable", 1) |
| 60 | return |
| 61 | data = f.read() |
| 62 | f.close() |
| 63 | self.addstats(ext, "bytes", len(data)) |
Guido van Rossum | 1ce5d18 | 2011-06-28 10:33:38 -0700 | [diff] [blame] | 64 | if b'\0' in data: |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 65 | self.addstats(ext, "binary", 1) |
| 66 | return |
| 67 | if not data: |
| 68 | self.addstats(ext, "empty", 1) |
| 69 | #self.addstats(ext, "chars", len(data)) |
| 70 | lines = data.splitlines() |
| 71 | self.addstats(ext, "lines", len(lines)) |
| 72 | del lines |
| 73 | words = data.split() |
| 74 | self.addstats(ext, "words", len(words)) |
| 75 | |
| 76 | def addstats(self, ext, key, n): |
| 77 | d = self.stats.setdefault(ext, {}) |
| 78 | d[key] = d.get(key, 0) + n |
| 79 | |
| 80 | def report(self): |
Guido van Rossum | 1ce5d18 | 2011-06-28 10:33:38 -0700 | [diff] [blame] | 81 | exts = sorted(self.stats.keys()) |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 82 | # Get the column keys |
| 83 | columns = {} |
| 84 | for ext in exts: |
| 85 | columns.update(self.stats[ext]) |
Guido van Rossum | 1ce5d18 | 2011-06-28 10:33:38 -0700 | [diff] [blame] | 86 | cols = sorted(columns.keys()) |
Guido van Rossum | 21123ab | 2003-04-09 19:10:46 +0000 | [diff] [blame] | 87 | colwidth = {} |
| 88 | colwidth["ext"] = max([len(ext) for ext in exts]) |
| 89 | minwidth = 6 |
| 90 | self.stats["TOTAL"] = {} |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 91 | for col in cols: |
Guido van Rossum | 21123ab | 2003-04-09 19:10:46 +0000 | [diff] [blame] | 92 | total = 0 |
| 93 | cw = max(minwidth, len(col)) |
| 94 | for ext in exts: |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 95 | value = self.stats[ext].get(col) |
| 96 | if value is None: |
Guido van Rossum | 21123ab | 2003-04-09 19:10:46 +0000 | [diff] [blame] | 97 | w = 0 |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 98 | else: |
Guido van Rossum | 21123ab | 2003-04-09 19:10:46 +0000 | [diff] [blame] | 99 | w = len("%d" % value) |
| 100 | total += value |
| 101 | cw = max(cw, w) |
| 102 | cw = max(cw, len(str(total))) |
| 103 | colwidth[col] = cw |
| 104 | self.stats["TOTAL"][col] = total |
| 105 | exts.append("TOTAL") |
| 106 | for ext in exts: |
| 107 | self.stats[ext]["ext"] = ext |
| 108 | cols.insert(0, "ext") |
| 109 | def printheader(): |
| 110 | for col in cols: |
Guido van Rossum | 1ce5d18 | 2011-06-28 10:33:38 -0700 | [diff] [blame] | 111 | print("%*s" % (colwidth[col], col), end=" ") |
| 112 | print() |
Guido van Rossum | 21123ab | 2003-04-09 19:10:46 +0000 | [diff] [blame] | 113 | printheader() |
| 114 | for ext in exts: |
| 115 | for col in cols: |
| 116 | value = self.stats[ext].get(col, "") |
Guido van Rossum | 1ce5d18 | 2011-06-28 10:33:38 -0700 | [diff] [blame] | 117 | print("%*s" % (colwidth[col], value), end=" ") |
| 118 | print() |
Guido van Rossum | 21123ab | 2003-04-09 19:10:46 +0000 | [diff] [blame] | 119 | printheader() # Another header at the bottom |
Guido van Rossum | 6a8f7b7 | 2003-01-01 14:41:25 +0000 | [diff] [blame] | 120 | |
| 121 | def main(): |
| 122 | args = sys.argv[1:] |
| 123 | if not args: |
| 124 | args = [os.curdir] |
| 125 | s = Stats() |
| 126 | s.statargs(args) |
| 127 | s.report() |
| 128 | |
| 129 | if __name__ == "__main__": |
| 130 | main() |