| 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 | |
| 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" % file) |
| 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" % (file, err)) |
| 29 | self.addstats(ext, "unlistable", 1) |
| 30 | return |
| 31 | names.sort() |
| 32 | for name in names: |
| 33 | full = os.path.join(dir, name) |
| 34 | if os.path.islink(full): |
| 35 | self.addstats("<lnk>", "links", 1) |
| 36 | elif os.path.isdir(full): |
| 37 | self.statdir(full) |
| 38 | else: |
| 39 | self.statfile(full) |
| 40 | |
| 41 | def statfile(self, file): |
| 42 | head, ext = os.path.splitext(file) |
| 43 | head, base = os.path.split(file) |
| 44 | if ext == base: |
| 45 | ext = "" # .cvsignore is deemed not to have an extension |
| 46 | self.addstats(ext, "files", 1) |
| 47 | try: |
| 48 | f = open(file, "rb") |
| 49 | except IOError, err: |
| 50 | sys.stderr.write("Can't open %s: %s\n" % (file, err)) |
| 51 | self.addstats(ext, "unopenable", 1) |
| 52 | return |
| 53 | data = f.read() |
| 54 | f.close() |
| 55 | self.addstats(ext, "bytes", len(data)) |
| 56 | if '\0' in data: |
| 57 | self.addstats(ext, "binary", 1) |
| 58 | return |
| 59 | if not data: |
| 60 | self.addstats(ext, "empty", 1) |
| 61 | #self.addstats(ext, "chars", len(data)) |
| 62 | lines = data.splitlines() |
| 63 | self.addstats(ext, "lines", len(lines)) |
| 64 | del lines |
| 65 | words = data.split() |
| 66 | self.addstats(ext, "words", len(words)) |
| 67 | |
| 68 | def addstats(self, ext, key, n): |
| 69 | d = self.stats.setdefault(ext, {}) |
| 70 | d[key] = d.get(key, 0) + n |
| 71 | |
| 72 | def report(self): |
| 73 | totals = {} |
| 74 | exts = self.stats.keys() |
| 75 | exts.sort() |
| 76 | # Get the column keys |
| 77 | columns = {} |
| 78 | for ext in exts: |
| 79 | columns.update(self.stats[ext]) |
| 80 | cols = columns.keys() |
| 81 | cols.sort() |
| 82 | minwidth = 7 |
| 83 | extwidth = max([len(ext) for ext in exts]) |
| 84 | print "%*s" % (extwidth, "ext"), |
| 85 | for col in cols: |
| 86 | width = max(len(col), minwidth) |
| 87 | print "%*s" % (width, col), |
| 88 | print |
| 89 | for ext in exts: |
| 90 | print "%*s" % (extwidth, ext), |
| 91 | for col in cols: |
| 92 | width = max(len(col), minwidth) |
| 93 | value = self.stats[ext].get(col) |
| 94 | if value is None: |
| 95 | s = "" |
| 96 | else: |
| 97 | s = "%d" % value |
| 98 | totals[col] = totals.get(col, 0) + value |
| 99 | print "%*s" % (width, s), |
| 100 | print |
| 101 | print "%*s" % (extwidth, "TOTAL"), |
| 102 | for col in cols: |
| 103 | width = max(len(col), minwidth) |
| 104 | print "%*s" % (width, totals[col]), |
| 105 | print |
| 106 | |
| 107 | def main(): |
| 108 | args = sys.argv[1:] |
| 109 | if not args: |
| 110 | args = [os.curdir] |
| 111 | s = Stats() |
| 112 | s.statargs(args) |
| 113 | s.report() |
| 114 | |
| 115 | if __name__ == "__main__": |
| 116 | main() |