| 1 | n/a | #! /usr/bin/env python3 |
|---|
| 2 | n/a | |
|---|
| 3 | n/a | """Show file statistics by extension.""" |
|---|
| 4 | n/a | |
|---|
| 5 | n/a | import os |
|---|
| 6 | n/a | import sys |
|---|
| 7 | n/a | |
|---|
| 8 | n/a | |
|---|
| 9 | n/a | class Stats: |
|---|
| 10 | n/a | |
|---|
| 11 | n/a | def __init__(self): |
|---|
| 12 | n/a | self.stats = {} |
|---|
| 13 | n/a | |
|---|
| 14 | n/a | def statargs(self, args): |
|---|
| 15 | n/a | for arg in args: |
|---|
| 16 | n/a | if os.path.isdir(arg): |
|---|
| 17 | n/a | self.statdir(arg) |
|---|
| 18 | n/a | elif os.path.isfile(arg): |
|---|
| 19 | n/a | self.statfile(arg) |
|---|
| 20 | n/a | else: |
|---|
| 21 | n/a | sys.stderr.write("Can't find %s\n" % arg) |
|---|
| 22 | n/a | self.addstats("<???>", "unknown", 1) |
|---|
| 23 | n/a | |
|---|
| 24 | n/a | def statdir(self, dir): |
|---|
| 25 | n/a | self.addstats("<dir>", "dirs", 1) |
|---|
| 26 | n/a | try: |
|---|
| 27 | n/a | names = os.listdir(dir) |
|---|
| 28 | n/a | except OSError as err: |
|---|
| 29 | n/a | sys.stderr.write("Can't list %s: %s\n" % (dir, err)) |
|---|
| 30 | n/a | self.addstats("<dir>", "unlistable", 1) |
|---|
| 31 | n/a | return |
|---|
| 32 | n/a | for name in sorted(names): |
|---|
| 33 | n/a | if name.startswith(".#"): |
|---|
| 34 | n/a | continue # Skip CVS temp files |
|---|
| 35 | n/a | if name.endswith("~"): |
|---|
| 36 | n/a | continue # Skip Emacs backup files |
|---|
| 37 | n/a | full = os.path.join(dir, name) |
|---|
| 38 | n/a | if os.path.islink(full): |
|---|
| 39 | n/a | self.addstats("<lnk>", "links", 1) |
|---|
| 40 | n/a | elif os.path.isdir(full): |
|---|
| 41 | n/a | self.statdir(full) |
|---|
| 42 | n/a | else: |
|---|
| 43 | n/a | self.statfile(full) |
|---|
| 44 | n/a | |
|---|
| 45 | n/a | def statfile(self, filename): |
|---|
| 46 | n/a | head, ext = os.path.splitext(filename) |
|---|
| 47 | n/a | head, base = os.path.split(filename) |
|---|
| 48 | n/a | if ext == base: |
|---|
| 49 | n/a | ext = "" # E.g. .cvsignore is deemed not to have an extension |
|---|
| 50 | n/a | ext = os.path.normcase(ext) |
|---|
| 51 | n/a | if not ext: |
|---|
| 52 | n/a | ext = "<none>" |
|---|
| 53 | n/a | self.addstats(ext, "files", 1) |
|---|
| 54 | n/a | try: |
|---|
| 55 | n/a | with open(filename, "rb") as f: |
|---|
| 56 | n/a | data = f.read() |
|---|
| 57 | n/a | except IOError as err: |
|---|
| 58 | n/a | sys.stderr.write("Can't open %s: %s\n" % (filename, err)) |
|---|
| 59 | n/a | self.addstats(ext, "unopenable", 1) |
|---|
| 60 | n/a | return |
|---|
| 61 | n/a | self.addstats(ext, "bytes", len(data)) |
|---|
| 62 | n/a | if b'\0' in data: |
|---|
| 63 | n/a | self.addstats(ext, "binary", 1) |
|---|
| 64 | n/a | return |
|---|
| 65 | n/a | if not data: |
|---|
| 66 | n/a | self.addstats(ext, "empty", 1) |
|---|
| 67 | n/a | # self.addstats(ext, "chars", len(data)) |
|---|
| 68 | n/a | lines = str(data, "latin-1").splitlines() |
|---|
| 69 | n/a | self.addstats(ext, "lines", len(lines)) |
|---|
| 70 | n/a | del lines |
|---|
| 71 | n/a | words = data.split() |
|---|
| 72 | n/a | self.addstats(ext, "words", len(words)) |
|---|
| 73 | n/a | |
|---|
| 74 | n/a | def addstats(self, ext, key, n): |
|---|
| 75 | n/a | d = self.stats.setdefault(ext, {}) |
|---|
| 76 | n/a | d[key] = d.get(key, 0) + n |
|---|
| 77 | n/a | |
|---|
| 78 | n/a | def report(self): |
|---|
| 79 | n/a | exts = sorted(self.stats) |
|---|
| 80 | n/a | # Get the column keys |
|---|
| 81 | n/a | columns = {} |
|---|
| 82 | n/a | for ext in exts: |
|---|
| 83 | n/a | columns.update(self.stats[ext]) |
|---|
| 84 | n/a | cols = sorted(columns) |
|---|
| 85 | n/a | colwidth = {} |
|---|
| 86 | n/a | colwidth["ext"] = max([len(ext) for ext in exts]) |
|---|
| 87 | n/a | minwidth = 6 |
|---|
| 88 | n/a | self.stats["TOTAL"] = {} |
|---|
| 89 | n/a | for col in cols: |
|---|
| 90 | n/a | total = 0 |
|---|
| 91 | n/a | cw = max(minwidth, len(col)) |
|---|
| 92 | n/a | for ext in exts: |
|---|
| 93 | n/a | value = self.stats[ext].get(col) |
|---|
| 94 | n/a | if value is None: |
|---|
| 95 | n/a | w = 0 |
|---|
| 96 | n/a | else: |
|---|
| 97 | n/a | w = len("%d" % value) |
|---|
| 98 | n/a | total += value |
|---|
| 99 | n/a | cw = max(cw, w) |
|---|
| 100 | n/a | cw = max(cw, len(str(total))) |
|---|
| 101 | n/a | colwidth[col] = cw |
|---|
| 102 | n/a | self.stats["TOTAL"][col] = total |
|---|
| 103 | n/a | exts.append("TOTAL") |
|---|
| 104 | n/a | for ext in exts: |
|---|
| 105 | n/a | self.stats[ext]["ext"] = ext |
|---|
| 106 | n/a | cols.insert(0, "ext") |
|---|
| 107 | n/a | |
|---|
| 108 | n/a | def printheader(): |
|---|
| 109 | n/a | for col in cols: |
|---|
| 110 | n/a | print("%*s" % (colwidth[col], col), end=' ') |
|---|
| 111 | n/a | print() |
|---|
| 112 | n/a | |
|---|
| 113 | n/a | printheader() |
|---|
| 114 | n/a | for ext in exts: |
|---|
| 115 | n/a | for col in cols: |
|---|
| 116 | n/a | value = self.stats[ext].get(col, "") |
|---|
| 117 | n/a | print("%*s" % (colwidth[col], value), end=' ') |
|---|
| 118 | n/a | print() |
|---|
| 119 | n/a | printheader() # Another header at the bottom |
|---|
| 120 | n/a | |
|---|
| 121 | n/a | |
|---|
| 122 | n/a | def main(): |
|---|
| 123 | n/a | args = sys.argv[1:] |
|---|
| 124 | n/a | if not args: |
|---|
| 125 | n/a | args = [os.curdir] |
|---|
| 126 | n/a | s = Stats() |
|---|
| 127 | n/a | s.statargs(args) |
|---|
| 128 | n/a | s.report() |
|---|
| 129 | n/a | |
|---|
| 130 | n/a | |
|---|
| 131 | n/a | if __name__ == "__main__": |
|---|
| 132 | n/a | main() |
|---|