Doc/tools/listmodules

   1 #! /usr/bin/env python
   2 #  -*- Python -*-
   3 #
   4 #  This script can be used to identify undocumented modules in the Python
   5 #  standard library.  Use it like this:
   6 #
   7 #  .../Doc/tools/listmodules --ignore-from .../Doc/paper-<paper>/modlib.idx
   8
   9 """%(program)s - list modules in the Python standard library
  10
  11 -a, --annotate    Annotate the module names with the subdirectory they
  12                   live in
  13 -c, --categorize  Group the modules by subdirectory
  14 -i <file>,
  15
  16 --ignore-from <file>    Ignore the modules listed in <file>.  <file> may
  17                   contain a list of module names or a module index file
  18                   as produced when formatting the Python documentation
  19                   (.idx or .html flavor).
  20
  21 If neither -a nor -c are given, the modules are listed in alphabetical
  22 order.
  23
  24 Note that -a and -c are mutually exclusive.
  25
  26 Limitation: Modules loadable as shared objects may not be listed,
  27 though this script attempts to locate such modules.
  28
  29 """
  30
  31 __version__ = '$Revision$'
  32
  33 import getopt
  34 import glob
  35 import os
  36 import re
  37 import string
  38 import sys
  39
  40
  41 REMOVE_DIRS = ["dos-8x3", "encodings", "distutils",
  42                "lib-old", "lib-stdwin", "test"]
  43
  44
  45 def main():
  46     args = sys.argv[1:]
  47     annotate = 0
  48     builtin = 0
  49     categorize = 0
  50     ignore_dict = {}
  51     ignore = ignore_dict.has_key
  52     try:
  53         opts, args = getopt.getopt(
  54             args, "abchi:",
  55             ["annotate", "built-in", "categorize", "help", "ignore-from="])
  56     except getopt.error, msg:
  57         sys.stdout = sys.stderr
  58         print msg
  59         print
  60         usage()
  61         sys.exit(2)
  62     for opt, arg in opts:
  63         if opt in ("-a", "--annotate"):
  64             annotate = 1
  65         elif opt in ("-b", "--built-in"):
  66             builtin = 1
  67         elif opt in ("-c", "--categorize"):
  68             categorize = 1
  69         elif opt in ("-h", "--help"):
  70             usage()
  71             sys.exit()
  72         elif opt in ("-i", "--ignore-from"):
  73             data = open(arg).read()
  74             if data[:1] == "\\":
  75                 ignore_from_idx(data, ignore_dict)
  76             else:
  77                 ignore_from_modulelist(data, ignore_dict)
  78     if args or (annotate and categorize):
  79         usage()
  80         sys.exit(2)
  81     #
  82     # Populate the database:
  83     #
  84     srcdir = os.path.normpath(os.path.join(
  85         os.path.dirname(sys.argv[0]), os.pardir, os.pardir))
  86     os.chdir(srcdir)
  87     modules_by_name = {}
  88     modules_by_dir = {}
  89     if builtin:
  90         l = []
  91         modules_by_dir["<builtin>"] = l
  92         for name in sys.builtin_module_names:
  93             if not ignore(name):
  94                 modules_by_name[name] = "<built-in>"
  95                 l.append(name)
  96     rx = re.compile("Lib/plat-[a-zA-Z0-9]*/")
  97     fp = os.popen("find Lib -name \*.py -print", "r")
  98     while 1:
  99         line = fp.readline()
 100         if not line:
 101             break
 102         m = rx.match(line)
 103         if m:
 104             line = "Lib/plat-*/" + line[m.end():]
 105         line = line[4:-4]                # strip off 'Lib/' and '.py\n'
 106         dir, name = os.path.split(line)
 107         dir = dir or "<standard>"
 108         if ignore(name):
 109             continue
 110         if dir not in REMOVE_DIRS:
 111             modules_by_name[name] = dir
 112             l = modules_by_dir.get(dir, [])
 113             modules_by_dir[dir] = l
 114             if name not in l:
 115                 l.append(name)
 116     # load up extension modules:
 117     pwd = os.getcwd()
 118     try:
 119         os.chdir("Modules")
 120         dir = "<extension>"
 121         for line in glob.glob("*module.c"):
 122             name = line[:-8]
 123             if ignore(name) or modules_by_name.has_key(name) or name == "xx":
 124                 continue
 125             modules_by_name[name] = dir
 126             l = modules_by_dir.get(dir, [])
 127             modules_by_dir[dir] = l
 128             if name not in l:
 129                 l.append(name)
 130     finally:
 131         os.chdir(pwd)
 132     #
 133     # Dump the results:
 134     #
 135     if annotate:
 136         modules = modules_by_name.items()
 137         modules.sort()
 138         width = max(map(len, modules_by_name.keys()))
 139         format = "%%-%ds  %%s" % width
 140         for name, dir in modules:
 141             if dir and dir[0] != "<":
 142                 print format % (name, dir)
 143             else:
 144                 print name
 145     elif categorize:
 146         modules = modules_by_dir.items()
 147         modules.sort()
 148         width = max(map(len, modules_by_dir.keys()))
 149         format = "%%-%ds  %%s" % width
 150         for dir, names in modules:
 151             names.sort()
 152             print format % (dir, names[0])
 153             for name in names[1:]:
 154                 print format % ('', name)
 155             print
 156     else:
 157         modules = modules_by_name.keys()
 158         modules.sort()
 159         print string.join(modules, "\n")
 160
 161
 162 def ignore_from_modulelist(data, ignore_dict):
 163     for name in string.split(data):
 164         ignore_dict[name] = name
 165
 166 def ignore_from_idx(data, ignore_dict):
 167     data = string.replace(data, r"\hackscore  {}", "_")
 168     rx = re.compile(r"\\indexentry\s*{([^@]*)@")
 169     for line in string.split(data, "\n"):
 170         m = rx.match(line)
 171         if m:
 172             name = m.group(1)
 173             ignore_dict[name] = name
 174
 175
 176 def usage():
 177     vars = {}
 178     vars["program"] = os.path.basename(sys.argv[0])
 179     print __doc__ % vars
 180
 181
 182 if __name__ == "__main__":
 183     main()