Tools/freeze/findmodules.py

   1 # Determine the names and filenames of the modules imported by a
   2 # script, recursively.  This is done by scanning for lines containing
   3 # import statements.  (The scanning has only superficial knowledge of
   4 # Python syntax and no knowledge of semantics, so in theory the result
   5 # may be incorrect -- however this is quite unlikely if you don't
   6 # intentionally obscure your Python code.)
   7
   8 import os
   9 import regex
  10 import string
  11 import sys
  12
  13
  14 # Top-level interface.
  15 # First argument is the main program (script).
  16 # Second optional argument is list of modules to be searched as well.
  17
  18 def findmodules(scriptfile, modules = [], path = sys.path):
  19         todo = {}
  20         todo['__main__'] = scriptfile
  21         for name in modules:
  22                 mod = os.path.basename(name)
  23                 if mod[-3:] == '.py': mod = mod[:-3]
  24                 elif mod[-4:] == '.pyc': mod = mod[:-4]
  25                 todo[mod] = name
  26         done = closure(todo)
  27         return done
  28
  29
  30 # Compute the closure of scanfile() and findmodule().
  31 # Return a dictionary mapping module names to filenames.
  32 # Writes to stderr if a file can't be or read.
  33
  34 def closure(todo):
  35         done = {}
  36         while todo:
  37                 newtodo = {}
  38                 for modname in todo.keys():
  39                         if not done.has_key(modname):
  40                                 filename = todo[modname]
  41                                 if filename is None:
  42                                         filename = findmodule(modname)
  43                                 done[modname] = filename
  44                                 if filename in ('<builtin>', '<unknown>'):
  45                                         continue
  46                                 try:
  47                                         modules = scanfile(filename)
  48                                 except IOError, msg:
  49                                         sys.stderr.write("%s: %s\n" %
  50                                                          (filename, str(msg)))
  51                                         continue
  52                                 for m in modules:
  53                                         if not done.has_key(m):
  54                                                 newtodo[m] = None
  55                 todo = newtodo
  56         return done
  57
  58
  59 # Scan a file looking for import statements.
  60 # Return list of module names.
  61 # Can raise IOError.
  62
  63 importstr = '\(^\|:\)[ \t]*import[ \t]+\([a-zA-Z0-9_, \t]+\)'
  64 fromstr   = '\(^\|:\)[ \t]*from[ \t]+\([a-zA-Z0-9_]+\)[ \t]+import[ \t]+'
  65 isimport = regex.compile(importstr)
  66 isfrom = regex.compile(fromstr)
  67
  68 def scanfile(filename):
  69         allmodules = {}
  70         f = open(filename, 'r')
  71         try:
  72                 while 1:
  73                         line = f.readline()
  74                         if not line: break # EOF
  75                         while line[-2:] == '\\\n': # Continuation line
  76                                 line = line[:-2] + ' '
  77                                 line = line + f.readline()
  78                         if isimport.search(line) >= 0:
  79                                 rawmodules = isimport.group(2)
  80                                 modules = string.splitfields(rawmodules, ',')
  81                                 for i in range(len(modules)):
  82                                         modules[i] = string.strip(modules[i])
  83                         elif isfrom.search(line) >= 0:
  84                                 modules = [isfrom.group(2)]
  85                         else:
  86                                 continue
  87                         for mod in modules:
  88                                 allmodules[mod] = None
  89         finally:
  90                 f.close()
  91         return allmodules.keys()
  92
  93
  94 # Find the file containing a module, given its name.
  95 # Return filename, or '<builtin>', or '<unknown>'.
  96
  97 builtins = sys.builtin_module_names
  98 tails = ['.py', '.pyc']
  99
 100 def findmodule(modname, path = sys.path):
 101         if modname in builtins: return '<builtin>'
 102         for dirname in path:
 103                 for tail in tails:
 104                         fullname = os.path.join(dirname, modname + tail)
 105                         try:
 106                                 f = open(fullname, 'r')
 107                         except IOError:
 108                                 continue
 109                         f.close()
 110                         return fullname
 111         return '<unknown>'
 112
 113
 114 # Test the above functions.
 115
 116 def test():
 117         if not sys.argv[1:]:
 118                 print 'usage: python findmodules.py scriptfile [morefiles ...]'
 119                 sys.exit(2)
 120         done = findmodules(sys.argv[1], sys.argv[2:])
 121         items = done.items()
 122         items.sort()
 123         for mod, file in [('Module', 'File')] + items:
 124                 print "%-15s %s" % (mod, file)
 125
 126 if __name__ == '__main__':
 127         test()