blob: b19941e61e82bdde3616650c5e371b04738a9b28 [file] [log] [blame]
Guido van Rossum75dc4961998-03-05 03:42:00 +00001"""Find modules used by a script, using introspection."""
Thomas Heller919000e2002-11-25 20:21:59 +00002
Guido van Rossum75dc4961998-03-05 03:42:00 +00003import dis
Brett Cannon73b969e2012-12-22 19:34:21 -05004import importlib.machinery
Guido van Rossum75dc4961998-03-05 03:42:00 +00005import marshal
6import os
Guido van Rossum75dc4961998-03-05 03:42:00 +00007import sys
Christian Heimes45f9af32007-11-27 21:50:00 +00008import types
Guido van Rossumfc2a0a82006-10-27 23:06:01 +00009import struct
Brett Cannone4f41de2013-06-16 13:13:40 -040010import warnings
11with warnings.catch_warnings():
12 warnings.simplefilter('ignore', PendingDeprecationWarning)
13 import imp
Guido van Rossum75dc4961998-03-05 03:42:00 +000014
Brett Cannon40430012007-10-22 20:24:51 +000015# XXX Clean up once str8's cstor matches bytes.
Guido van Rossum98297ee2007-11-06 21:34:58 +000016LOAD_CONST = bytes([dis.opname.index('LOAD_CONST')])
17IMPORT_NAME = bytes([dis.opname.index('IMPORT_NAME')])
18STORE_NAME = bytes([dis.opname.index('STORE_NAME')])
19STORE_GLOBAL = bytes([dis.opname.index('STORE_GLOBAL')])
Just van Rossume29310a2002-12-31 16:33:00 +000020STORE_OPS = [STORE_NAME, STORE_GLOBAL]
Guido van Rossum98297ee2007-11-06 21:34:58 +000021HAVE_ARGUMENT = bytes([dis.HAVE_ARGUMENT])
Guido van Rossum75dc4961998-03-05 03:42:00 +000022
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000023# Modulefinder does a good job at simulating Python's, but it can not
24# handle __path__ modifications packages make at runtime. Therefore there
25# is a mechanism whereby you can register extra paths in this map for a
Thomas Wouters7e474022000-07-16 12:04:32 +000026# package, and it will be honored.
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000027
28# Note this is a mapping is lists of paths.
29packagePathMap = {}
30
31# A Public interface
32def AddPackagePath(packagename, path):
Éric Araujocee6bb52011-08-01 15:29:07 +020033 packagePathMap.setdefault(packagename, []).append(path)
Guido van Rossum75dc4961998-03-05 03:42:00 +000034
Thomas Hellerc7aaf952002-11-14 18:45:11 +000035replacePackageMap = {}
36
Martin v. Löwis2f48d892011-05-09 08:05:43 +020037# This ReplacePackage mechanism allows modulefinder to work around
38# situations in which a package injects itself under the name
39# of another package into sys.modules at runtime by calling
40# ReplacePackage("real_package_name", "faked_package_name")
Thomas Hellerc7aaf952002-11-14 18:45:11 +000041# before running ModuleFinder.
42
43def ReplacePackage(oldname, newname):
44 replacePackageMap[oldname] = newname
45
46
Guido van Rossum75dc4961998-03-05 03:42:00 +000047class Module:
48
49 def __init__(self, name, file=None, path=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +000050 self.__name__ = name
51 self.__file__ = file
52 self.__path__ = path
53 self.__code__ = None
Just van Rossume29310a2002-12-31 16:33:00 +000054 # The set of global names that are assigned to in the module.
55 # This includes those names imported through starimports of
56 # Python modules.
57 self.globalnames = {}
58 # The set of starimports this module did that could not be
59 # resolved, ie. a starimport from a non-Python module.
60 self.starimports = {}
Guido van Rossum75dc4961998-03-05 03:42:00 +000061
62 def __repr__(self):
Neil Schemenauer32d23c92004-02-15 16:43:20 +000063 s = "Module(%r" % (self.__name__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000064 if self.__file__ is not None:
Walter Dörwald70a6b492004-02-12 17:35:32 +000065 s = s + ", %r" % (self.__file__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000066 if self.__path__ is not None:
Walter Dörwald70a6b492004-02-12 17:35:32 +000067 s = s + ", %r" % (self.__path__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000068 s = s + ")"
69 return s
Guido van Rossum75dc4961998-03-05 03:42:00 +000070
Guido van Rossum75dc4961998-03-05 03:42:00 +000071class ModuleFinder:
72
Just van Rossume29310a2002-12-31 16:33:00 +000073 def __init__(self, path=None, debug=0, excludes=[], replace_paths=[]):
Guido van Rossum912a14c1998-03-05 04:56:37 +000074 if path is None:
75 path = sys.path
76 self.path = path
77 self.modules = {}
78 self.badmodules = {}
79 self.debug = debug
80 self.indent = 0
Guido van Rossum78fc3631998-03-20 17:37:24 +000081 self.excludes = excludes
Guido van Rossum6b767ac2001-03-20 20:43:34 +000082 self.replace_paths = replace_paths
83 self.processed_paths = [] # Used in debugging only
Guido van Rossum75dc4961998-03-05 03:42:00 +000084
85 def msg(self, level, str, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000086 if level <= self.debug:
87 for i in range(self.indent):
Guido van Rossumbe19ed72007-02-09 05:37:30 +000088 print(" ", end=' ')
89 print(str, end=' ')
Guido van Rossum912a14c1998-03-05 04:56:37 +000090 for arg in args:
Guido van Rossumbe19ed72007-02-09 05:37:30 +000091 print(repr(arg), end=' ')
92 print()
Guido van Rossum75dc4961998-03-05 03:42:00 +000093
94 def msgin(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000095 level = args[0]
96 if level <= self.debug:
97 self.indent = self.indent + 1
Guido van Rossum68468eb2003-02-27 20:14:51 +000098 self.msg(*args)
Guido van Rossum75dc4961998-03-05 03:42:00 +000099
100 def msgout(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000101 level = args[0]
102 if level <= self.debug:
103 self.indent = self.indent - 1
Guido van Rossum68468eb2003-02-27 20:14:51 +0000104 self.msg(*args)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000105
106 def run_script(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000107 self.msg(2, "run_script", pathname)
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200108 with open(pathname) as fp:
Brett Cannon028011f2010-10-30 00:26:48 +0000109 stuff = ("", "r", imp.PY_SOURCE)
110 self.load_module('__main__', fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000111
112 def load_file(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000113 dir, name = os.path.split(pathname)
114 name, ext = os.path.splitext(name)
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200115 with open(pathname) as fp:
Brett Cannon028011f2010-10-30 00:26:48 +0000116 stuff = (ext, "r", imp.PY_SOURCE)
117 self.load_module(name, fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000118
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000119 def import_hook(self, name, caller=None, fromlist=None, level=-1):
120 self.msg(3, "import_hook", name, caller, fromlist, level)
121 parent = self.determine_parent(caller, level=level)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000122 q, tail = self.find_head_package(parent, name)
123 m = self.load_tail(q, tail)
124 if not fromlist:
125 return q
126 if m.__path__:
127 self.ensure_fromlist(m, fromlist)
Thomas Heller318b7b92002-11-26 08:06:50 +0000128 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000129
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000130 def determine_parent(self, caller, level=-1):
131 self.msgin(4, "determine_parent", caller, level)
132 if not caller or level == 0:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000133 self.msgout(4, "determine_parent -> None")
134 return None
135 pname = caller.__name__
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000136 if level >= 1: # relative import
137 if caller.__path__:
138 level -= 1
139 if level == 0:
140 parent = self.modules[pname]
141 assert parent is caller
142 self.msgout(4, "determine_parent ->", parent)
143 return parent
144 if pname.count(".") < level:
Collin Winterce36ad82007-08-30 01:19:48 +0000145 raise ImportError("relative importpath too deep")
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000146 pname = ".".join(pname.split(".")[:-level])
147 parent = self.modules[pname]
148 self.msgout(4, "determine_parent ->", parent)
149 return parent
Guido van Rossum912a14c1998-03-05 04:56:37 +0000150 if caller.__path__:
151 parent = self.modules[pname]
152 assert caller is parent
153 self.msgout(4, "determine_parent ->", parent)
154 return parent
155 if '.' in pname:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000156 i = pname.rfind('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000157 pname = pname[:i]
158 parent = self.modules[pname]
159 assert parent.__name__ == pname
160 self.msgout(4, "determine_parent ->", parent)
161 return parent
162 self.msgout(4, "determine_parent -> None")
163 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000164
165 def find_head_package(self, parent, name):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000166 self.msgin(4, "find_head_package", parent, name)
167 if '.' in name:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000168 i = name.find('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000169 head = name[:i]
170 tail = name[i+1:]
171 else:
172 head = name
173 tail = ""
174 if parent:
175 qname = "%s.%s" % (parent.__name__, head)
176 else:
177 qname = head
178 q = self.import_module(head, qname, parent)
179 if q:
180 self.msgout(4, "find_head_package ->", (q, tail))
181 return q, tail
182 if parent:
183 qname = head
184 parent = None
185 q = self.import_module(head, qname, parent)
186 if q:
187 self.msgout(4, "find_head_package ->", (q, tail))
188 return q, tail
189 self.msgout(4, "raise ImportError: No module named", qname)
Collin Winterce36ad82007-08-30 01:19:48 +0000190 raise ImportError("No module named " + qname)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000191
192 def load_tail(self, q, tail):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000193 self.msgin(4, "load_tail", q, tail)
194 m = q
195 while tail:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000196 i = tail.find('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000197 if i < 0: i = len(tail)
198 head, tail = tail[:i], tail[i+1:]
199 mname = "%s.%s" % (m.__name__, head)
200 m = self.import_module(head, mname, m)
201 if not m:
202 self.msgout(4, "raise ImportError: No module named", mname)
Collin Winterce36ad82007-08-30 01:19:48 +0000203 raise ImportError("No module named " + mname)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000204 self.msgout(4, "load_tail ->", m)
205 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000206
207 def ensure_fromlist(self, m, fromlist, recursive=0):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000208 self.msg(4, "ensure_fromlist", m, fromlist, recursive)
209 for sub in fromlist:
210 if sub == "*":
211 if not recursive:
212 all = self.find_all_submodules(m)
213 if all:
214 self.ensure_fromlist(m, all, 1)
215 elif not hasattr(m, sub):
216 subname = "%s.%s" % (m.__name__, sub)
217 submod = self.import_module(sub, subname, m)
218 if not submod:
Collin Winterce36ad82007-08-30 01:19:48 +0000219 raise ImportError("No module named " + subname)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000220
221 def find_all_submodules(self, m):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000222 if not m.__path__:
223 return
224 modules = {}
Thomas Helleraaf1c8d2003-11-14 10:28:42 +0000225 # 'suffixes' used to be a list hardcoded to [".py", ".pyc", ".pyo"].
226 # But we must also collect Python extension modules - although
227 # we cannot separate normal dlls from Python extensions.
228 suffixes = []
Brett Cannoncb66eb02012-05-11 12:58:42 -0400229 suffixes += importlib.machinery.EXTENSION_SUFFIXES[:]
230 suffixes += importlib.machinery.SOURCE_SUFFIXES[:]
231 suffixes += importlib.machinery.BYTECODE_SUFFIXES[:]
Guido van Rossum912a14c1998-03-05 04:56:37 +0000232 for dir in m.__path__:
233 try:
234 names = os.listdir(dir)
Andrew Svetlovad28c7f2012-12-18 22:02:39 +0200235 except OSError:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000236 self.msg(2, "can't list directory", dir)
237 continue
238 for name in names:
239 mod = None
240 for suff in suffixes:
241 n = len(suff)
242 if name[-n:] == suff:
243 mod = name[:-n]
244 break
245 if mod and mod != "__init__":
246 modules[mod] = mod
247 return modules.keys()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000248
249 def import_module(self, partname, fqname, parent):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000250 self.msgin(3, "import_module", partname, fqname, parent)
251 try:
252 m = self.modules[fqname]
253 except KeyError:
254 pass
255 else:
256 self.msgout(3, "import_module ->", m)
257 return m
Guido van Rossume2b70bc2006-08-18 22:13:04 +0000258 if fqname in self.badmodules:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000259 self.msgout(3, "import_module -> None")
260 return None
Thomas Heller2e7c8322004-05-11 15:10:59 +0000261 if parent and parent.__path__ is None:
262 self.msgout(3, "import_module -> None")
263 return None
Guido van Rossum912a14c1998-03-05 04:56:37 +0000264 try:
265 fp, pathname, stuff = self.find_module(partname,
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000266 parent and parent.__path__, parent)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000267 except ImportError:
268 self.msgout(3, "import_module ->", None)
269 return None
270 try:
271 m = self.load_module(fqname, fp, pathname, stuff)
272 finally:
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200273 if fp:
274 fp.close()
Guido van Rossum912a14c1998-03-05 04:56:37 +0000275 if parent:
276 setattr(parent, partname, m)
277 self.msgout(3, "import_module ->", m)
278 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000279
Guido van Rossum1bc535d2007-05-15 18:46:22 +0000280 def load_module(self, fqname, fp, pathname, file_info):
281 suffix, mode, type = file_info
Guido van Rossum912a14c1998-03-05 04:56:37 +0000282 self.msgin(2, "load_module", fqname, fp and "fp", pathname)
283 if type == imp.PKG_DIRECTORY:
284 m = self.load_package(fqname, pathname)
285 self.msgout(2, "load_module ->", m)
286 return m
287 if type == imp.PY_SOURCE:
Guido van Rossum78fc3631998-03-20 17:37:24 +0000288 co = compile(fp.read()+'\n', pathname, 'exec')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000289 elif type == imp.PY_COMPILED:
290 if fp.read(4) != imp.get_magic():
291 self.msgout(2, "raise ImportError: Bad magic number", pathname)
Collin Winterce36ad82007-08-30 01:19:48 +0000292 raise ImportError("Bad magic number in %s" % pathname)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000293 fp.read(4)
294 co = marshal.load(fp)
295 else:
296 co = None
297 m = self.add_module(fqname)
Guido van Rossumab045f91998-03-06 19:55:10 +0000298 m.__file__ = pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000299 if co:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000300 if self.replace_paths:
301 co = self.replace_paths_in_code(co)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000302 m.__code__ = co
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000303 self.scan_code(co, m)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000304 self.msgout(2, "load_module ->", m)
305 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000306
Just van Rossume29310a2002-12-31 16:33:00 +0000307 def _add_badmodule(self, name, caller):
308 if name not in self.badmodules:
309 self.badmodules[name] = {}
Benjamin Petersonc0747cf2008-11-03 20:31:38 +0000310 if caller:
311 self.badmodules[name][caller.__name__] = 1
312 else:
313 self.badmodules[name]["-"] = 1
Just van Rossume29310a2002-12-31 16:33:00 +0000314
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000315 def _safe_import_hook(self, name, caller, fromlist, level=-1):
Just van Rossume29310a2002-12-31 16:33:00 +0000316 # wrapper for self.import_hook() that won't raise ImportError
317 if name in self.badmodules:
318 self._add_badmodule(name, caller)
319 return
320 try:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000321 self.import_hook(name, caller, level=level)
Guido van Rossumb940e112007-01-10 16:19:56 +0000322 except ImportError as msg:
Just van Rossume29310a2002-12-31 16:33:00 +0000323 self.msg(2, "ImportError:", str(msg))
324 self._add_badmodule(name, caller)
325 else:
326 if fromlist:
327 for sub in fromlist:
328 if sub in self.badmodules:
329 self._add_badmodule(sub, caller)
330 continue
331 try:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000332 self.import_hook(name, caller, [sub], level=level)
Guido van Rossumb940e112007-01-10 16:19:56 +0000333 except ImportError as msg:
Just van Rossume29310a2002-12-31 16:33:00 +0000334 self.msg(2, "ImportError:", str(msg))
335 fullname = name + "." + sub
336 self._add_badmodule(fullname, caller)
337
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000338 def scan_opcodes_25(self, co,
339 unpack = struct.unpack):
340 # Scan the code, and yield 'interesting' opcode combinations
341 # Python 2.5 version (has absolute and relative imports)
342 code = co.co_code
343 names = co.co_names
344 consts = co.co_consts
345 LOAD_LOAD_AND_IMPORT = LOAD_CONST + LOAD_CONST + IMPORT_NAME
346 while code:
Guido van Rossum98297ee2007-11-06 21:34:58 +0000347 c = bytes([code[0]])
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000348 if c in STORE_OPS:
349 oparg, = unpack('<H', code[1:3])
350 yield "store", (names[oparg],)
351 code = code[3:]
352 continue
353 if code[:9:3] == LOAD_LOAD_AND_IMPORT:
354 oparg_1, oparg_2, oparg_3 = unpack('<xHxHxH', code[:9])
355 level = consts[oparg_1]
356 if level == 0: # absolute import
357 yield "absolute_import", (consts[oparg_2], names[oparg_3])
358 else: # relative import
359 yield "relative_import", (level, consts[oparg_2], names[oparg_3])
360 code = code[9:]
361 continue
362 if c >= HAVE_ARGUMENT:
363 code = code[3:]
364 else:
365 code = code[1:]
366
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000367 def scan_code(self, co, m):
368 code = co.co_code
Serhiy Storchaka7e527052014-01-20 21:29:31 +0200369 scanner = self.scan_opcodes_25
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000370 for what, args in scanner(co):
371 if what == "store":
372 name, = args
373 m.globalnames[name] = 1
374 elif what == "absolute_import":
375 fromlist, name = args
Just van Rossume29310a2002-12-31 16:33:00 +0000376 have_star = 0
377 if fromlist is not None:
378 if "*" in fromlist:
379 have_star = 1
380 fromlist = [f for f in fromlist if f != "*"]
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000381 self._safe_import_hook(name, m, fromlist, level=0)
Just van Rossume29310a2002-12-31 16:33:00 +0000382 if have_star:
383 # We've encountered an "import *". If it is a Python module,
384 # the code has already been parsed and we can suck out the
385 # global names.
386 mm = None
387 if m.__path__:
388 # At this point we don't know whether 'name' is a
389 # submodule of 'm' or a global module. Let's just try
390 # the full name first.
391 mm = self.modules.get(m.__name__ + "." + name)
392 if mm is None:
393 mm = self.modules.get(name)
394 if mm is not None:
395 m.globalnames.update(mm.globalnames)
396 m.starimports.update(mm.starimports)
397 if mm.__code__ is None:
398 m.starimports[name] = 1
399 else:
400 m.starimports[name] = 1
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000401 elif what == "relative_import":
402 level, fromlist, name = args
403 if name:
404 self._safe_import_hook(name, m, fromlist, level=level)
405 else:
406 parent = self.determine_parent(m, level=level)
407 self._safe_import_hook(parent.__name__, None, fromlist, level=0)
408 else:
409 # We don't expect anything else from the generator.
410 raise RuntimeError(what)
411
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000412 for c in co.co_consts:
413 if isinstance(c, type(co)):
414 self.scan_code(c, m)
415
Guido van Rossum75dc4961998-03-05 03:42:00 +0000416 def load_package(self, fqname, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000417 self.msgin(2, "load_package", fqname, pathname)
Thomas Hellerc7aaf952002-11-14 18:45:11 +0000418 newname = replacePackageMap.get(fqname)
419 if newname:
420 fqname = newname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000421 m = self.add_module(fqname)
422 m.__file__ = pathname
423 m.__path__ = [pathname]
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000424
Guido van Rossume7e632a1998-09-14 16:02:28 +0000425 # As per comment at top of file, simulate runtime __path__ additions.
426 m.__path__ = m.__path__ + packagePathMap.get(fqname, [])
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000427
Guido van Rossum912a14c1998-03-05 04:56:37 +0000428 fp, buf, stuff = self.find_module("__init__", m.__path__)
Brett Cannon028011f2010-10-30 00:26:48 +0000429 try:
430 self.load_module(fqname, fp, buf, stuff)
431 self.msgout(2, "load_package ->", m)
432 return m
433 finally:
434 if fp:
435 fp.close()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000436
437 def add_module(self, fqname):
Guido van Rossume2b70bc2006-08-18 22:13:04 +0000438 if fqname in self.modules:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000439 return self.modules[fqname]
440 self.modules[fqname] = m = Module(fqname)
441 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000442
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000443 def find_module(self, name, path, parent=None):
444 if parent is not None:
Thomas Heller2e7c8322004-05-11 15:10:59 +0000445 # assert path is not None
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000446 fullname = parent.__name__+'.'+name
Guido van Rossum03f7f082001-10-18 19:15:32 +0000447 else:
448 fullname = name
449 if fullname in self.excludes:
450 self.msgout(3, "find_module -> Excluded", fullname)
Collin Winterce36ad82007-08-30 01:19:48 +0000451 raise ImportError(name)
Guido van Rossum78fc3631998-03-20 17:37:24 +0000452
Guido van Rossum912a14c1998-03-05 04:56:37 +0000453 if path is None:
454 if name in sys.builtin_module_names:
455 return (None, None, ("", "", imp.C_BUILTIN))
Guido van Rossum78fc3631998-03-20 17:37:24 +0000456
Guido van Rossum912a14c1998-03-05 04:56:37 +0000457 path = self.path
458 return imp.find_module(name, path)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000459
460 def report(self):
Just van Rossume29310a2002-12-31 16:33:00 +0000461 """Print a report to stdout, listing the found modules with their
462 paths, as well as modules that are missing, or seem to be missing.
463 """
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000464 print()
465 print(" %-25s %s" % ("Name", "File"))
466 print(" %-25s %s" % ("----", "----"))
Guido van Rossum912a14c1998-03-05 04:56:37 +0000467 # Print modules found
Guido van Rossumd59cde82007-06-12 00:25:08 +0000468 keys = sorted(self.modules.keys())
Guido van Rossum912a14c1998-03-05 04:56:37 +0000469 for key in keys:
470 m = self.modules[key]
471 if m.__path__:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000472 print("P", end=' ')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000473 else:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000474 print("m", end=' ')
475 print("%-25s" % key, m.__file__ or "")
Guido van Rossum75dc4961998-03-05 03:42:00 +0000476
Guido van Rossum912a14c1998-03-05 04:56:37 +0000477 # Print missing modules
Just van Rossume29310a2002-12-31 16:33:00 +0000478 missing, maybe = self.any_missing_maybe()
479 if missing:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000480 print()
481 print("Missing modules:")
Just van Rossume29310a2002-12-31 16:33:00 +0000482 for name in missing:
Guido van Rossumd59cde82007-06-12 00:25:08 +0000483 mods = sorted(self.badmodules[name].keys())
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000484 print("?", name, "imported from", ', '.join(mods))
Just van Rossume29310a2002-12-31 16:33:00 +0000485 # Print modules that may be missing, but then again, maybe not...
486 if maybe:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000487 print()
Ezio Melotti30b9d5d2013-08-17 15:50:46 +0300488 print("Submodules that appear to be missing, but could also be", end=' ')
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000489 print("global names in the parent package:")
Just van Rossume29310a2002-12-31 16:33:00 +0000490 for name in maybe:
Guido van Rossumd59cde82007-06-12 00:25:08 +0000491 mods = sorted(self.badmodules[name].keys())
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000492 print("?", name, "imported from", ', '.join(mods))
Guido van Rossum75dc4961998-03-05 03:42:00 +0000493
Guido van Rossum03f7f082001-10-18 19:15:32 +0000494 def any_missing(self):
Just van Rossume29310a2002-12-31 16:33:00 +0000495 """Return a list of modules that appear to be missing. Use
496 any_missing_maybe() if you want to know which modules are
497 certain to be missing, and which *may* be missing.
498 """
499 missing, maybe = self.any_missing_maybe()
500 return missing + maybe
501
502 def any_missing_maybe(self):
503 """Return two lists, one with modules that are certainly missing
504 and one with modules that *may* be missing. The latter names could
505 either be submodules *or* just global names in the package.
506
507 The reason it can't always be determined is that it's impossible to
508 tell which names are imported when "from module import *" is done
509 with an extension module, short of actually importing it.
510 """
Guido van Rossum03f7f082001-10-18 19:15:32 +0000511 missing = []
Just van Rossume29310a2002-12-31 16:33:00 +0000512 maybe = []
513 for name in self.badmodules:
514 if name in self.excludes:
515 continue
516 i = name.rfind(".")
517 if i < 0:
518 missing.append(name)
519 continue
520 subname = name[i+1:]
521 pkgname = name[:i]
522 pkg = self.modules.get(pkgname)
523 if pkg is not None:
524 if pkgname in self.badmodules[name]:
525 # The package tried to import this module itself and
526 # failed. It's definitely missing.
527 missing.append(name)
528 elif subname in pkg.globalnames:
529 # It's a global in the package: definitely not missing.
530 pass
531 elif pkg.starimports:
532 # It could be missing, but the package did an "import *"
533 # from a non-Python module, so we simply can't be sure.
534 maybe.append(name)
535 else:
536 # It's not a global in the package, the package didn't
537 # do funny star imports, it's very likely to be missing.
538 # The symbol could be inserted into the package from the
539 # outside, but since that's not good style we simply list
540 # it missing.
541 missing.append(name)
542 else:
543 missing.append(name)
544 missing.sort()
545 maybe.sort()
546 return missing, maybe
Guido van Rossum03f7f082001-10-18 19:15:32 +0000547
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000548 def replace_paths_in_code(self, co):
549 new_filename = original_filename = os.path.normpath(co.co_filename)
Just van Rossume29310a2002-12-31 16:33:00 +0000550 for f, r in self.replace_paths:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000551 if original_filename.startswith(f):
Just van Rossume29310a2002-12-31 16:33:00 +0000552 new_filename = r + original_filename[len(f):]
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000553 break
554
555 if self.debug and original_filename not in self.processed_paths:
Just van Rossume29310a2002-12-31 16:33:00 +0000556 if new_filename != original_filename:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000557 self.msgout(2, "co_filename %r changed to %r" \
558 % (original_filename,new_filename,))
559 else:
560 self.msgout(2, "co_filename %r remains unchanged" \
561 % (original_filename,))
562 self.processed_paths.append(original_filename)
563
564 consts = list(co.co_consts)
565 for i in range(len(consts)):
566 if isinstance(consts[i], type(co)):
567 consts[i] = self.replace_paths_in_code(consts[i])
568
Christian Heimes45f9af32007-11-27 21:50:00 +0000569 return types.CodeType(co.co_argcount, co.co_nlocals, co.co_stacksize,
Tim Peters2c60f7a2003-01-29 03:49:43 +0000570 co.co_flags, co.co_code, tuple(consts), co.co_names,
571 co.co_varnames, new_filename, co.co_name,
Neal Norwitz82c72312002-11-12 23:09:12 +0000572 co.co_firstlineno, co.co_lnotab,
573 co.co_freevars, co.co_cellvars)
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000574
Guido van Rossum75dc4961998-03-05 03:42:00 +0000575
576def test():
577 # Parse command line
578 import getopt
579 try:
Guido van Rossumbaf06031998-08-25 14:06:55 +0000580 opts, args = getopt.getopt(sys.argv[1:], "dmp:qx:")
Guido van Rossumb940e112007-01-10 16:19:56 +0000581 except getopt.error as msg:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000582 print(msg)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000583 return
Guido van Rossum75dc4961998-03-05 03:42:00 +0000584
585 # Process options
586 debug = 1
587 domods = 0
588 addpath = []
Guido van Rossumbaf06031998-08-25 14:06:55 +0000589 exclude = []
Guido van Rossum75dc4961998-03-05 03:42:00 +0000590 for o, a in opts:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000591 if o == '-d':
592 debug = debug + 1
593 if o == '-m':
594 domods = 1
595 if o == '-p':
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000596 addpath = addpath + a.split(os.pathsep)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000597 if o == '-q':
598 debug = 0
Guido van Rossumbaf06031998-08-25 14:06:55 +0000599 if o == '-x':
600 exclude.append(a)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000601
602 # Provide default arguments
603 if not args:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000604 script = "hello.py"
Guido van Rossum75dc4961998-03-05 03:42:00 +0000605 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000606 script = args[0]
Guido van Rossum75dc4961998-03-05 03:42:00 +0000607
608 # Set the path based on sys.path and the script directory
609 path = sys.path[:]
610 path[0] = os.path.dirname(script)
611 path = addpath + path
612 if debug > 1:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000613 print("path:")
Guido van Rossum912a14c1998-03-05 04:56:37 +0000614 for item in path:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000615 print(" ", repr(item))
Guido van Rossum75dc4961998-03-05 03:42:00 +0000616
617 # Create the module finder and turn its crank
Guido van Rossumbaf06031998-08-25 14:06:55 +0000618 mf = ModuleFinder(path, debug, exclude)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000619 for arg in args[1:]:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000620 if arg == '-m':
621 domods = 1
622 continue
623 if domods:
624 if arg[-2:] == '.*':
625 mf.import_hook(arg[:-2], None, ["*"])
626 else:
627 mf.import_hook(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000628 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000629 mf.load_file(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000630 mf.run_script(script)
631 mf.report()
Just van Rossume29310a2002-12-31 16:33:00 +0000632 return mf # for -i debugging
Guido van Rossum75dc4961998-03-05 03:42:00 +0000633
634
635if __name__ == '__main__':
636 try:
Just van Rossume29310a2002-12-31 16:33:00 +0000637 mf = test()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000638 except KeyboardInterrupt:
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200639 print("\n[interrupted]")