blob: da5b56d84c72705c2ff2e68d5f6765703ead6300 [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
4import imp
5import 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
Guido van Rossum75dc4961998-03-05 03:42:00 +000010
Brett Cannon40430012007-10-22 20:24:51 +000011# XXX Clean up once str8's cstor matches bytes.
Guido van Rossum98297ee2007-11-06 21:34:58 +000012LOAD_CONST = bytes([dis.opname.index('LOAD_CONST')])
13IMPORT_NAME = bytes([dis.opname.index('IMPORT_NAME')])
14STORE_NAME = bytes([dis.opname.index('STORE_NAME')])
15STORE_GLOBAL = bytes([dis.opname.index('STORE_GLOBAL')])
Just van Rossume29310a2002-12-31 16:33:00 +000016STORE_OPS = [STORE_NAME, STORE_GLOBAL]
Guido van Rossum98297ee2007-11-06 21:34:58 +000017HAVE_ARGUMENT = bytes([dis.HAVE_ARGUMENT])
Guido van Rossum75dc4961998-03-05 03:42:00 +000018
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000019# Modulefinder does a good job at simulating Python's, but it can not
20# handle __path__ modifications packages make at runtime. Therefore there
21# is a mechanism whereby you can register extra paths in this map for a
Thomas Wouters7e474022000-07-16 12:04:32 +000022# package, and it will be honored.
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000023
24# Note this is a mapping is lists of paths.
25packagePathMap = {}
26
27# A Public interface
28def AddPackagePath(packagename, path):
Éric Araujo1e3a68d2011-07-28 23:35:29 +020029 paths = packagePathMap.setdefault(packagename, []).append(path)
Guido van Rossume7e632a1998-09-14 16:02:28 +000030 packagePathMap[packagename] = paths
Guido van Rossum75dc4961998-03-05 03:42:00 +000031
Thomas Hellerc7aaf952002-11-14 18:45:11 +000032replacePackageMap = {}
33
Martin v. Löwis2f48d892011-05-09 08:05:43 +020034# This ReplacePackage mechanism allows modulefinder to work around
35# situations in which a package injects itself under the name
36# of another package into sys.modules at runtime by calling
37# ReplacePackage("real_package_name", "faked_package_name")
Thomas Hellerc7aaf952002-11-14 18:45:11 +000038# before running ModuleFinder.
39
40def ReplacePackage(oldname, newname):
41 replacePackageMap[oldname] = newname
42
43
Guido van Rossum75dc4961998-03-05 03:42:00 +000044class Module:
45
46 def __init__(self, name, file=None, path=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +000047 self.__name__ = name
48 self.__file__ = file
49 self.__path__ = path
50 self.__code__ = None
Just van Rossume29310a2002-12-31 16:33:00 +000051 # The set of global names that are assigned to in the module.
52 # This includes those names imported through starimports of
53 # Python modules.
54 self.globalnames = {}
55 # The set of starimports this module did that could not be
56 # resolved, ie. a starimport from a non-Python module.
57 self.starimports = {}
Guido van Rossum75dc4961998-03-05 03:42:00 +000058
59 def __repr__(self):
Neil Schemenauer32d23c92004-02-15 16:43:20 +000060 s = "Module(%r" % (self.__name__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000061 if self.__file__ is not None:
Walter Dörwald70a6b492004-02-12 17:35:32 +000062 s = s + ", %r" % (self.__file__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000063 if self.__path__ is not None:
Walter Dörwald70a6b492004-02-12 17:35:32 +000064 s = s + ", %r" % (self.__path__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000065 s = s + ")"
66 return s
Guido van Rossum75dc4961998-03-05 03:42:00 +000067
Guido van Rossum75dc4961998-03-05 03:42:00 +000068class ModuleFinder:
69
Just van Rossume29310a2002-12-31 16:33:00 +000070 def __init__(self, path=None, debug=0, excludes=[], replace_paths=[]):
Guido van Rossum912a14c1998-03-05 04:56:37 +000071 if path is None:
72 path = sys.path
73 self.path = path
74 self.modules = {}
75 self.badmodules = {}
76 self.debug = debug
77 self.indent = 0
Guido van Rossum78fc3631998-03-20 17:37:24 +000078 self.excludes = excludes
Guido van Rossum6b767ac2001-03-20 20:43:34 +000079 self.replace_paths = replace_paths
80 self.processed_paths = [] # Used in debugging only
Guido van Rossum75dc4961998-03-05 03:42:00 +000081
82 def msg(self, level, str, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000083 if level <= self.debug:
84 for i in range(self.indent):
Guido van Rossumbe19ed72007-02-09 05:37:30 +000085 print(" ", end=' ')
86 print(str, end=' ')
Guido van Rossum912a14c1998-03-05 04:56:37 +000087 for arg in args:
Guido van Rossumbe19ed72007-02-09 05:37:30 +000088 print(repr(arg), end=' ')
89 print()
Guido van Rossum75dc4961998-03-05 03:42:00 +000090
91 def msgin(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000092 level = args[0]
93 if level <= self.debug:
94 self.indent = self.indent + 1
Guido van Rossum68468eb2003-02-27 20:14:51 +000095 self.msg(*args)
Guido van Rossum75dc4961998-03-05 03:42:00 +000096
97 def msgout(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000098 level = args[0]
99 if level <= self.debug:
100 self.indent = self.indent - 1
Guido van Rossum68468eb2003-02-27 20:14:51 +0000101 self.msg(*args)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000102
103 def run_script(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000104 self.msg(2, "run_script", pathname)
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200105 with open(pathname) as fp:
Brett Cannon028011f2010-10-30 00:26:48 +0000106 stuff = ("", "r", imp.PY_SOURCE)
107 self.load_module('__main__', fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000108
109 def load_file(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000110 dir, name = os.path.split(pathname)
111 name, ext = os.path.splitext(name)
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200112 with open(pathname) as fp:
Brett Cannon028011f2010-10-30 00:26:48 +0000113 stuff = (ext, "r", imp.PY_SOURCE)
114 self.load_module(name, fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000115
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000116 def import_hook(self, name, caller=None, fromlist=None, level=-1):
117 self.msg(3, "import_hook", name, caller, fromlist, level)
118 parent = self.determine_parent(caller, level=level)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000119 q, tail = self.find_head_package(parent, name)
120 m = self.load_tail(q, tail)
121 if not fromlist:
122 return q
123 if m.__path__:
124 self.ensure_fromlist(m, fromlist)
Thomas Heller318b7b92002-11-26 08:06:50 +0000125 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000126
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000127 def determine_parent(self, caller, level=-1):
128 self.msgin(4, "determine_parent", caller, level)
129 if not caller or level == 0:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000130 self.msgout(4, "determine_parent -> None")
131 return None
132 pname = caller.__name__
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000133 if level >= 1: # relative import
134 if caller.__path__:
135 level -= 1
136 if level == 0:
137 parent = self.modules[pname]
138 assert parent is caller
139 self.msgout(4, "determine_parent ->", parent)
140 return parent
141 if pname.count(".") < level:
Collin Winterce36ad82007-08-30 01:19:48 +0000142 raise ImportError("relative importpath too deep")
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000143 pname = ".".join(pname.split(".")[:-level])
144 parent = self.modules[pname]
145 self.msgout(4, "determine_parent ->", parent)
146 return parent
Guido van Rossum912a14c1998-03-05 04:56:37 +0000147 if caller.__path__:
148 parent = self.modules[pname]
149 assert caller is parent
150 self.msgout(4, "determine_parent ->", parent)
151 return parent
152 if '.' in pname:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000153 i = pname.rfind('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000154 pname = pname[:i]
155 parent = self.modules[pname]
156 assert parent.__name__ == pname
157 self.msgout(4, "determine_parent ->", parent)
158 return parent
159 self.msgout(4, "determine_parent -> None")
160 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000161
162 def find_head_package(self, parent, name):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000163 self.msgin(4, "find_head_package", parent, name)
164 if '.' in name:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000165 i = name.find('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000166 head = name[:i]
167 tail = name[i+1:]
168 else:
169 head = name
170 tail = ""
171 if parent:
172 qname = "%s.%s" % (parent.__name__, head)
173 else:
174 qname = head
175 q = self.import_module(head, qname, parent)
176 if q:
177 self.msgout(4, "find_head_package ->", (q, tail))
178 return q, tail
179 if parent:
180 qname = head
181 parent = None
182 q = self.import_module(head, qname, parent)
183 if q:
184 self.msgout(4, "find_head_package ->", (q, tail))
185 return q, tail
186 self.msgout(4, "raise ImportError: No module named", qname)
Collin Winterce36ad82007-08-30 01:19:48 +0000187 raise ImportError("No module named " + qname)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000188
189 def load_tail(self, q, tail):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000190 self.msgin(4, "load_tail", q, tail)
191 m = q
192 while tail:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000193 i = tail.find('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000194 if i < 0: i = len(tail)
195 head, tail = tail[:i], tail[i+1:]
196 mname = "%s.%s" % (m.__name__, head)
197 m = self.import_module(head, mname, m)
198 if not m:
199 self.msgout(4, "raise ImportError: No module named", mname)
Collin Winterce36ad82007-08-30 01:19:48 +0000200 raise ImportError("No module named " + mname)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000201 self.msgout(4, "load_tail ->", m)
202 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000203
204 def ensure_fromlist(self, m, fromlist, recursive=0):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000205 self.msg(4, "ensure_fromlist", m, fromlist, recursive)
206 for sub in fromlist:
207 if sub == "*":
208 if not recursive:
209 all = self.find_all_submodules(m)
210 if all:
211 self.ensure_fromlist(m, all, 1)
212 elif not hasattr(m, sub):
213 subname = "%s.%s" % (m.__name__, sub)
214 submod = self.import_module(sub, subname, m)
215 if not submod:
Collin Winterce36ad82007-08-30 01:19:48 +0000216 raise ImportError("No module named " + subname)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000217
218 def find_all_submodules(self, m):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000219 if not m.__path__:
220 return
221 modules = {}
Thomas Helleraaf1c8d2003-11-14 10:28:42 +0000222 # 'suffixes' used to be a list hardcoded to [".py", ".pyc", ".pyo"].
223 # But we must also collect Python extension modules - although
224 # we cannot separate normal dlls from Python extensions.
225 suffixes = []
226 for triple in imp.get_suffixes():
227 suffixes.append(triple[0])
Guido van Rossum912a14c1998-03-05 04:56:37 +0000228 for dir in m.__path__:
229 try:
230 names = os.listdir(dir)
231 except os.error:
232 self.msg(2, "can't list directory", dir)
233 continue
234 for name in names:
235 mod = None
236 for suff in suffixes:
237 n = len(suff)
238 if name[-n:] == suff:
239 mod = name[:-n]
240 break
241 if mod and mod != "__init__":
242 modules[mod] = mod
243 return modules.keys()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000244
245 def import_module(self, partname, fqname, parent):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000246 self.msgin(3, "import_module", partname, fqname, parent)
247 try:
248 m = self.modules[fqname]
249 except KeyError:
250 pass
251 else:
252 self.msgout(3, "import_module ->", m)
253 return m
Guido van Rossume2b70bc2006-08-18 22:13:04 +0000254 if fqname in self.badmodules:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000255 self.msgout(3, "import_module -> None")
256 return None
Thomas Heller2e7c8322004-05-11 15:10:59 +0000257 if parent and parent.__path__ is None:
258 self.msgout(3, "import_module -> None")
259 return None
Guido van Rossum912a14c1998-03-05 04:56:37 +0000260 try:
261 fp, pathname, stuff = self.find_module(partname,
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000262 parent and parent.__path__, parent)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000263 except ImportError:
264 self.msgout(3, "import_module ->", None)
265 return None
266 try:
267 m = self.load_module(fqname, fp, pathname, stuff)
268 finally:
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200269 if fp:
270 fp.close()
Guido van Rossum912a14c1998-03-05 04:56:37 +0000271 if parent:
272 setattr(parent, partname, m)
273 self.msgout(3, "import_module ->", m)
274 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000275
Guido van Rossum1bc535d2007-05-15 18:46:22 +0000276 def load_module(self, fqname, fp, pathname, file_info):
277 suffix, mode, type = file_info
Guido van Rossum912a14c1998-03-05 04:56:37 +0000278 self.msgin(2, "load_module", fqname, fp and "fp", pathname)
279 if type == imp.PKG_DIRECTORY:
280 m = self.load_package(fqname, pathname)
281 self.msgout(2, "load_module ->", m)
282 return m
283 if type == imp.PY_SOURCE:
Guido van Rossum78fc3631998-03-20 17:37:24 +0000284 co = compile(fp.read()+'\n', pathname, 'exec')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000285 elif type == imp.PY_COMPILED:
286 if fp.read(4) != imp.get_magic():
287 self.msgout(2, "raise ImportError: Bad magic number", pathname)
Collin Winterce36ad82007-08-30 01:19:48 +0000288 raise ImportError("Bad magic number in %s" % pathname)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000289 fp.read(4)
290 co = marshal.load(fp)
291 else:
292 co = None
293 m = self.add_module(fqname)
Guido van Rossumab045f91998-03-06 19:55:10 +0000294 m.__file__ = pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000295 if co:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000296 if self.replace_paths:
297 co = self.replace_paths_in_code(co)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000298 m.__code__ = co
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000299 self.scan_code(co, m)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000300 self.msgout(2, "load_module ->", m)
301 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000302
Just van Rossume29310a2002-12-31 16:33:00 +0000303 def _add_badmodule(self, name, caller):
304 if name not in self.badmodules:
305 self.badmodules[name] = {}
Benjamin Petersonc0747cf2008-11-03 20:31:38 +0000306 if caller:
307 self.badmodules[name][caller.__name__] = 1
308 else:
309 self.badmodules[name]["-"] = 1
Just van Rossume29310a2002-12-31 16:33:00 +0000310
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000311 def _safe_import_hook(self, name, caller, fromlist, level=-1):
Just van Rossume29310a2002-12-31 16:33:00 +0000312 # wrapper for self.import_hook() that won't raise ImportError
313 if name in self.badmodules:
314 self._add_badmodule(name, caller)
315 return
316 try:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000317 self.import_hook(name, caller, level=level)
Guido van Rossumb940e112007-01-10 16:19:56 +0000318 except ImportError as msg:
Just van Rossume29310a2002-12-31 16:33:00 +0000319 self.msg(2, "ImportError:", str(msg))
320 self._add_badmodule(name, caller)
321 else:
322 if fromlist:
323 for sub in fromlist:
324 if sub in self.badmodules:
325 self._add_badmodule(sub, caller)
326 continue
327 try:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000328 self.import_hook(name, caller, [sub], level=level)
Guido van Rossumb940e112007-01-10 16:19:56 +0000329 except ImportError as msg:
Just van Rossume29310a2002-12-31 16:33:00 +0000330 self.msg(2, "ImportError:", str(msg))
331 fullname = name + "." + sub
332 self._add_badmodule(fullname, caller)
333
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000334 def scan_opcodes(self, co,
335 unpack = struct.unpack):
336 # Scan the code, and yield 'interesting' opcode combinations
337 # Version for Python 2.4 and older
338 code = co.co_code
339 names = co.co_names
340 consts = co.co_consts
341 while code:
342 c = code[0]
343 if c in STORE_OPS:
344 oparg, = unpack('<H', code[1:3])
345 yield "store", (names[oparg],)
346 code = code[3:]
347 continue
348 if c == LOAD_CONST and code[3] == IMPORT_NAME:
349 oparg_1, oparg_2 = unpack('<xHxH', code[:6])
350 yield "import", (consts[oparg_1], names[oparg_2])
351 code = code[6:]
352 continue
353 if c >= HAVE_ARGUMENT:
354 code = code[3:]
355 else:
356 code = code[1:]
357
358 def scan_opcodes_25(self, co,
359 unpack = struct.unpack):
360 # Scan the code, and yield 'interesting' opcode combinations
361 # Python 2.5 version (has absolute and relative imports)
362 code = co.co_code
363 names = co.co_names
364 consts = co.co_consts
365 LOAD_LOAD_AND_IMPORT = LOAD_CONST + LOAD_CONST + IMPORT_NAME
366 while code:
Guido van Rossum98297ee2007-11-06 21:34:58 +0000367 c = bytes([code[0]])
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000368 if c in STORE_OPS:
369 oparg, = unpack('<H', code[1:3])
370 yield "store", (names[oparg],)
371 code = code[3:]
372 continue
373 if code[:9:3] == LOAD_LOAD_AND_IMPORT:
374 oparg_1, oparg_2, oparg_3 = unpack('<xHxHxH', code[:9])
375 level = consts[oparg_1]
376 if level == 0: # absolute import
377 yield "absolute_import", (consts[oparg_2], names[oparg_3])
378 else: # relative import
379 yield "relative_import", (level, consts[oparg_2], names[oparg_3])
380 code = code[9:]
381 continue
382 if c >= HAVE_ARGUMENT:
383 code = code[3:]
384 else:
385 code = code[1:]
386
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000387 def scan_code(self, co, m):
388 code = co.co_code
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000389 if sys.version_info >= (2, 5):
390 scanner = self.scan_opcodes_25
391 else:
392 scanner = self.scan_opcodes
393 for what, args in scanner(co):
394 if what == "store":
395 name, = args
396 m.globalnames[name] = 1
397 elif what == "absolute_import":
398 fromlist, name = args
Just van Rossume29310a2002-12-31 16:33:00 +0000399 have_star = 0
400 if fromlist is not None:
401 if "*" in fromlist:
402 have_star = 1
403 fromlist = [f for f in fromlist if f != "*"]
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000404 self._safe_import_hook(name, m, fromlist, level=0)
Just van Rossume29310a2002-12-31 16:33:00 +0000405 if have_star:
406 # We've encountered an "import *". If it is a Python module,
407 # the code has already been parsed and we can suck out the
408 # global names.
409 mm = None
410 if m.__path__:
411 # At this point we don't know whether 'name' is a
412 # submodule of 'm' or a global module. Let's just try
413 # the full name first.
414 mm = self.modules.get(m.__name__ + "." + name)
415 if mm is None:
416 mm = self.modules.get(name)
417 if mm is not None:
418 m.globalnames.update(mm.globalnames)
419 m.starimports.update(mm.starimports)
420 if mm.__code__ is None:
421 m.starimports[name] = 1
422 else:
423 m.starimports[name] = 1
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000424 elif what == "relative_import":
425 level, fromlist, name = args
426 if name:
427 self._safe_import_hook(name, m, fromlist, level=level)
428 else:
429 parent = self.determine_parent(m, level=level)
430 self._safe_import_hook(parent.__name__, None, fromlist, level=0)
431 else:
432 # We don't expect anything else from the generator.
433 raise RuntimeError(what)
434
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000435 for c in co.co_consts:
436 if isinstance(c, type(co)):
437 self.scan_code(c, m)
438
Guido van Rossum75dc4961998-03-05 03:42:00 +0000439 def load_package(self, fqname, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000440 self.msgin(2, "load_package", fqname, pathname)
Thomas Hellerc7aaf952002-11-14 18:45:11 +0000441 newname = replacePackageMap.get(fqname)
442 if newname:
443 fqname = newname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000444 m = self.add_module(fqname)
445 m.__file__ = pathname
446 m.__path__ = [pathname]
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000447
Guido van Rossume7e632a1998-09-14 16:02:28 +0000448 # As per comment at top of file, simulate runtime __path__ additions.
449 m.__path__ = m.__path__ + packagePathMap.get(fqname, [])
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000450
Guido van Rossum912a14c1998-03-05 04:56:37 +0000451 fp, buf, stuff = self.find_module("__init__", m.__path__)
Brett Cannon028011f2010-10-30 00:26:48 +0000452 try:
453 self.load_module(fqname, fp, buf, stuff)
454 self.msgout(2, "load_package ->", m)
455 return m
456 finally:
457 if fp:
458 fp.close()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000459
460 def add_module(self, fqname):
Guido van Rossume2b70bc2006-08-18 22:13:04 +0000461 if fqname in self.modules:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000462 return self.modules[fqname]
463 self.modules[fqname] = m = Module(fqname)
464 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000465
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000466 def find_module(self, name, path, parent=None):
467 if parent is not None:
Thomas Heller2e7c8322004-05-11 15:10:59 +0000468 # assert path is not None
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000469 fullname = parent.__name__+'.'+name
Guido van Rossum03f7f082001-10-18 19:15:32 +0000470 else:
471 fullname = name
472 if fullname in self.excludes:
473 self.msgout(3, "find_module -> Excluded", fullname)
Collin Winterce36ad82007-08-30 01:19:48 +0000474 raise ImportError(name)
Guido van Rossum78fc3631998-03-20 17:37:24 +0000475
Guido van Rossum912a14c1998-03-05 04:56:37 +0000476 if path is None:
477 if name in sys.builtin_module_names:
478 return (None, None, ("", "", imp.C_BUILTIN))
Guido van Rossum78fc3631998-03-20 17:37:24 +0000479
Guido van Rossum912a14c1998-03-05 04:56:37 +0000480 path = self.path
481 return imp.find_module(name, path)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000482
483 def report(self):
Just van Rossume29310a2002-12-31 16:33:00 +0000484 """Print a report to stdout, listing the found modules with their
485 paths, as well as modules that are missing, or seem to be missing.
486 """
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000487 print()
488 print(" %-25s %s" % ("Name", "File"))
489 print(" %-25s %s" % ("----", "----"))
Guido van Rossum912a14c1998-03-05 04:56:37 +0000490 # Print modules found
Guido van Rossumd59cde82007-06-12 00:25:08 +0000491 keys = sorted(self.modules.keys())
Guido van Rossum912a14c1998-03-05 04:56:37 +0000492 for key in keys:
493 m = self.modules[key]
494 if m.__path__:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000495 print("P", end=' ')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000496 else:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000497 print("m", end=' ')
498 print("%-25s" % key, m.__file__ or "")
Guido van Rossum75dc4961998-03-05 03:42:00 +0000499
Guido van Rossum912a14c1998-03-05 04:56:37 +0000500 # Print missing modules
Just van Rossume29310a2002-12-31 16:33:00 +0000501 missing, maybe = self.any_missing_maybe()
502 if missing:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000503 print()
504 print("Missing modules:")
Just van Rossume29310a2002-12-31 16:33:00 +0000505 for name in missing:
Guido van Rossumd59cde82007-06-12 00:25:08 +0000506 mods = sorted(self.badmodules[name].keys())
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000507 print("?", name, "imported from", ', '.join(mods))
Just van Rossume29310a2002-12-31 16:33:00 +0000508 # Print modules that may be missing, but then again, maybe not...
509 if maybe:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000510 print()
511 print("Submodules thay appear to be missing, but could also be", end=' ')
512 print("global names in the parent package:")
Just van Rossume29310a2002-12-31 16:33:00 +0000513 for name in maybe:
Guido van Rossumd59cde82007-06-12 00:25:08 +0000514 mods = sorted(self.badmodules[name].keys())
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000515 print("?", name, "imported from", ', '.join(mods))
Guido van Rossum75dc4961998-03-05 03:42:00 +0000516
Guido van Rossum03f7f082001-10-18 19:15:32 +0000517 def any_missing(self):
Just van Rossume29310a2002-12-31 16:33:00 +0000518 """Return a list of modules that appear to be missing. Use
519 any_missing_maybe() if you want to know which modules are
520 certain to be missing, and which *may* be missing.
521 """
522 missing, maybe = self.any_missing_maybe()
523 return missing + maybe
524
525 def any_missing_maybe(self):
526 """Return two lists, one with modules that are certainly missing
527 and one with modules that *may* be missing. The latter names could
528 either be submodules *or* just global names in the package.
529
530 The reason it can't always be determined is that it's impossible to
531 tell which names are imported when "from module import *" is done
532 with an extension module, short of actually importing it.
533 """
Guido van Rossum03f7f082001-10-18 19:15:32 +0000534 missing = []
Just van Rossume29310a2002-12-31 16:33:00 +0000535 maybe = []
536 for name in self.badmodules:
537 if name in self.excludes:
538 continue
539 i = name.rfind(".")
540 if i < 0:
541 missing.append(name)
542 continue
543 subname = name[i+1:]
544 pkgname = name[:i]
545 pkg = self.modules.get(pkgname)
546 if pkg is not None:
547 if pkgname in self.badmodules[name]:
548 # The package tried to import this module itself and
549 # failed. It's definitely missing.
550 missing.append(name)
551 elif subname in pkg.globalnames:
552 # It's a global in the package: definitely not missing.
553 pass
554 elif pkg.starimports:
555 # It could be missing, but the package did an "import *"
556 # from a non-Python module, so we simply can't be sure.
557 maybe.append(name)
558 else:
559 # It's not a global in the package, the package didn't
560 # do funny star imports, it's very likely to be missing.
561 # The symbol could be inserted into the package from the
562 # outside, but since that's not good style we simply list
563 # it missing.
564 missing.append(name)
565 else:
566 missing.append(name)
567 missing.sort()
568 maybe.sort()
569 return missing, maybe
Guido van Rossum03f7f082001-10-18 19:15:32 +0000570
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000571 def replace_paths_in_code(self, co):
572 new_filename = original_filename = os.path.normpath(co.co_filename)
Just van Rossume29310a2002-12-31 16:33:00 +0000573 for f, r in self.replace_paths:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000574 if original_filename.startswith(f):
Just van Rossume29310a2002-12-31 16:33:00 +0000575 new_filename = r + original_filename[len(f):]
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000576 break
577
578 if self.debug and original_filename not in self.processed_paths:
Just van Rossume29310a2002-12-31 16:33:00 +0000579 if new_filename != original_filename:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000580 self.msgout(2, "co_filename %r changed to %r" \
581 % (original_filename,new_filename,))
582 else:
583 self.msgout(2, "co_filename %r remains unchanged" \
584 % (original_filename,))
585 self.processed_paths.append(original_filename)
586
587 consts = list(co.co_consts)
588 for i in range(len(consts)):
589 if isinstance(consts[i], type(co)):
590 consts[i] = self.replace_paths_in_code(consts[i])
591
Christian Heimes45f9af32007-11-27 21:50:00 +0000592 return types.CodeType(co.co_argcount, co.co_nlocals, co.co_stacksize,
Tim Peters2c60f7a2003-01-29 03:49:43 +0000593 co.co_flags, co.co_code, tuple(consts), co.co_names,
594 co.co_varnames, new_filename, co.co_name,
Neal Norwitz82c72312002-11-12 23:09:12 +0000595 co.co_firstlineno, co.co_lnotab,
596 co.co_freevars, co.co_cellvars)
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000597
Guido van Rossum75dc4961998-03-05 03:42:00 +0000598
599def test():
600 # Parse command line
601 import getopt
602 try:
Guido van Rossumbaf06031998-08-25 14:06:55 +0000603 opts, args = getopt.getopt(sys.argv[1:], "dmp:qx:")
Guido van Rossumb940e112007-01-10 16:19:56 +0000604 except getopt.error as msg:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000605 print(msg)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000606 return
Guido van Rossum75dc4961998-03-05 03:42:00 +0000607
608 # Process options
609 debug = 1
610 domods = 0
611 addpath = []
Guido van Rossumbaf06031998-08-25 14:06:55 +0000612 exclude = []
Guido van Rossum75dc4961998-03-05 03:42:00 +0000613 for o, a in opts:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000614 if o == '-d':
615 debug = debug + 1
616 if o == '-m':
617 domods = 1
618 if o == '-p':
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000619 addpath = addpath + a.split(os.pathsep)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000620 if o == '-q':
621 debug = 0
Guido van Rossumbaf06031998-08-25 14:06:55 +0000622 if o == '-x':
623 exclude.append(a)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000624
625 # Provide default arguments
626 if not args:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000627 script = "hello.py"
Guido van Rossum75dc4961998-03-05 03:42:00 +0000628 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000629 script = args[0]
Guido van Rossum75dc4961998-03-05 03:42:00 +0000630
631 # Set the path based on sys.path and the script directory
632 path = sys.path[:]
633 path[0] = os.path.dirname(script)
634 path = addpath + path
635 if debug > 1:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000636 print("path:")
Guido van Rossum912a14c1998-03-05 04:56:37 +0000637 for item in path:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000638 print(" ", repr(item))
Guido van Rossum75dc4961998-03-05 03:42:00 +0000639
640 # Create the module finder and turn its crank
Guido van Rossumbaf06031998-08-25 14:06:55 +0000641 mf = ModuleFinder(path, debug, exclude)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000642 for arg in args[1:]:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000643 if arg == '-m':
644 domods = 1
645 continue
646 if domods:
647 if arg[-2:] == '.*':
648 mf.import_hook(arg[:-2], None, ["*"])
649 else:
650 mf.import_hook(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000651 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000652 mf.load_file(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000653 mf.run_script(script)
654 mf.report()
Just van Rossume29310a2002-12-31 16:33:00 +0000655 return mf # for -i debugging
Guido van Rossum75dc4961998-03-05 03:42:00 +0000656
657
658if __name__ == '__main__':
659 try:
Just van Rossume29310a2002-12-31 16:33:00 +0000660 mf = test()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000661 except KeyboardInterrupt:
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200662 print("\n[interrupted]")