blob: f033ba98efb3038810452ebf30b0c05f23194db1 [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 Rossumfc2a0a82006-10-27 23:06:01 +00003from __future__ import generators
Guido van Rossum75dc4961998-03-05 03:42:00 +00004import dis
5import imp
6import marshal
7import os
Guido van Rossum75dc4961998-03-05 03:42:00 +00008import sys
Christian Heimes45f9af32007-11-27 21:50:00 +00009import types
Guido van Rossumfc2a0a82006-10-27 23:06:01 +000010import struct
Guido van Rossum75dc4961998-03-05 03:42:00 +000011
Benjamin Peterson821d0f82009-02-16 16:20:10 +000012READ_MODE = "rU"
Just van Rossum5d0bd1e2002-11-26 09:53:16 +000013
Brett Cannon40430012007-10-22 20:24:51 +000014# XXX Clean up once str8's cstor matches bytes.
Guido van Rossum98297ee2007-11-06 21:34:58 +000015LOAD_CONST = bytes([dis.opname.index('LOAD_CONST')])
16IMPORT_NAME = bytes([dis.opname.index('IMPORT_NAME')])
17STORE_NAME = bytes([dis.opname.index('STORE_NAME')])
18STORE_GLOBAL = bytes([dis.opname.index('STORE_GLOBAL')])
Just van Rossume29310a2002-12-31 16:33:00 +000019STORE_OPS = [STORE_NAME, STORE_GLOBAL]
Guido van Rossum98297ee2007-11-06 21:34:58 +000020HAVE_ARGUMENT = bytes([dis.HAVE_ARGUMENT])
Guido van Rossum75dc4961998-03-05 03:42:00 +000021
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000022# Modulefinder does a good job at simulating Python's, but it can not
23# handle __path__ modifications packages make at runtime. Therefore there
24# is a mechanism whereby you can register extra paths in this map for a
Thomas Wouters7e474022000-07-16 12:04:32 +000025# package, and it will be honored.
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000026
27# Note this is a mapping is lists of paths.
28packagePathMap = {}
29
30# A Public interface
31def AddPackagePath(packagename, path):
Guido van Rossume7e632a1998-09-14 16:02:28 +000032 paths = packagePathMap.get(packagename, [])
33 paths.append(path)
34 packagePathMap[packagename] = paths
Guido van Rossum75dc4961998-03-05 03:42:00 +000035
Thomas Hellerc7aaf952002-11-14 18:45:11 +000036replacePackageMap = {}
37
Martin v. Löwis2f48d892011-05-09 08:05:43 +020038# This ReplacePackage mechanism allows modulefinder to work around
39# situations in which a package injects itself under the name
40# of another package into sys.modules at runtime by calling
41# ReplacePackage("real_package_name", "faked_package_name")
Thomas Hellerc7aaf952002-11-14 18:45:11 +000042# before running ModuleFinder.
43
44def ReplacePackage(oldname, newname):
45 replacePackageMap[oldname] = newname
46
47
Guido van Rossum75dc4961998-03-05 03:42:00 +000048class Module:
49
50 def __init__(self, name, file=None, path=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +000051 self.__name__ = name
52 self.__file__ = file
53 self.__path__ = path
54 self.__code__ = None
Just van Rossume29310a2002-12-31 16:33:00 +000055 # The set of global names that are assigned to in the module.
56 # This includes those names imported through starimports of
57 # Python modules.
58 self.globalnames = {}
59 # The set of starimports this module did that could not be
60 # resolved, ie. a starimport from a non-Python module.
61 self.starimports = {}
Guido van Rossum75dc4961998-03-05 03:42:00 +000062
63 def __repr__(self):
Neil Schemenauer32d23c92004-02-15 16:43:20 +000064 s = "Module(%r" % (self.__name__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000065 if self.__file__ is not None:
Walter Dörwald70a6b492004-02-12 17:35:32 +000066 s = s + ", %r" % (self.__file__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000067 if self.__path__ is not None:
Walter Dörwald70a6b492004-02-12 17:35:32 +000068 s = s + ", %r" % (self.__path__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000069 s = s + ")"
70 return s
Guido van Rossum75dc4961998-03-05 03:42:00 +000071
Guido van Rossum75dc4961998-03-05 03:42:00 +000072class ModuleFinder:
73
Just van Rossume29310a2002-12-31 16:33:00 +000074 def __init__(self, path=None, debug=0, excludes=[], replace_paths=[]):
Guido van Rossum912a14c1998-03-05 04:56:37 +000075 if path is None:
76 path = sys.path
77 self.path = path
78 self.modules = {}
79 self.badmodules = {}
80 self.debug = debug
81 self.indent = 0
Guido van Rossum78fc3631998-03-20 17:37:24 +000082 self.excludes = excludes
Guido van Rossum6b767ac2001-03-20 20:43:34 +000083 self.replace_paths = replace_paths
84 self.processed_paths = [] # Used in debugging only
Guido van Rossum75dc4961998-03-05 03:42:00 +000085
86 def msg(self, level, str, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000087 if level <= self.debug:
88 for i in range(self.indent):
Guido van Rossumbe19ed72007-02-09 05:37:30 +000089 print(" ", end=' ')
90 print(str, end=' ')
Guido van Rossum912a14c1998-03-05 04:56:37 +000091 for arg in args:
Guido van Rossumbe19ed72007-02-09 05:37:30 +000092 print(repr(arg), end=' ')
93 print()
Guido van Rossum75dc4961998-03-05 03:42:00 +000094
95 def msgin(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000096 level = args[0]
97 if level <= self.debug:
98 self.indent = self.indent + 1
Guido van Rossum68468eb2003-02-27 20:14:51 +000099 self.msg(*args)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000100
101 def msgout(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000102 level = args[0]
103 if level <= self.debug:
104 self.indent = self.indent - 1
Guido van Rossum68468eb2003-02-27 20:14:51 +0000105 self.msg(*args)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000106
107 def run_script(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000108 self.msg(2, "run_script", pathname)
Brett Cannon028011f2010-10-30 00:26:48 +0000109 with open(pathname, READ_MODE) as fp:
110 stuff = ("", "r", imp.PY_SOURCE)
111 self.load_module('__main__', fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000112
113 def load_file(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000114 dir, name = os.path.split(pathname)
115 name, ext = os.path.splitext(name)
Brett Cannon028011f2010-10-30 00:26:48 +0000116 with open(pathname, READ_MODE) as fp:
117 stuff = (ext, "r", imp.PY_SOURCE)
118 self.load_module(name, fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000119
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000120 def import_hook(self, name, caller=None, fromlist=None, level=-1):
121 self.msg(3, "import_hook", name, caller, fromlist, level)
122 parent = self.determine_parent(caller, level=level)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000123 q, tail = self.find_head_package(parent, name)
124 m = self.load_tail(q, tail)
125 if not fromlist:
126 return q
127 if m.__path__:
128 self.ensure_fromlist(m, fromlist)
Thomas Heller318b7b92002-11-26 08:06:50 +0000129 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000130
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000131 def determine_parent(self, caller, level=-1):
132 self.msgin(4, "determine_parent", caller, level)
133 if not caller or level == 0:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000134 self.msgout(4, "determine_parent -> None")
135 return None
136 pname = caller.__name__
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000137 if level >= 1: # relative import
138 if caller.__path__:
139 level -= 1
140 if level == 0:
141 parent = self.modules[pname]
142 assert parent is caller
143 self.msgout(4, "determine_parent ->", parent)
144 return parent
145 if pname.count(".") < level:
Collin Winterce36ad82007-08-30 01:19:48 +0000146 raise ImportError("relative importpath too deep")
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000147 pname = ".".join(pname.split(".")[:-level])
148 parent = self.modules[pname]
149 self.msgout(4, "determine_parent ->", parent)
150 return parent
Guido van Rossum912a14c1998-03-05 04:56:37 +0000151 if caller.__path__:
152 parent = self.modules[pname]
153 assert caller is parent
154 self.msgout(4, "determine_parent ->", parent)
155 return parent
156 if '.' in pname:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000157 i = pname.rfind('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000158 pname = pname[:i]
159 parent = self.modules[pname]
160 assert parent.__name__ == pname
161 self.msgout(4, "determine_parent ->", parent)
162 return parent
163 self.msgout(4, "determine_parent -> None")
164 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000165
166 def find_head_package(self, parent, name):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000167 self.msgin(4, "find_head_package", parent, name)
168 if '.' in name:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000169 i = name.find('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000170 head = name[:i]
171 tail = name[i+1:]
172 else:
173 head = name
174 tail = ""
175 if parent:
176 qname = "%s.%s" % (parent.__name__, head)
177 else:
178 qname = head
179 q = self.import_module(head, qname, parent)
180 if q:
181 self.msgout(4, "find_head_package ->", (q, tail))
182 return q, tail
183 if parent:
184 qname = head
185 parent = None
186 q = self.import_module(head, qname, parent)
187 if q:
188 self.msgout(4, "find_head_package ->", (q, tail))
189 return q, tail
190 self.msgout(4, "raise ImportError: No module named", qname)
Collin Winterce36ad82007-08-30 01:19:48 +0000191 raise ImportError("No module named " + qname)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000192
193 def load_tail(self, q, tail):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000194 self.msgin(4, "load_tail", q, tail)
195 m = q
196 while tail:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000197 i = tail.find('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000198 if i < 0: i = len(tail)
199 head, tail = tail[:i], tail[i+1:]
200 mname = "%s.%s" % (m.__name__, head)
201 m = self.import_module(head, mname, m)
202 if not m:
203 self.msgout(4, "raise ImportError: No module named", mname)
Collin Winterce36ad82007-08-30 01:19:48 +0000204 raise ImportError("No module named " + mname)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000205 self.msgout(4, "load_tail ->", m)
206 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000207
208 def ensure_fromlist(self, m, fromlist, recursive=0):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000209 self.msg(4, "ensure_fromlist", m, fromlist, recursive)
210 for sub in fromlist:
211 if sub == "*":
212 if not recursive:
213 all = self.find_all_submodules(m)
214 if all:
215 self.ensure_fromlist(m, all, 1)
216 elif not hasattr(m, sub):
217 subname = "%s.%s" % (m.__name__, sub)
218 submod = self.import_module(sub, subname, m)
219 if not submod:
Collin Winterce36ad82007-08-30 01:19:48 +0000220 raise ImportError("No module named " + subname)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000221
222 def find_all_submodules(self, m):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000223 if not m.__path__:
224 return
225 modules = {}
Thomas Helleraaf1c8d2003-11-14 10:28:42 +0000226 # 'suffixes' used to be a list hardcoded to [".py", ".pyc", ".pyo"].
227 # But we must also collect Python extension modules - although
228 # we cannot separate normal dlls from Python extensions.
229 suffixes = []
230 for triple in imp.get_suffixes():
231 suffixes.append(triple[0])
Guido van Rossum912a14c1998-03-05 04:56:37 +0000232 for dir in m.__path__:
233 try:
234 names = os.listdir(dir)
235 except os.error:
236 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:
273 if fp: fp.close()
274 if parent:
275 setattr(parent, partname, m)
276 self.msgout(3, "import_module ->", m)
277 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000278
Guido van Rossum1bc535d2007-05-15 18:46:22 +0000279 def load_module(self, fqname, fp, pathname, file_info):
280 suffix, mode, type = file_info
Guido van Rossum912a14c1998-03-05 04:56:37 +0000281 self.msgin(2, "load_module", fqname, fp and "fp", pathname)
282 if type == imp.PKG_DIRECTORY:
283 m = self.load_package(fqname, pathname)
284 self.msgout(2, "load_module ->", m)
285 return m
286 if type == imp.PY_SOURCE:
Guido van Rossum78fc3631998-03-20 17:37:24 +0000287 co = compile(fp.read()+'\n', pathname, 'exec')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000288 elif type == imp.PY_COMPILED:
289 if fp.read(4) != imp.get_magic():
290 self.msgout(2, "raise ImportError: Bad magic number", pathname)
Collin Winterce36ad82007-08-30 01:19:48 +0000291 raise ImportError("Bad magic number in %s" % pathname)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000292 fp.read(4)
293 co = marshal.load(fp)
294 else:
295 co = None
296 m = self.add_module(fqname)
Guido van Rossumab045f91998-03-06 19:55:10 +0000297 m.__file__ = pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000298 if co:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000299 if self.replace_paths:
300 co = self.replace_paths_in_code(co)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000301 m.__code__ = co
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000302 self.scan_code(co, m)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000303 self.msgout(2, "load_module ->", m)
304 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000305
Just van Rossume29310a2002-12-31 16:33:00 +0000306 def _add_badmodule(self, name, caller):
307 if name not in self.badmodules:
308 self.badmodules[name] = {}
Benjamin Petersonc0747cf2008-11-03 20:31:38 +0000309 if caller:
310 self.badmodules[name][caller.__name__] = 1
311 else:
312 self.badmodules[name]["-"] = 1
Just van Rossume29310a2002-12-31 16:33:00 +0000313
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000314 def _safe_import_hook(self, name, caller, fromlist, level=-1):
Just van Rossume29310a2002-12-31 16:33:00 +0000315 # wrapper for self.import_hook() that won't raise ImportError
316 if name in self.badmodules:
317 self._add_badmodule(name, caller)
318 return
319 try:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000320 self.import_hook(name, caller, level=level)
Guido van Rossumb940e112007-01-10 16:19:56 +0000321 except ImportError as msg:
Just van Rossume29310a2002-12-31 16:33:00 +0000322 self.msg(2, "ImportError:", str(msg))
323 self._add_badmodule(name, caller)
324 else:
325 if fromlist:
326 for sub in fromlist:
327 if sub in self.badmodules:
328 self._add_badmodule(sub, caller)
329 continue
330 try:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000331 self.import_hook(name, caller, [sub], level=level)
Guido van Rossumb940e112007-01-10 16:19:56 +0000332 except ImportError as msg:
Just van Rossume29310a2002-12-31 16:33:00 +0000333 self.msg(2, "ImportError:", str(msg))
334 fullname = name + "." + sub
335 self._add_badmodule(fullname, caller)
336
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000337 def scan_opcodes(self, co,
338 unpack = struct.unpack):
339 # Scan the code, and yield 'interesting' opcode combinations
340 # Version for Python 2.4 and older
341 code = co.co_code
342 names = co.co_names
343 consts = co.co_consts
344 while code:
345 c = code[0]
346 if c in STORE_OPS:
347 oparg, = unpack('<H', code[1:3])
348 yield "store", (names[oparg],)
349 code = code[3:]
350 continue
351 if c == LOAD_CONST and code[3] == IMPORT_NAME:
352 oparg_1, oparg_2 = unpack('<xHxH', code[:6])
353 yield "import", (consts[oparg_1], names[oparg_2])
354 code = code[6:]
355 continue
356 if c >= HAVE_ARGUMENT:
357 code = code[3:]
358 else:
359 code = code[1:]
360
361 def scan_opcodes_25(self, co,
362 unpack = struct.unpack):
363 # Scan the code, and yield 'interesting' opcode combinations
364 # Python 2.5 version (has absolute and relative imports)
365 code = co.co_code
366 names = co.co_names
367 consts = co.co_consts
368 LOAD_LOAD_AND_IMPORT = LOAD_CONST + LOAD_CONST + IMPORT_NAME
369 while code:
Guido van Rossum98297ee2007-11-06 21:34:58 +0000370 c = bytes([code[0]])
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000371 if c in STORE_OPS:
372 oparg, = unpack('<H', code[1:3])
373 yield "store", (names[oparg],)
374 code = code[3:]
375 continue
376 if code[:9:3] == LOAD_LOAD_AND_IMPORT:
377 oparg_1, oparg_2, oparg_3 = unpack('<xHxHxH', code[:9])
378 level = consts[oparg_1]
379 if level == 0: # absolute import
380 yield "absolute_import", (consts[oparg_2], names[oparg_3])
381 else: # relative import
382 yield "relative_import", (level, consts[oparg_2], names[oparg_3])
383 code = code[9:]
384 continue
385 if c >= HAVE_ARGUMENT:
386 code = code[3:]
387 else:
388 code = code[1:]
389
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000390 def scan_code(self, co, m):
391 code = co.co_code
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000392 if sys.version_info >= (2, 5):
393 scanner = self.scan_opcodes_25
394 else:
395 scanner = self.scan_opcodes
396 for what, args in scanner(co):
397 if what == "store":
398 name, = args
399 m.globalnames[name] = 1
400 elif what == "absolute_import":
401 fromlist, name = args
Just van Rossume29310a2002-12-31 16:33:00 +0000402 have_star = 0
403 if fromlist is not None:
404 if "*" in fromlist:
405 have_star = 1
406 fromlist = [f for f in fromlist if f != "*"]
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000407 self._safe_import_hook(name, m, fromlist, level=0)
Just van Rossume29310a2002-12-31 16:33:00 +0000408 if have_star:
409 # We've encountered an "import *". If it is a Python module,
410 # the code has already been parsed and we can suck out the
411 # global names.
412 mm = None
413 if m.__path__:
414 # At this point we don't know whether 'name' is a
415 # submodule of 'm' or a global module. Let's just try
416 # the full name first.
417 mm = self.modules.get(m.__name__ + "." + name)
418 if mm is None:
419 mm = self.modules.get(name)
420 if mm is not None:
421 m.globalnames.update(mm.globalnames)
422 m.starimports.update(mm.starimports)
423 if mm.__code__ is None:
424 m.starimports[name] = 1
425 else:
426 m.starimports[name] = 1
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000427 elif what == "relative_import":
428 level, fromlist, name = args
429 if name:
430 self._safe_import_hook(name, m, fromlist, level=level)
431 else:
432 parent = self.determine_parent(m, level=level)
433 self._safe_import_hook(parent.__name__, None, fromlist, level=0)
434 else:
435 # We don't expect anything else from the generator.
436 raise RuntimeError(what)
437
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000438 for c in co.co_consts:
439 if isinstance(c, type(co)):
440 self.scan_code(c, m)
441
Guido van Rossum75dc4961998-03-05 03:42:00 +0000442 def load_package(self, fqname, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000443 self.msgin(2, "load_package", fqname, pathname)
Thomas Hellerc7aaf952002-11-14 18:45:11 +0000444 newname = replacePackageMap.get(fqname)
445 if newname:
446 fqname = newname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000447 m = self.add_module(fqname)
448 m.__file__ = pathname
449 m.__path__ = [pathname]
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000450
Guido van Rossume7e632a1998-09-14 16:02:28 +0000451 # As per comment at top of file, simulate runtime __path__ additions.
452 m.__path__ = m.__path__ + packagePathMap.get(fqname, [])
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000453
Guido van Rossum912a14c1998-03-05 04:56:37 +0000454 fp, buf, stuff = self.find_module("__init__", m.__path__)
Brett Cannon028011f2010-10-30 00:26:48 +0000455 try:
456 self.load_module(fqname, fp, buf, stuff)
457 self.msgout(2, "load_package ->", m)
458 return m
459 finally:
460 if fp:
461 fp.close()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000462
463 def add_module(self, fqname):
Guido van Rossume2b70bc2006-08-18 22:13:04 +0000464 if fqname in self.modules:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000465 return self.modules[fqname]
466 self.modules[fqname] = m = Module(fqname)
467 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000468
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000469 def find_module(self, name, path, parent=None):
470 if parent is not None:
Thomas Heller2e7c8322004-05-11 15:10:59 +0000471 # assert path is not None
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000472 fullname = parent.__name__+'.'+name
Guido van Rossum03f7f082001-10-18 19:15:32 +0000473 else:
474 fullname = name
475 if fullname in self.excludes:
476 self.msgout(3, "find_module -> Excluded", fullname)
Collin Winterce36ad82007-08-30 01:19:48 +0000477 raise ImportError(name)
Guido van Rossum78fc3631998-03-20 17:37:24 +0000478
Guido van Rossum912a14c1998-03-05 04:56:37 +0000479 if path is None:
480 if name in sys.builtin_module_names:
481 return (None, None, ("", "", imp.C_BUILTIN))
Guido van Rossum78fc3631998-03-20 17:37:24 +0000482
Guido van Rossum912a14c1998-03-05 04:56:37 +0000483 path = self.path
484 return imp.find_module(name, path)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000485
486 def report(self):
Just van Rossume29310a2002-12-31 16:33:00 +0000487 """Print a report to stdout, listing the found modules with their
488 paths, as well as modules that are missing, or seem to be missing.
489 """
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000490 print()
491 print(" %-25s %s" % ("Name", "File"))
492 print(" %-25s %s" % ("----", "----"))
Guido van Rossum912a14c1998-03-05 04:56:37 +0000493 # Print modules found
Guido van Rossumd59cde82007-06-12 00:25:08 +0000494 keys = sorted(self.modules.keys())
Guido van Rossum912a14c1998-03-05 04:56:37 +0000495 for key in keys:
496 m = self.modules[key]
497 if m.__path__:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000498 print("P", end=' ')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000499 else:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000500 print("m", end=' ')
501 print("%-25s" % key, m.__file__ or "")
Guido van Rossum75dc4961998-03-05 03:42:00 +0000502
Guido van Rossum912a14c1998-03-05 04:56:37 +0000503 # Print missing modules
Just van Rossume29310a2002-12-31 16:33:00 +0000504 missing, maybe = self.any_missing_maybe()
505 if missing:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000506 print()
507 print("Missing modules:")
Just van Rossume29310a2002-12-31 16:33:00 +0000508 for name in missing:
Guido van Rossumd59cde82007-06-12 00:25:08 +0000509 mods = sorted(self.badmodules[name].keys())
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000510 print("?", name, "imported from", ', '.join(mods))
Just van Rossume29310a2002-12-31 16:33:00 +0000511 # Print modules that may be missing, but then again, maybe not...
512 if maybe:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000513 print()
514 print("Submodules thay appear to be missing, but could also be", end=' ')
515 print("global names in the parent package:")
Just van Rossume29310a2002-12-31 16:33:00 +0000516 for name in maybe:
Guido van Rossumd59cde82007-06-12 00:25:08 +0000517 mods = sorted(self.badmodules[name].keys())
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000518 print("?", name, "imported from", ', '.join(mods))
Guido van Rossum75dc4961998-03-05 03:42:00 +0000519
Guido van Rossum03f7f082001-10-18 19:15:32 +0000520 def any_missing(self):
Just van Rossume29310a2002-12-31 16:33:00 +0000521 """Return a list of modules that appear to be missing. Use
522 any_missing_maybe() if you want to know which modules are
523 certain to be missing, and which *may* be missing.
524 """
525 missing, maybe = self.any_missing_maybe()
526 return missing + maybe
527
528 def any_missing_maybe(self):
529 """Return two lists, one with modules that are certainly missing
530 and one with modules that *may* be missing. The latter names could
531 either be submodules *or* just global names in the package.
532
533 The reason it can't always be determined is that it's impossible to
534 tell which names are imported when "from module import *" is done
535 with an extension module, short of actually importing it.
536 """
Guido van Rossum03f7f082001-10-18 19:15:32 +0000537 missing = []
Just van Rossume29310a2002-12-31 16:33:00 +0000538 maybe = []
539 for name in self.badmodules:
540 if name in self.excludes:
541 continue
542 i = name.rfind(".")
543 if i < 0:
544 missing.append(name)
545 continue
546 subname = name[i+1:]
547 pkgname = name[:i]
548 pkg = self.modules.get(pkgname)
549 if pkg is not None:
550 if pkgname in self.badmodules[name]:
551 # The package tried to import this module itself and
552 # failed. It's definitely missing.
553 missing.append(name)
554 elif subname in pkg.globalnames:
555 # It's a global in the package: definitely not missing.
556 pass
557 elif pkg.starimports:
558 # It could be missing, but the package did an "import *"
559 # from a non-Python module, so we simply can't be sure.
560 maybe.append(name)
561 else:
562 # It's not a global in the package, the package didn't
563 # do funny star imports, it's very likely to be missing.
564 # The symbol could be inserted into the package from the
565 # outside, but since that's not good style we simply list
566 # it missing.
567 missing.append(name)
568 else:
569 missing.append(name)
570 missing.sort()
571 maybe.sort()
572 return missing, maybe
Guido van Rossum03f7f082001-10-18 19:15:32 +0000573
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000574 def replace_paths_in_code(self, co):
575 new_filename = original_filename = os.path.normpath(co.co_filename)
Just van Rossume29310a2002-12-31 16:33:00 +0000576 for f, r in self.replace_paths:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000577 if original_filename.startswith(f):
Just van Rossume29310a2002-12-31 16:33:00 +0000578 new_filename = r + original_filename[len(f):]
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000579 break
580
581 if self.debug and original_filename not in self.processed_paths:
Just van Rossume29310a2002-12-31 16:33:00 +0000582 if new_filename != original_filename:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000583 self.msgout(2, "co_filename %r changed to %r" \
584 % (original_filename,new_filename,))
585 else:
586 self.msgout(2, "co_filename %r remains unchanged" \
587 % (original_filename,))
588 self.processed_paths.append(original_filename)
589
590 consts = list(co.co_consts)
591 for i in range(len(consts)):
592 if isinstance(consts[i], type(co)):
593 consts[i] = self.replace_paths_in_code(consts[i])
594
Christian Heimes45f9af32007-11-27 21:50:00 +0000595 return types.CodeType(co.co_argcount, co.co_nlocals, co.co_stacksize,
Tim Peters2c60f7a2003-01-29 03:49:43 +0000596 co.co_flags, co.co_code, tuple(consts), co.co_names,
597 co.co_varnames, new_filename, co.co_name,
Neal Norwitz82c72312002-11-12 23:09:12 +0000598 co.co_firstlineno, co.co_lnotab,
599 co.co_freevars, co.co_cellvars)
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000600
Guido van Rossum75dc4961998-03-05 03:42:00 +0000601
602def test():
603 # Parse command line
604 import getopt
605 try:
Guido van Rossumbaf06031998-08-25 14:06:55 +0000606 opts, args = getopt.getopt(sys.argv[1:], "dmp:qx:")
Guido van Rossumb940e112007-01-10 16:19:56 +0000607 except getopt.error as msg:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000608 print(msg)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000609 return
Guido van Rossum75dc4961998-03-05 03:42:00 +0000610
611 # Process options
612 debug = 1
613 domods = 0
614 addpath = []
Guido van Rossumbaf06031998-08-25 14:06:55 +0000615 exclude = []
Guido van Rossum75dc4961998-03-05 03:42:00 +0000616 for o, a in opts:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000617 if o == '-d':
618 debug = debug + 1
619 if o == '-m':
620 domods = 1
621 if o == '-p':
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000622 addpath = addpath + a.split(os.pathsep)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000623 if o == '-q':
624 debug = 0
Guido van Rossumbaf06031998-08-25 14:06:55 +0000625 if o == '-x':
626 exclude.append(a)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000627
628 # Provide default arguments
629 if not args:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000630 script = "hello.py"
Guido van Rossum75dc4961998-03-05 03:42:00 +0000631 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000632 script = args[0]
Guido van Rossum75dc4961998-03-05 03:42:00 +0000633
634 # Set the path based on sys.path and the script directory
635 path = sys.path[:]
636 path[0] = os.path.dirname(script)
637 path = addpath + path
638 if debug > 1:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000639 print("path:")
Guido van Rossum912a14c1998-03-05 04:56:37 +0000640 for item in path:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000641 print(" ", repr(item))
Guido van Rossum75dc4961998-03-05 03:42:00 +0000642
643 # Create the module finder and turn its crank
Guido van Rossumbaf06031998-08-25 14:06:55 +0000644 mf = ModuleFinder(path, debug, exclude)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000645 for arg in args[1:]:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000646 if arg == '-m':
647 domods = 1
648 continue
649 if domods:
650 if arg[-2:] == '.*':
651 mf.import_hook(arg[:-2], None, ["*"])
652 else:
653 mf.import_hook(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000654 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000655 mf.load_file(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000656 mf.run_script(script)
657 mf.report()
Just van Rossume29310a2002-12-31 16:33:00 +0000658 return mf # for -i debugging
Guido van Rossum75dc4961998-03-05 03:42:00 +0000659
660
661if __name__ == '__main__':
662 try:
Just van Rossume29310a2002-12-31 16:33:00 +0000663 mf = test()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000664 except KeyboardInterrupt:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000665 print("\n[interrupt]")