blob: c0910aa8774f10bec3082d206d1b376f25331e2a [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 Araujocee6bb52011-08-01 15:29:07 +020029 packagePathMap.setdefault(packagename, []).append(path)
Guido van Rossum75dc4961998-03-05 03:42:00 +000030
Thomas Hellerc7aaf952002-11-14 18:45:11 +000031replacePackageMap = {}
32
Martin v. Löwis2f48d892011-05-09 08:05:43 +020033# This ReplacePackage mechanism allows modulefinder to work around
34# situations in which a package injects itself under the name
35# of another package into sys.modules at runtime by calling
36# ReplacePackage("real_package_name", "faked_package_name")
Thomas Hellerc7aaf952002-11-14 18:45:11 +000037# before running ModuleFinder.
38
39def ReplacePackage(oldname, newname):
40 replacePackageMap[oldname] = newname
41
42
Guido van Rossum75dc4961998-03-05 03:42:00 +000043class Module:
44
45 def __init__(self, name, file=None, path=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +000046 self.__name__ = name
47 self.__file__ = file
48 self.__path__ = path
49 self.__code__ = None
Just van Rossume29310a2002-12-31 16:33:00 +000050 # The set of global names that are assigned to in the module.
51 # This includes those names imported through starimports of
52 # Python modules.
53 self.globalnames = {}
54 # The set of starimports this module did that could not be
55 # resolved, ie. a starimport from a non-Python module.
56 self.starimports = {}
Guido van Rossum75dc4961998-03-05 03:42:00 +000057
58 def __repr__(self):
Neil Schemenauer32d23c92004-02-15 16:43:20 +000059 s = "Module(%r" % (self.__name__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000060 if self.__file__ is not None:
Walter Dörwald70a6b492004-02-12 17:35:32 +000061 s = s + ", %r" % (self.__file__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000062 if self.__path__ is not None:
Walter Dörwald70a6b492004-02-12 17:35:32 +000063 s = s + ", %r" % (self.__path__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +000064 s = s + ")"
65 return s
Guido van Rossum75dc4961998-03-05 03:42:00 +000066
Guido van Rossum75dc4961998-03-05 03:42:00 +000067class ModuleFinder:
68
Just van Rossume29310a2002-12-31 16:33:00 +000069 def __init__(self, path=None, debug=0, excludes=[], replace_paths=[]):
Guido van Rossum912a14c1998-03-05 04:56:37 +000070 if path is None:
71 path = sys.path
72 self.path = path
73 self.modules = {}
74 self.badmodules = {}
75 self.debug = debug
76 self.indent = 0
Guido van Rossum78fc3631998-03-20 17:37:24 +000077 self.excludes = excludes
Guido van Rossum6b767ac2001-03-20 20:43:34 +000078 self.replace_paths = replace_paths
79 self.processed_paths = [] # Used in debugging only
Guido van Rossum75dc4961998-03-05 03:42:00 +000080
81 def msg(self, level, str, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000082 if level <= self.debug:
83 for i in range(self.indent):
Guido van Rossumbe19ed72007-02-09 05:37:30 +000084 print(" ", end=' ')
85 print(str, end=' ')
Guido van Rossum912a14c1998-03-05 04:56:37 +000086 for arg in args:
Guido van Rossumbe19ed72007-02-09 05:37:30 +000087 print(repr(arg), end=' ')
88 print()
Guido van Rossum75dc4961998-03-05 03:42:00 +000089
90 def msgin(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000091 level = args[0]
92 if level <= self.debug:
93 self.indent = self.indent + 1
Guido van Rossum68468eb2003-02-27 20:14:51 +000094 self.msg(*args)
Guido van Rossum75dc4961998-03-05 03:42:00 +000095
96 def msgout(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000097 level = args[0]
98 if level <= self.debug:
99 self.indent = self.indent - 1
Guido van Rossum68468eb2003-02-27 20:14:51 +0000100 self.msg(*args)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000101
102 def run_script(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000103 self.msg(2, "run_script", pathname)
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200104 with open(pathname) as fp:
Brett Cannon028011f2010-10-30 00:26:48 +0000105 stuff = ("", "r", imp.PY_SOURCE)
106 self.load_module('__main__', fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000107
108 def load_file(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000109 dir, name = os.path.split(pathname)
110 name, ext = os.path.splitext(name)
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200111 with open(pathname) as fp:
Brett Cannon028011f2010-10-30 00:26:48 +0000112 stuff = (ext, "r", imp.PY_SOURCE)
113 self.load_module(name, fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000114
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000115 def import_hook(self, name, caller=None, fromlist=None, level=-1):
116 self.msg(3, "import_hook", name, caller, fromlist, level)
117 parent = self.determine_parent(caller, level=level)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000118 q, tail = self.find_head_package(parent, name)
119 m = self.load_tail(q, tail)
120 if not fromlist:
121 return q
122 if m.__path__:
123 self.ensure_fromlist(m, fromlist)
Thomas Heller318b7b92002-11-26 08:06:50 +0000124 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000125
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000126 def determine_parent(self, caller, level=-1):
127 self.msgin(4, "determine_parent", caller, level)
128 if not caller or level == 0:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000129 self.msgout(4, "determine_parent -> None")
130 return None
131 pname = caller.__name__
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000132 if level >= 1: # relative import
133 if caller.__path__:
134 level -= 1
135 if level == 0:
136 parent = self.modules[pname]
137 assert parent is caller
138 self.msgout(4, "determine_parent ->", parent)
139 return parent
140 if pname.count(".") < level:
Collin Winterce36ad82007-08-30 01:19:48 +0000141 raise ImportError("relative importpath too deep")
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000142 pname = ".".join(pname.split(".")[:-level])
143 parent = self.modules[pname]
144 self.msgout(4, "determine_parent ->", parent)
145 return parent
Guido van Rossum912a14c1998-03-05 04:56:37 +0000146 if caller.__path__:
147 parent = self.modules[pname]
148 assert caller is parent
149 self.msgout(4, "determine_parent ->", parent)
150 return parent
151 if '.' in pname:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000152 i = pname.rfind('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000153 pname = pname[:i]
154 parent = self.modules[pname]
155 assert parent.__name__ == pname
156 self.msgout(4, "determine_parent ->", parent)
157 return parent
158 self.msgout(4, "determine_parent -> None")
159 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000160
161 def find_head_package(self, parent, name):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000162 self.msgin(4, "find_head_package", parent, name)
163 if '.' in name:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000164 i = name.find('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000165 head = name[:i]
166 tail = name[i+1:]
167 else:
168 head = name
169 tail = ""
170 if parent:
171 qname = "%s.%s" % (parent.__name__, head)
172 else:
173 qname = head
174 q = self.import_module(head, qname, parent)
175 if q:
176 self.msgout(4, "find_head_package ->", (q, tail))
177 return q, tail
178 if parent:
179 qname = head
180 parent = None
181 q = self.import_module(head, qname, parent)
182 if q:
183 self.msgout(4, "find_head_package ->", (q, tail))
184 return q, tail
185 self.msgout(4, "raise ImportError: No module named", qname)
Collin Winterce36ad82007-08-30 01:19:48 +0000186 raise ImportError("No module named " + qname)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000187
188 def load_tail(self, q, tail):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000189 self.msgin(4, "load_tail", q, tail)
190 m = q
191 while tail:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000192 i = tail.find('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000193 if i < 0: i = len(tail)
194 head, tail = tail[:i], tail[i+1:]
195 mname = "%s.%s" % (m.__name__, head)
196 m = self.import_module(head, mname, m)
197 if not m:
198 self.msgout(4, "raise ImportError: No module named", mname)
Collin Winterce36ad82007-08-30 01:19:48 +0000199 raise ImportError("No module named " + mname)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000200 self.msgout(4, "load_tail ->", m)
201 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000202
203 def ensure_fromlist(self, m, fromlist, recursive=0):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000204 self.msg(4, "ensure_fromlist", m, fromlist, recursive)
205 for sub in fromlist:
206 if sub == "*":
207 if not recursive:
208 all = self.find_all_submodules(m)
209 if all:
210 self.ensure_fromlist(m, all, 1)
211 elif not hasattr(m, sub):
212 subname = "%s.%s" % (m.__name__, sub)
213 submod = self.import_module(sub, subname, m)
214 if not submod:
Collin Winterce36ad82007-08-30 01:19:48 +0000215 raise ImportError("No module named " + subname)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000216
217 def find_all_submodules(self, m):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000218 if not m.__path__:
219 return
220 modules = {}
Thomas Helleraaf1c8d2003-11-14 10:28:42 +0000221 # 'suffixes' used to be a list hardcoded to [".py", ".pyc", ".pyo"].
222 # But we must also collect Python extension modules - although
223 # we cannot separate normal dlls from Python extensions.
224 suffixes = []
225 for triple in imp.get_suffixes():
226 suffixes.append(triple[0])
Guido van Rossum912a14c1998-03-05 04:56:37 +0000227 for dir in m.__path__:
228 try:
229 names = os.listdir(dir)
230 except os.error:
231 self.msg(2, "can't list directory", dir)
232 continue
233 for name in names:
234 mod = None
235 for suff in suffixes:
236 n = len(suff)
237 if name[-n:] == suff:
238 mod = name[:-n]
239 break
240 if mod and mod != "__init__":
241 modules[mod] = mod
242 return modules.keys()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000243
244 def import_module(self, partname, fqname, parent):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000245 self.msgin(3, "import_module", partname, fqname, parent)
246 try:
247 m = self.modules[fqname]
248 except KeyError:
249 pass
250 else:
251 self.msgout(3, "import_module ->", m)
252 return m
Guido van Rossume2b70bc2006-08-18 22:13:04 +0000253 if fqname in self.badmodules:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000254 self.msgout(3, "import_module -> None")
255 return None
Thomas Heller2e7c8322004-05-11 15:10:59 +0000256 if parent and parent.__path__ is None:
257 self.msgout(3, "import_module -> None")
258 return None
Guido van Rossum912a14c1998-03-05 04:56:37 +0000259 try:
260 fp, pathname, stuff = self.find_module(partname,
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000261 parent and parent.__path__, parent)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000262 except ImportError:
263 self.msgout(3, "import_module ->", None)
264 return None
265 try:
266 m = self.load_module(fqname, fp, pathname, stuff)
267 finally:
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200268 if fp:
269 fp.close()
Guido van Rossum912a14c1998-03-05 04:56:37 +0000270 if parent:
271 setattr(parent, partname, m)
272 self.msgout(3, "import_module ->", m)
273 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000274
Guido van Rossum1bc535d2007-05-15 18:46:22 +0000275 def load_module(self, fqname, fp, pathname, file_info):
276 suffix, mode, type = file_info
Guido van Rossum912a14c1998-03-05 04:56:37 +0000277 self.msgin(2, "load_module", fqname, fp and "fp", pathname)
278 if type == imp.PKG_DIRECTORY:
279 m = self.load_package(fqname, pathname)
280 self.msgout(2, "load_module ->", m)
281 return m
282 if type == imp.PY_SOURCE:
Guido van Rossum78fc3631998-03-20 17:37:24 +0000283 co = compile(fp.read()+'\n', pathname, 'exec')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000284 elif type == imp.PY_COMPILED:
285 if fp.read(4) != imp.get_magic():
286 self.msgout(2, "raise ImportError: Bad magic number", pathname)
Collin Winterce36ad82007-08-30 01:19:48 +0000287 raise ImportError("Bad magic number in %s" % pathname)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000288 fp.read(4)
289 co = marshal.load(fp)
290 else:
291 co = None
292 m = self.add_module(fqname)
Guido van Rossumab045f91998-03-06 19:55:10 +0000293 m.__file__ = pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000294 if co:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000295 if self.replace_paths:
296 co = self.replace_paths_in_code(co)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000297 m.__code__ = co
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000298 self.scan_code(co, m)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000299 self.msgout(2, "load_module ->", m)
300 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000301
Just van Rossume29310a2002-12-31 16:33:00 +0000302 def _add_badmodule(self, name, caller):
303 if name not in self.badmodules:
304 self.badmodules[name] = {}
Benjamin Petersonc0747cf2008-11-03 20:31:38 +0000305 if caller:
306 self.badmodules[name][caller.__name__] = 1
307 else:
308 self.badmodules[name]["-"] = 1
Just van Rossume29310a2002-12-31 16:33:00 +0000309
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000310 def _safe_import_hook(self, name, caller, fromlist, level=-1):
Just van Rossume29310a2002-12-31 16:33:00 +0000311 # wrapper for self.import_hook() that won't raise ImportError
312 if name in self.badmodules:
313 self._add_badmodule(name, caller)
314 return
315 try:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000316 self.import_hook(name, caller, level=level)
Guido van Rossumb940e112007-01-10 16:19:56 +0000317 except ImportError as msg:
Just van Rossume29310a2002-12-31 16:33:00 +0000318 self.msg(2, "ImportError:", str(msg))
319 self._add_badmodule(name, caller)
320 else:
321 if fromlist:
322 for sub in fromlist:
323 if sub in self.badmodules:
324 self._add_badmodule(sub, caller)
325 continue
326 try:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000327 self.import_hook(name, caller, [sub], level=level)
Guido van Rossumb940e112007-01-10 16:19:56 +0000328 except ImportError as msg:
Just van Rossume29310a2002-12-31 16:33:00 +0000329 self.msg(2, "ImportError:", str(msg))
330 fullname = name + "." + sub
331 self._add_badmodule(fullname, caller)
332
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000333 def scan_opcodes(self, co,
334 unpack = struct.unpack):
335 # Scan the code, and yield 'interesting' opcode combinations
336 # Version for Python 2.4 and older
337 code = co.co_code
338 names = co.co_names
339 consts = co.co_consts
340 while code:
341 c = code[0]
342 if c in STORE_OPS:
343 oparg, = unpack('<H', code[1:3])
344 yield "store", (names[oparg],)
345 code = code[3:]
346 continue
347 if c == LOAD_CONST and code[3] == IMPORT_NAME:
348 oparg_1, oparg_2 = unpack('<xHxH', code[:6])
349 yield "import", (consts[oparg_1], names[oparg_2])
350 code = code[6:]
351 continue
352 if c >= HAVE_ARGUMENT:
353 code = code[3:]
354 else:
355 code = code[1:]
356
357 def scan_opcodes_25(self, co,
358 unpack = struct.unpack):
359 # Scan the code, and yield 'interesting' opcode combinations
360 # Python 2.5 version (has absolute and relative imports)
361 code = co.co_code
362 names = co.co_names
363 consts = co.co_consts
364 LOAD_LOAD_AND_IMPORT = LOAD_CONST + LOAD_CONST + IMPORT_NAME
365 while code:
Guido van Rossum98297ee2007-11-06 21:34:58 +0000366 c = bytes([code[0]])
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000367 if c in STORE_OPS:
368 oparg, = unpack('<H', code[1:3])
369 yield "store", (names[oparg],)
370 code = code[3:]
371 continue
372 if code[:9:3] == LOAD_LOAD_AND_IMPORT:
373 oparg_1, oparg_2, oparg_3 = unpack('<xHxHxH', code[:9])
374 level = consts[oparg_1]
375 if level == 0: # absolute import
376 yield "absolute_import", (consts[oparg_2], names[oparg_3])
377 else: # relative import
378 yield "relative_import", (level, consts[oparg_2], names[oparg_3])
379 code = code[9:]
380 continue
381 if c >= HAVE_ARGUMENT:
382 code = code[3:]
383 else:
384 code = code[1:]
385
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000386 def scan_code(self, co, m):
387 code = co.co_code
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000388 if sys.version_info >= (2, 5):
389 scanner = self.scan_opcodes_25
390 else:
391 scanner = self.scan_opcodes
392 for what, args in scanner(co):
393 if what == "store":
394 name, = args
395 m.globalnames[name] = 1
396 elif what == "absolute_import":
397 fromlist, name = args
Just van Rossume29310a2002-12-31 16:33:00 +0000398 have_star = 0
399 if fromlist is not None:
400 if "*" in fromlist:
401 have_star = 1
402 fromlist = [f for f in fromlist if f != "*"]
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000403 self._safe_import_hook(name, m, fromlist, level=0)
Just van Rossume29310a2002-12-31 16:33:00 +0000404 if have_star:
405 # We've encountered an "import *". If it is a Python module,
406 # the code has already been parsed and we can suck out the
407 # global names.
408 mm = None
409 if m.__path__:
410 # At this point we don't know whether 'name' is a
411 # submodule of 'm' or a global module. Let's just try
412 # the full name first.
413 mm = self.modules.get(m.__name__ + "." + name)
414 if mm is None:
415 mm = self.modules.get(name)
416 if mm is not None:
417 m.globalnames.update(mm.globalnames)
418 m.starimports.update(mm.starimports)
419 if mm.__code__ is None:
420 m.starimports[name] = 1
421 else:
422 m.starimports[name] = 1
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000423 elif what == "relative_import":
424 level, fromlist, name = args
425 if name:
426 self._safe_import_hook(name, m, fromlist, level=level)
427 else:
428 parent = self.determine_parent(m, level=level)
429 self._safe_import_hook(parent.__name__, None, fromlist, level=0)
430 else:
431 # We don't expect anything else from the generator.
432 raise RuntimeError(what)
433
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000434 for c in co.co_consts:
435 if isinstance(c, type(co)):
436 self.scan_code(c, m)
437
Guido van Rossum75dc4961998-03-05 03:42:00 +0000438 def load_package(self, fqname, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000439 self.msgin(2, "load_package", fqname, pathname)
Thomas Hellerc7aaf952002-11-14 18:45:11 +0000440 newname = replacePackageMap.get(fqname)
441 if newname:
442 fqname = newname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000443 m = self.add_module(fqname)
444 m.__file__ = pathname
445 m.__path__ = [pathname]
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000446
Guido van Rossume7e632a1998-09-14 16:02:28 +0000447 # As per comment at top of file, simulate runtime __path__ additions.
448 m.__path__ = m.__path__ + packagePathMap.get(fqname, [])
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000449
Guido van Rossum912a14c1998-03-05 04:56:37 +0000450 fp, buf, stuff = self.find_module("__init__", m.__path__)
Brett Cannon028011f2010-10-30 00:26:48 +0000451 try:
452 self.load_module(fqname, fp, buf, stuff)
453 self.msgout(2, "load_package ->", m)
454 return m
455 finally:
456 if fp:
457 fp.close()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000458
459 def add_module(self, fqname):
Guido van Rossume2b70bc2006-08-18 22:13:04 +0000460 if fqname in self.modules:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000461 return self.modules[fqname]
462 self.modules[fqname] = m = Module(fqname)
463 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000464
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000465 def find_module(self, name, path, parent=None):
466 if parent is not None:
Thomas Heller2e7c8322004-05-11 15:10:59 +0000467 # assert path is not None
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000468 fullname = parent.__name__+'.'+name
Guido van Rossum03f7f082001-10-18 19:15:32 +0000469 else:
470 fullname = name
471 if fullname in self.excludes:
472 self.msgout(3, "find_module -> Excluded", fullname)
Collin Winterce36ad82007-08-30 01:19:48 +0000473 raise ImportError(name)
Guido van Rossum78fc3631998-03-20 17:37:24 +0000474
Guido van Rossum912a14c1998-03-05 04:56:37 +0000475 if path is None:
476 if name in sys.builtin_module_names:
477 return (None, None, ("", "", imp.C_BUILTIN))
Guido van Rossum78fc3631998-03-20 17:37:24 +0000478
Guido van Rossum912a14c1998-03-05 04:56:37 +0000479 path = self.path
480 return imp.find_module(name, path)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000481
482 def report(self):
Just van Rossume29310a2002-12-31 16:33:00 +0000483 """Print a report to stdout, listing the found modules with their
484 paths, as well as modules that are missing, or seem to be missing.
485 """
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000486 print()
487 print(" %-25s %s" % ("Name", "File"))
488 print(" %-25s %s" % ("----", "----"))
Guido van Rossum912a14c1998-03-05 04:56:37 +0000489 # Print modules found
Guido van Rossumd59cde82007-06-12 00:25:08 +0000490 keys = sorted(self.modules.keys())
Guido van Rossum912a14c1998-03-05 04:56:37 +0000491 for key in keys:
492 m = self.modules[key]
493 if m.__path__:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000494 print("P", end=' ')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000495 else:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000496 print("m", end=' ')
497 print("%-25s" % key, m.__file__ or "")
Guido van Rossum75dc4961998-03-05 03:42:00 +0000498
Guido van Rossum912a14c1998-03-05 04:56:37 +0000499 # Print missing modules
Just van Rossume29310a2002-12-31 16:33:00 +0000500 missing, maybe = self.any_missing_maybe()
501 if missing:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000502 print()
503 print("Missing modules:")
Just van Rossume29310a2002-12-31 16:33:00 +0000504 for name in missing:
Guido van Rossumd59cde82007-06-12 00:25:08 +0000505 mods = sorted(self.badmodules[name].keys())
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000506 print("?", name, "imported from", ', '.join(mods))
Just van Rossume29310a2002-12-31 16:33:00 +0000507 # Print modules that may be missing, but then again, maybe not...
508 if maybe:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000509 print()
510 print("Submodules thay appear to be missing, but could also be", end=' ')
511 print("global names in the parent package:")
Just van Rossume29310a2002-12-31 16:33:00 +0000512 for name in maybe:
Guido van Rossumd59cde82007-06-12 00:25:08 +0000513 mods = sorted(self.badmodules[name].keys())
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000514 print("?", name, "imported from", ', '.join(mods))
Guido van Rossum75dc4961998-03-05 03:42:00 +0000515
Guido van Rossum03f7f082001-10-18 19:15:32 +0000516 def any_missing(self):
Just van Rossume29310a2002-12-31 16:33:00 +0000517 """Return a list of modules that appear to be missing. Use
518 any_missing_maybe() if you want to know which modules are
519 certain to be missing, and which *may* be missing.
520 """
521 missing, maybe = self.any_missing_maybe()
522 return missing + maybe
523
524 def any_missing_maybe(self):
525 """Return two lists, one with modules that are certainly missing
526 and one with modules that *may* be missing. The latter names could
527 either be submodules *or* just global names in the package.
528
529 The reason it can't always be determined is that it's impossible to
530 tell which names are imported when "from module import *" is done
531 with an extension module, short of actually importing it.
532 """
Guido van Rossum03f7f082001-10-18 19:15:32 +0000533 missing = []
Just van Rossume29310a2002-12-31 16:33:00 +0000534 maybe = []
535 for name in self.badmodules:
536 if name in self.excludes:
537 continue
538 i = name.rfind(".")
539 if i < 0:
540 missing.append(name)
541 continue
542 subname = name[i+1:]
543 pkgname = name[:i]
544 pkg = self.modules.get(pkgname)
545 if pkg is not None:
546 if pkgname in self.badmodules[name]:
547 # The package tried to import this module itself and
548 # failed. It's definitely missing.
549 missing.append(name)
550 elif subname in pkg.globalnames:
551 # It's a global in the package: definitely not missing.
552 pass
553 elif pkg.starimports:
554 # It could be missing, but the package did an "import *"
555 # from a non-Python module, so we simply can't be sure.
556 maybe.append(name)
557 else:
558 # It's not a global in the package, the package didn't
559 # do funny star imports, it's very likely to be missing.
560 # The symbol could be inserted into the package from the
561 # outside, but since that's not good style we simply list
562 # it missing.
563 missing.append(name)
564 else:
565 missing.append(name)
566 missing.sort()
567 maybe.sort()
568 return missing, maybe
Guido van Rossum03f7f082001-10-18 19:15:32 +0000569
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000570 def replace_paths_in_code(self, co):
571 new_filename = original_filename = os.path.normpath(co.co_filename)
Just van Rossume29310a2002-12-31 16:33:00 +0000572 for f, r in self.replace_paths:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000573 if original_filename.startswith(f):
Just van Rossume29310a2002-12-31 16:33:00 +0000574 new_filename = r + original_filename[len(f):]
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000575 break
576
577 if self.debug and original_filename not in self.processed_paths:
Just van Rossume29310a2002-12-31 16:33:00 +0000578 if new_filename != original_filename:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000579 self.msgout(2, "co_filename %r changed to %r" \
580 % (original_filename,new_filename,))
581 else:
582 self.msgout(2, "co_filename %r remains unchanged" \
583 % (original_filename,))
584 self.processed_paths.append(original_filename)
585
586 consts = list(co.co_consts)
587 for i in range(len(consts)):
588 if isinstance(consts[i], type(co)):
589 consts[i] = self.replace_paths_in_code(consts[i])
590
Christian Heimes45f9af32007-11-27 21:50:00 +0000591 return types.CodeType(co.co_argcount, co.co_nlocals, co.co_stacksize,
Tim Peters2c60f7a2003-01-29 03:49:43 +0000592 co.co_flags, co.co_code, tuple(consts), co.co_names,
593 co.co_varnames, new_filename, co.co_name,
Neal Norwitz82c72312002-11-12 23:09:12 +0000594 co.co_firstlineno, co.co_lnotab,
595 co.co_freevars, co.co_cellvars)
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000596
Guido van Rossum75dc4961998-03-05 03:42:00 +0000597
598def test():
599 # Parse command line
600 import getopt
601 try:
Guido van Rossumbaf06031998-08-25 14:06:55 +0000602 opts, args = getopt.getopt(sys.argv[1:], "dmp:qx:")
Guido van Rossumb940e112007-01-10 16:19:56 +0000603 except getopt.error as msg:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000604 print(msg)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000605 return
Guido van Rossum75dc4961998-03-05 03:42:00 +0000606
607 # Process options
608 debug = 1
609 domods = 0
610 addpath = []
Guido van Rossumbaf06031998-08-25 14:06:55 +0000611 exclude = []
Guido van Rossum75dc4961998-03-05 03:42:00 +0000612 for o, a in opts:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000613 if o == '-d':
614 debug = debug + 1
615 if o == '-m':
616 domods = 1
617 if o == '-p':
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000618 addpath = addpath + a.split(os.pathsep)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000619 if o == '-q':
620 debug = 0
Guido van Rossumbaf06031998-08-25 14:06:55 +0000621 if o == '-x':
622 exclude.append(a)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000623
624 # Provide default arguments
625 if not args:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000626 script = "hello.py"
Guido van Rossum75dc4961998-03-05 03:42:00 +0000627 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000628 script = args[0]
Guido van Rossum75dc4961998-03-05 03:42:00 +0000629
630 # Set the path based on sys.path and the script directory
631 path = sys.path[:]
632 path[0] = os.path.dirname(script)
633 path = addpath + path
634 if debug > 1:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000635 print("path:")
Guido van Rossum912a14c1998-03-05 04:56:37 +0000636 for item in path:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000637 print(" ", repr(item))
Guido van Rossum75dc4961998-03-05 03:42:00 +0000638
639 # Create the module finder and turn its crank
Guido van Rossumbaf06031998-08-25 14:06:55 +0000640 mf = ModuleFinder(path, debug, exclude)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000641 for arg in args[1:]:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000642 if arg == '-m':
643 domods = 1
644 continue
645 if domods:
646 if arg[-2:] == '.*':
647 mf.import_hook(arg[:-2], None, ["*"])
648 else:
649 mf.import_hook(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000650 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000651 mf.load_file(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000652 mf.run_script(script)
653 mf.report()
Just van Rossume29310a2002-12-31 16:33:00 +0000654 return mf # for -i debugging
Guido van Rossum75dc4961998-03-05 03:42:00 +0000655
656
657if __name__ == '__main__':
658 try:
Just van Rossume29310a2002-12-31 16:33:00 +0000659 mf = test()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000660 except KeyboardInterrupt:
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200661 print("\n[interrupted]")