blob: a36f1b6d58c4461eed7bc1f39b639ca8f3d3f76f [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
Eric Snow32439d62015-05-02 19:15:18 -06004import importlib._bootstrap_external
Brett Cannon73b969e2012-12-22 19:34:21 -05005import importlib.machinery
Guido van Rossum75dc4961998-03-05 03:42:00 +00006import marshal
7import os
Guido van Rossum75dc4961998-03-05 03:42:00 +00008import sys
Christian Heimes45f9af32007-11-27 21:50:00 +00009import types
Brett Cannone4f41de2013-06-16 13:13:40 -040010import warnings
Brandt Bucher9d7b2c02019-04-07 01:00:41 -070011
Guido van Rossum75dc4961998-03-05 03:42:00 +000012
Serhiy Storchaka02d9f5e2016-05-08 23:43:50 +030013LOAD_CONST = dis.opmap['LOAD_CONST']
14IMPORT_NAME = dis.opmap['IMPORT_NAME']
15STORE_NAME = dis.opmap['STORE_NAME']
16STORE_GLOBAL = dis.opmap['STORE_GLOBAL']
Victor Stinner7e509782016-04-12 18:17:06 +020017STORE_OPS = STORE_NAME, STORE_GLOBAL
Serhiy Storchaka02d9f5e2016-05-08 23:43:50 +030018EXTENDED_ARG = dis.EXTENDED_ARG
Guido van Rossum75dc4961998-03-05 03:42:00 +000019
Brandt Bucher9d7b2c02019-04-07 01:00:41 -070020# Old imp constants:
21
22_SEARCH_ERROR = 0
23_PY_SOURCE = 1
24_PY_COMPILED = 2
25_C_EXTENSION = 3
26_PKG_DIRECTORY = 5
27_C_BUILTIN = 6
28_PY_FROZEN = 7
29
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000030# Modulefinder does a good job at simulating Python's, but it can not
31# handle __path__ modifications packages make at runtime. Therefore there
32# is a mechanism whereby you can register extra paths in this map for a
Thomas Wouters7e474022000-07-16 12:04:32 +000033# package, and it will be honored.
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000034
35# Note this is a mapping is lists of paths.
36packagePathMap = {}
37
38# A Public interface
39def AddPackagePath(packagename, path):
Éric Araujocee6bb52011-08-01 15:29:07 +020040 packagePathMap.setdefault(packagename, []).append(path)
Guido van Rossum75dc4961998-03-05 03:42:00 +000041
Thomas Hellerc7aaf952002-11-14 18:45:11 +000042replacePackageMap = {}
43
Martin v. Löwis2f48d892011-05-09 08:05:43 +020044# This ReplacePackage mechanism allows modulefinder to work around
45# situations in which a package injects itself under the name
46# of another package into sys.modules at runtime by calling
47# ReplacePackage("real_package_name", "faked_package_name")
Thomas Hellerc7aaf952002-11-14 18:45:11 +000048# before running ModuleFinder.
49
50def ReplacePackage(oldname, newname):
51 replacePackageMap[oldname] = newname
52
53
Brandt Bucher9d7b2c02019-04-07 01:00:41 -070054def _find_module(name, path=None):
55 """An importlib reimplementation of imp.find_module (for our purposes)."""
56
57 # It's necessary to clear the caches for our Finder first, in case any
58 # modules are being added/deleted/modified at runtime. In particular,
59 # test_modulefinder.py changes file tree contents in a cache-breaking way:
60
61 importlib.machinery.PathFinder.invalidate_caches()
62
63 spec = importlib.machinery.PathFinder.find_spec(name, path)
64
65 if spec is None:
66 raise ImportError("No module named {name!r}".format(name=name), name=name)
67
68 # Some special cases:
69
70 if spec.loader is importlib.machinery.BuiltinImporter:
71 return None, None, ("", "", _C_BUILTIN)
72
73 if spec.loader is importlib.machinery.FrozenImporter:
74 return None, None, ("", "", _PY_FROZEN)
75
76 file_path = spec.origin
77
78 if spec.loader.is_package(name):
79 return None, os.path.dirname(file_path), ("", "", _PKG_DIRECTORY)
80
81 if isinstance(spec.loader, importlib.machinery.SourceFileLoader):
82 kind = _PY_SOURCE
83 mode = "r"
84
85 elif isinstance(spec.loader, importlib.machinery.ExtensionFileLoader):
86 kind = _C_EXTENSION
87 mode = "rb"
88
89 elif isinstance(spec.loader, importlib.machinery.SourcelessFileLoader):
90 kind = _PY_COMPILED
91 mode = "rb"
92
93 else: # Should never happen.
94 return None, None, ("", "", _SEARCH_ERROR)
95
96 file = open(file_path, mode)
97 suffix = os.path.splitext(file_path)[-1]
98
99 return file, file_path, (suffix, mode, kind)
100
101
Guido van Rossum75dc4961998-03-05 03:42:00 +0000102class Module:
103
104 def __init__(self, name, file=None, path=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000105 self.__name__ = name
106 self.__file__ = file
107 self.__path__ = path
108 self.__code__ = None
Just van Rossume29310a2002-12-31 16:33:00 +0000109 # The set of global names that are assigned to in the module.
110 # This includes those names imported through starimports of
111 # Python modules.
112 self.globalnames = {}
113 # The set of starimports this module did that could not be
114 # resolved, ie. a starimport from a non-Python module.
115 self.starimports = {}
Guido van Rossum75dc4961998-03-05 03:42:00 +0000116
117 def __repr__(self):
Neil Schemenauer32d23c92004-02-15 16:43:20 +0000118 s = "Module(%r" % (self.__name__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000119 if self.__file__ is not None:
Walter Dörwald70a6b492004-02-12 17:35:32 +0000120 s = s + ", %r" % (self.__file__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000121 if self.__path__ is not None:
Walter Dörwald70a6b492004-02-12 17:35:32 +0000122 s = s + ", %r" % (self.__path__,)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000123 s = s + ")"
124 return s
Guido van Rossum75dc4961998-03-05 03:42:00 +0000125
Guido van Rossum75dc4961998-03-05 03:42:00 +0000126class ModuleFinder:
127
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700128 def __init__(self, path=None, debug=0, excludes=None, replace_paths=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000129 if path is None:
130 path = sys.path
131 self.path = path
132 self.modules = {}
133 self.badmodules = {}
134 self.debug = debug
135 self.indent = 0
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700136 self.excludes = excludes if excludes is not None else []
137 self.replace_paths = replace_paths if replace_paths is not None else []
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000138 self.processed_paths = [] # Used in debugging only
Guido van Rossum75dc4961998-03-05 03:42:00 +0000139
140 def msg(self, level, str, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000141 if level <= self.debug:
142 for i in range(self.indent):
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000143 print(" ", end=' ')
144 print(str, end=' ')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000145 for arg in args:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000146 print(repr(arg), end=' ')
147 print()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000148
149 def msgin(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000150 level = args[0]
151 if level <= self.debug:
152 self.indent = self.indent + 1
Guido van Rossum68468eb2003-02-27 20:14:51 +0000153 self.msg(*args)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000154
155 def msgout(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000156 level = args[0]
157 if level <= self.debug:
158 self.indent = self.indent - 1
Guido van Rossum68468eb2003-02-27 20:14:51 +0000159 self.msg(*args)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000160
161 def run_script(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000162 self.msg(2, "run_script", pathname)
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200163 with open(pathname) as fp:
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700164 stuff = ("", "r", _PY_SOURCE)
Brett Cannon028011f2010-10-30 00:26:48 +0000165 self.load_module('__main__', fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000166
167 def load_file(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000168 dir, name = os.path.split(pathname)
169 name, ext = os.path.splitext(name)
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200170 with open(pathname) as fp:
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700171 stuff = (ext, "r", _PY_SOURCE)
Brett Cannon028011f2010-10-30 00:26:48 +0000172 self.load_module(name, fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000173
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000174 def import_hook(self, name, caller=None, fromlist=None, level=-1):
175 self.msg(3, "import_hook", name, caller, fromlist, level)
176 parent = self.determine_parent(caller, level=level)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000177 q, tail = self.find_head_package(parent, name)
178 m = self.load_tail(q, tail)
179 if not fromlist:
180 return q
181 if m.__path__:
182 self.ensure_fromlist(m, fromlist)
Thomas Heller318b7b92002-11-26 08:06:50 +0000183 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000184
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000185 def determine_parent(self, caller, level=-1):
186 self.msgin(4, "determine_parent", caller, level)
187 if not caller or level == 0:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000188 self.msgout(4, "determine_parent -> None")
189 return None
190 pname = caller.__name__
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000191 if level >= 1: # relative import
192 if caller.__path__:
193 level -= 1
194 if level == 0:
195 parent = self.modules[pname]
196 assert parent is caller
197 self.msgout(4, "determine_parent ->", parent)
198 return parent
199 if pname.count(".") < level:
Collin Winterce36ad82007-08-30 01:19:48 +0000200 raise ImportError("relative importpath too deep")
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000201 pname = ".".join(pname.split(".")[:-level])
202 parent = self.modules[pname]
203 self.msgout(4, "determine_parent ->", parent)
204 return parent
Guido van Rossum912a14c1998-03-05 04:56:37 +0000205 if caller.__path__:
206 parent = self.modules[pname]
207 assert caller is parent
208 self.msgout(4, "determine_parent ->", parent)
209 return parent
210 if '.' in pname:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000211 i = pname.rfind('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000212 pname = pname[:i]
213 parent = self.modules[pname]
214 assert parent.__name__ == pname
215 self.msgout(4, "determine_parent ->", parent)
216 return parent
217 self.msgout(4, "determine_parent -> None")
218 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000219
220 def find_head_package(self, parent, name):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000221 self.msgin(4, "find_head_package", parent, name)
222 if '.' in name:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000223 i = name.find('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000224 head = name[:i]
225 tail = name[i+1:]
226 else:
227 head = name
228 tail = ""
229 if parent:
230 qname = "%s.%s" % (parent.__name__, head)
231 else:
232 qname = head
233 q = self.import_module(head, qname, parent)
234 if q:
235 self.msgout(4, "find_head_package ->", (q, tail))
236 return q, tail
237 if parent:
238 qname = head
239 parent = None
240 q = self.import_module(head, qname, parent)
241 if q:
242 self.msgout(4, "find_head_package ->", (q, tail))
243 return q, tail
244 self.msgout(4, "raise ImportError: No module named", qname)
Collin Winterce36ad82007-08-30 01:19:48 +0000245 raise ImportError("No module named " + qname)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000246
247 def load_tail(self, q, tail):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000248 self.msgin(4, "load_tail", q, tail)
249 m = q
250 while tail:
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000251 i = tail.find('.')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000252 if i < 0: i = len(tail)
253 head, tail = tail[:i], tail[i+1:]
254 mname = "%s.%s" % (m.__name__, head)
255 m = self.import_module(head, mname, m)
256 if not m:
257 self.msgout(4, "raise ImportError: No module named", mname)
Collin Winterce36ad82007-08-30 01:19:48 +0000258 raise ImportError("No module named " + mname)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000259 self.msgout(4, "load_tail ->", m)
260 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000261
262 def ensure_fromlist(self, m, fromlist, recursive=0):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000263 self.msg(4, "ensure_fromlist", m, fromlist, recursive)
264 for sub in fromlist:
265 if sub == "*":
266 if not recursive:
267 all = self.find_all_submodules(m)
268 if all:
269 self.ensure_fromlist(m, all, 1)
270 elif not hasattr(m, sub):
271 subname = "%s.%s" % (m.__name__, sub)
272 submod = self.import_module(sub, subname, m)
273 if not submod:
Collin Winterce36ad82007-08-30 01:19:48 +0000274 raise ImportError("No module named " + subname)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000275
276 def find_all_submodules(self, m):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000277 if not m.__path__:
278 return
279 modules = {}
Brett Cannonf299abd2015-04-13 14:21:02 -0400280 # 'suffixes' used to be a list hardcoded to [".py", ".pyc"].
Thomas Helleraaf1c8d2003-11-14 10:28:42 +0000281 # But we must also collect Python extension modules - although
282 # we cannot separate normal dlls from Python extensions.
283 suffixes = []
Brett Cannoncb66eb02012-05-11 12:58:42 -0400284 suffixes += importlib.machinery.EXTENSION_SUFFIXES[:]
285 suffixes += importlib.machinery.SOURCE_SUFFIXES[:]
286 suffixes += importlib.machinery.BYTECODE_SUFFIXES[:]
Guido van Rossum912a14c1998-03-05 04:56:37 +0000287 for dir in m.__path__:
288 try:
289 names = os.listdir(dir)
Andrew Svetlovad28c7f2012-12-18 22:02:39 +0200290 except OSError:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000291 self.msg(2, "can't list directory", dir)
292 continue
293 for name in names:
294 mod = None
295 for suff in suffixes:
296 n = len(suff)
297 if name[-n:] == suff:
298 mod = name[:-n]
299 break
300 if mod and mod != "__init__":
301 modules[mod] = mod
302 return modules.keys()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000303
304 def import_module(self, partname, fqname, parent):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000305 self.msgin(3, "import_module", partname, fqname, parent)
306 try:
307 m = self.modules[fqname]
308 except KeyError:
309 pass
310 else:
311 self.msgout(3, "import_module ->", m)
312 return m
Guido van Rossume2b70bc2006-08-18 22:13:04 +0000313 if fqname in self.badmodules:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000314 self.msgout(3, "import_module -> None")
315 return None
Thomas Heller2e7c8322004-05-11 15:10:59 +0000316 if parent and parent.__path__ is None:
317 self.msgout(3, "import_module -> None")
318 return None
Guido van Rossum912a14c1998-03-05 04:56:37 +0000319 try:
320 fp, pathname, stuff = self.find_module(partname,
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000321 parent and parent.__path__, parent)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000322 except ImportError:
323 self.msgout(3, "import_module ->", None)
324 return None
325 try:
326 m = self.load_module(fqname, fp, pathname, stuff)
327 finally:
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200328 if fp:
329 fp.close()
Guido van Rossum912a14c1998-03-05 04:56:37 +0000330 if parent:
331 setattr(parent, partname, m)
332 self.msgout(3, "import_module ->", m)
333 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000334
Guido van Rossum1bc535d2007-05-15 18:46:22 +0000335 def load_module(self, fqname, fp, pathname, file_info):
336 suffix, mode, type = file_info
Guido van Rossum912a14c1998-03-05 04:56:37 +0000337 self.msgin(2, "load_module", fqname, fp and "fp", pathname)
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700338 if type == _PKG_DIRECTORY:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000339 m = self.load_package(fqname, pathname)
340 self.msgout(2, "load_module ->", m)
341 return m
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700342 if type == _PY_SOURCE:
Guido van Rossum78fc3631998-03-20 17:37:24 +0000343 co = compile(fp.read()+'\n', pathname, 'exec')
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700344 elif type == _PY_COMPILED:
Brett Cannon0f384782014-02-28 10:50:34 -0500345 try:
Benjamin Peterson42aa93b2017-12-09 10:26:52 -0800346 data = fp.read()
347 importlib._bootstrap_external._classify_pyc(data, fqname, {})
Brett Cannon0f384782014-02-28 10:50:34 -0500348 except ImportError as exc:
349 self.msgout(2, "raise ImportError: " + str(exc), pathname)
350 raise
Benjamin Peterson42aa93b2017-12-09 10:26:52 -0800351 co = marshal.loads(memoryview(data)[16:])
Guido van Rossum912a14c1998-03-05 04:56:37 +0000352 else:
353 co = None
354 m = self.add_module(fqname)
Guido van Rossumab045f91998-03-06 19:55:10 +0000355 m.__file__ = pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000356 if co:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000357 if self.replace_paths:
358 co = self.replace_paths_in_code(co)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000359 m.__code__ = co
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000360 self.scan_code(co, m)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000361 self.msgout(2, "load_module ->", m)
362 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000363
Just van Rossume29310a2002-12-31 16:33:00 +0000364 def _add_badmodule(self, name, caller):
365 if name not in self.badmodules:
366 self.badmodules[name] = {}
Benjamin Petersonc0747cf2008-11-03 20:31:38 +0000367 if caller:
368 self.badmodules[name][caller.__name__] = 1
369 else:
370 self.badmodules[name]["-"] = 1
Just van Rossume29310a2002-12-31 16:33:00 +0000371
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000372 def _safe_import_hook(self, name, caller, fromlist, level=-1):
Just van Rossume29310a2002-12-31 16:33:00 +0000373 # wrapper for self.import_hook() that won't raise ImportError
374 if name in self.badmodules:
375 self._add_badmodule(name, caller)
376 return
377 try:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000378 self.import_hook(name, caller, level=level)
Guido van Rossumb940e112007-01-10 16:19:56 +0000379 except ImportError as msg:
Just van Rossume29310a2002-12-31 16:33:00 +0000380 self.msg(2, "ImportError:", str(msg))
381 self._add_badmodule(name, caller)
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700382 except SyntaxError as msg:
383 self.msg(2, "SyntaxError:", str(msg))
384 self._add_badmodule(name, caller)
Just van Rossume29310a2002-12-31 16:33:00 +0000385 else:
386 if fromlist:
387 for sub in fromlist:
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700388 fullname = name + "." + sub
389 if fullname in self.badmodules:
390 self._add_badmodule(fullname, caller)
Just van Rossume29310a2002-12-31 16:33:00 +0000391 continue
392 try:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000393 self.import_hook(name, caller, [sub], level=level)
Guido van Rossumb940e112007-01-10 16:19:56 +0000394 except ImportError as msg:
Just van Rossume29310a2002-12-31 16:33:00 +0000395 self.msg(2, "ImportError:", str(msg))
Just van Rossume29310a2002-12-31 16:33:00 +0000396 self._add_badmodule(fullname, caller)
397
Serhiy Storchaka02d9f5e2016-05-08 23:43:50 +0300398 def scan_opcodes(self, co):
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000399 # Scan the code, and yield 'interesting' opcode combinations
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000400 code = co.co_code
401 names = co.co_names
402 consts = co.co_consts
Serhiy Storchaka02d9f5e2016-05-08 23:43:50 +0300403 opargs = [(op, arg) for _, op, arg in dis._unpack_opargs(code)
404 if op != EXTENDED_ARG]
405 for i, (op, oparg) in enumerate(opargs):
406 if op in STORE_OPS:
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000407 yield "store", (names[oparg],)
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000408 continue
Serhiy Storchaka02d9f5e2016-05-08 23:43:50 +0300409 if (op == IMPORT_NAME and i >= 2
410 and opargs[i-1][0] == opargs[i-2][0] == LOAD_CONST):
411 level = consts[opargs[i-2][1]]
412 fromlist = consts[opargs[i-1][1]]
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000413 if level == 0: # absolute import
Serhiy Storchaka02d9f5e2016-05-08 23:43:50 +0300414 yield "absolute_import", (fromlist, names[oparg])
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000415 else: # relative import
Serhiy Storchaka02d9f5e2016-05-08 23:43:50 +0300416 yield "relative_import", (level, fromlist, names[oparg])
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000417 continue
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000418
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000419 def scan_code(self, co, m):
420 code = co.co_code
Serhiy Storchaka02d9f5e2016-05-08 23:43:50 +0300421 scanner = self.scan_opcodes
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000422 for what, args in scanner(co):
423 if what == "store":
424 name, = args
425 m.globalnames[name] = 1
426 elif what == "absolute_import":
427 fromlist, name = args
Just van Rossume29310a2002-12-31 16:33:00 +0000428 have_star = 0
429 if fromlist is not None:
430 if "*" in fromlist:
431 have_star = 1
432 fromlist = [f for f in fromlist if f != "*"]
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000433 self._safe_import_hook(name, m, fromlist, level=0)
Just van Rossume29310a2002-12-31 16:33:00 +0000434 if have_star:
435 # We've encountered an "import *". If it is a Python module,
436 # the code has already been parsed and we can suck out the
437 # global names.
438 mm = None
439 if m.__path__:
440 # At this point we don't know whether 'name' is a
441 # submodule of 'm' or a global module. Let's just try
442 # the full name first.
443 mm = self.modules.get(m.__name__ + "." + name)
444 if mm is None:
445 mm = self.modules.get(name)
446 if mm is not None:
447 m.globalnames.update(mm.globalnames)
448 m.starimports.update(mm.starimports)
449 if mm.__code__ is None:
450 m.starimports[name] = 1
451 else:
452 m.starimports[name] = 1
Guido van Rossumfc2a0a82006-10-27 23:06:01 +0000453 elif what == "relative_import":
454 level, fromlist, name = args
455 if name:
456 self._safe_import_hook(name, m, fromlist, level=level)
457 else:
458 parent = self.determine_parent(m, level=level)
459 self._safe_import_hook(parent.__name__, None, fromlist, level=0)
460 else:
461 # We don't expect anything else from the generator.
462 raise RuntimeError(what)
463
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000464 for c in co.co_consts:
465 if isinstance(c, type(co)):
466 self.scan_code(c, m)
467
Guido van Rossum75dc4961998-03-05 03:42:00 +0000468 def load_package(self, fqname, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000469 self.msgin(2, "load_package", fqname, pathname)
Thomas Hellerc7aaf952002-11-14 18:45:11 +0000470 newname = replacePackageMap.get(fqname)
471 if newname:
472 fqname = newname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000473 m = self.add_module(fqname)
474 m.__file__ = pathname
475 m.__path__ = [pathname]
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000476
Guido van Rossume7e632a1998-09-14 16:02:28 +0000477 # As per comment at top of file, simulate runtime __path__ additions.
478 m.__path__ = m.__path__ + packagePathMap.get(fqname, [])
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000479
Guido van Rossum912a14c1998-03-05 04:56:37 +0000480 fp, buf, stuff = self.find_module("__init__", m.__path__)
Brett Cannon028011f2010-10-30 00:26:48 +0000481 try:
482 self.load_module(fqname, fp, buf, stuff)
483 self.msgout(2, "load_package ->", m)
484 return m
485 finally:
486 if fp:
487 fp.close()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000488
489 def add_module(self, fqname):
Guido van Rossume2b70bc2006-08-18 22:13:04 +0000490 if fqname in self.modules:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000491 return self.modules[fqname]
492 self.modules[fqname] = m = Module(fqname)
493 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000494
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000495 def find_module(self, name, path, parent=None):
496 if parent is not None:
Thomas Heller2e7c8322004-05-11 15:10:59 +0000497 # assert path is not None
Just van Rossumf0dfbaf2003-03-05 17:23:48 +0000498 fullname = parent.__name__+'.'+name
Guido van Rossum03f7f082001-10-18 19:15:32 +0000499 else:
500 fullname = name
501 if fullname in self.excludes:
502 self.msgout(3, "find_module -> Excluded", fullname)
Collin Winterce36ad82007-08-30 01:19:48 +0000503 raise ImportError(name)
Guido van Rossum78fc3631998-03-20 17:37:24 +0000504
Guido van Rossum912a14c1998-03-05 04:56:37 +0000505 if path is None:
506 if name in sys.builtin_module_names:
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700507 return (None, None, ("", "", _C_BUILTIN))
Guido van Rossum78fc3631998-03-20 17:37:24 +0000508
Guido van Rossum912a14c1998-03-05 04:56:37 +0000509 path = self.path
Brandt Bucher9d7b2c02019-04-07 01:00:41 -0700510
511 return _find_module(name, path)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000512
513 def report(self):
Just van Rossume29310a2002-12-31 16:33:00 +0000514 """Print a report to stdout, listing the found modules with their
515 paths, as well as modules that are missing, or seem to be missing.
516 """
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000517 print()
518 print(" %-25s %s" % ("Name", "File"))
519 print(" %-25s %s" % ("----", "----"))
Guido van Rossum912a14c1998-03-05 04:56:37 +0000520 # Print modules found
Guido van Rossumd59cde82007-06-12 00:25:08 +0000521 keys = sorted(self.modules.keys())
Guido van Rossum912a14c1998-03-05 04:56:37 +0000522 for key in keys:
523 m = self.modules[key]
524 if m.__path__:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000525 print("P", end=' ')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000526 else:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000527 print("m", end=' ')
528 print("%-25s" % key, m.__file__ or "")
Guido van Rossum75dc4961998-03-05 03:42:00 +0000529
Guido van Rossum912a14c1998-03-05 04:56:37 +0000530 # Print missing modules
Just van Rossume29310a2002-12-31 16:33:00 +0000531 missing, maybe = self.any_missing_maybe()
532 if missing:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000533 print()
534 print("Missing modules:")
Just van Rossume29310a2002-12-31 16:33:00 +0000535 for name in missing:
Guido van Rossumd59cde82007-06-12 00:25:08 +0000536 mods = sorted(self.badmodules[name].keys())
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000537 print("?", name, "imported from", ', '.join(mods))
Just van Rossume29310a2002-12-31 16:33:00 +0000538 # Print modules that may be missing, but then again, maybe not...
539 if maybe:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000540 print()
Ezio Melotti30b9d5d2013-08-17 15:50:46 +0300541 print("Submodules that appear to be missing, but could also be", end=' ')
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000542 print("global names in the parent package:")
Just van Rossume29310a2002-12-31 16:33:00 +0000543 for name in maybe:
Guido van Rossumd59cde82007-06-12 00:25:08 +0000544 mods = sorted(self.badmodules[name].keys())
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000545 print("?", name, "imported from", ', '.join(mods))
Guido van Rossum75dc4961998-03-05 03:42:00 +0000546
Guido van Rossum03f7f082001-10-18 19:15:32 +0000547 def any_missing(self):
Just van Rossume29310a2002-12-31 16:33:00 +0000548 """Return a list of modules that appear to be missing. Use
549 any_missing_maybe() if you want to know which modules are
550 certain to be missing, and which *may* be missing.
551 """
552 missing, maybe = self.any_missing_maybe()
553 return missing + maybe
554
555 def any_missing_maybe(self):
556 """Return two lists, one with modules that are certainly missing
557 and one with modules that *may* be missing. The latter names could
558 either be submodules *or* just global names in the package.
559
560 The reason it can't always be determined is that it's impossible to
561 tell which names are imported when "from module import *" is done
562 with an extension module, short of actually importing it.
563 """
Guido van Rossum03f7f082001-10-18 19:15:32 +0000564 missing = []
Just van Rossume29310a2002-12-31 16:33:00 +0000565 maybe = []
566 for name in self.badmodules:
567 if name in self.excludes:
568 continue
569 i = name.rfind(".")
570 if i < 0:
571 missing.append(name)
572 continue
573 subname = name[i+1:]
574 pkgname = name[:i]
575 pkg = self.modules.get(pkgname)
576 if pkg is not None:
577 if pkgname in self.badmodules[name]:
578 # The package tried to import this module itself and
579 # failed. It's definitely missing.
580 missing.append(name)
581 elif subname in pkg.globalnames:
582 # It's a global in the package: definitely not missing.
583 pass
584 elif pkg.starimports:
585 # It could be missing, but the package did an "import *"
586 # from a non-Python module, so we simply can't be sure.
587 maybe.append(name)
588 else:
589 # It's not a global in the package, the package didn't
590 # do funny star imports, it's very likely to be missing.
591 # The symbol could be inserted into the package from the
592 # outside, but since that's not good style we simply list
593 # it missing.
594 missing.append(name)
595 else:
596 missing.append(name)
597 missing.sort()
598 maybe.sort()
599 return missing, maybe
Guido van Rossum03f7f082001-10-18 19:15:32 +0000600
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000601 def replace_paths_in_code(self, co):
602 new_filename = original_filename = os.path.normpath(co.co_filename)
Just van Rossume29310a2002-12-31 16:33:00 +0000603 for f, r in self.replace_paths:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000604 if original_filename.startswith(f):
Just van Rossume29310a2002-12-31 16:33:00 +0000605 new_filename = r + original_filename[len(f):]
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000606 break
607
608 if self.debug and original_filename not in self.processed_paths:
Just van Rossume29310a2002-12-31 16:33:00 +0000609 if new_filename != original_filename:
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000610 self.msgout(2, "co_filename %r changed to %r" \
611 % (original_filename,new_filename,))
612 else:
613 self.msgout(2, "co_filename %r remains unchanged" \
614 % (original_filename,))
615 self.processed_paths.append(original_filename)
616
617 consts = list(co.co_consts)
618 for i in range(len(consts)):
619 if isinstance(consts[i], type(co)):
620 consts[i] = self.replace_paths_in_code(consts[i])
621
Pablo Galindo8c77b8c2019-04-29 13:36:57 +0100622 return types.CodeType(co.co_argcount, co.co_posonlyargcount,
623 co.co_kwonlyargcount, co.co_nlocals,
624 co.co_stacksize, co.co_flags,
Berker Peksag0a0d1da2014-07-07 14:58:12 +0300625 co.co_code, tuple(consts), co.co_names,
626 co.co_varnames, new_filename, co.co_name,
627 co.co_firstlineno, co.co_lnotab, co.co_freevars,
628 co.co_cellvars)
Guido van Rossum6b767ac2001-03-20 20:43:34 +0000629
Guido van Rossum75dc4961998-03-05 03:42:00 +0000630
631def test():
632 # Parse command line
633 import getopt
634 try:
Guido van Rossumbaf06031998-08-25 14:06:55 +0000635 opts, args = getopt.getopt(sys.argv[1:], "dmp:qx:")
Guido van Rossumb940e112007-01-10 16:19:56 +0000636 except getopt.error as msg:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000637 print(msg)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000638 return
Guido van Rossum75dc4961998-03-05 03:42:00 +0000639
640 # Process options
641 debug = 1
642 domods = 0
643 addpath = []
Guido van Rossumbaf06031998-08-25 14:06:55 +0000644 exclude = []
Guido van Rossum75dc4961998-03-05 03:42:00 +0000645 for o, a in opts:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000646 if o == '-d':
647 debug = debug + 1
648 if o == '-m':
649 domods = 1
650 if o == '-p':
Walter Dörwaldaaab30e2002-09-11 20:36:02 +0000651 addpath = addpath + a.split(os.pathsep)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000652 if o == '-q':
653 debug = 0
Guido van Rossumbaf06031998-08-25 14:06:55 +0000654 if o == '-x':
655 exclude.append(a)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000656
657 # Provide default arguments
658 if not args:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000659 script = "hello.py"
Guido van Rossum75dc4961998-03-05 03:42:00 +0000660 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000661 script = args[0]
Guido van Rossum75dc4961998-03-05 03:42:00 +0000662
663 # Set the path based on sys.path and the script directory
664 path = sys.path[:]
665 path[0] = os.path.dirname(script)
666 path = addpath + path
667 if debug > 1:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000668 print("path:")
Guido van Rossum912a14c1998-03-05 04:56:37 +0000669 for item in path:
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000670 print(" ", repr(item))
Guido van Rossum75dc4961998-03-05 03:42:00 +0000671
672 # Create the module finder and turn its crank
Guido van Rossumbaf06031998-08-25 14:06:55 +0000673 mf = ModuleFinder(path, debug, exclude)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000674 for arg in args[1:]:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000675 if arg == '-m':
676 domods = 1
677 continue
678 if domods:
679 if arg[-2:] == '.*':
680 mf.import_hook(arg[:-2], None, ["*"])
681 else:
682 mf.import_hook(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000683 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000684 mf.load_file(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000685 mf.run_script(script)
686 mf.report()
Just van Rossume29310a2002-12-31 16:33:00 +0000687 return mf # for -i debugging
Guido van Rossum75dc4961998-03-05 03:42:00 +0000688
689
690if __name__ == '__main__':
691 try:
Just van Rossume29310a2002-12-31 16:33:00 +0000692 mf = test()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000693 except KeyboardInterrupt:
Éric Araujo1e3a68d2011-07-28 23:35:29 +0200694 print("\n[interrupted]")