blob: a75e2f3f750cfba3e807de2d792365c8334e71f1 [file] [log] [blame]
Guido van Rossum75dc4961998-03-05 03:42:00 +00001"""Find modules used by a script, using introspection."""
2
3import dis
4import imp
5import marshal
6import os
7import re
8import string
9import sys
10
Guido van Rossum78fc3631998-03-20 17:37:24 +000011if sys.platform=="win32":
12 # On Windows, we can locate modules in the registry with
13 # the help of the win32api package.
14 try:
15 import win32api
16 except ImportError:
17 print "The win32api module is not available - modules listed"
18 print "in the registry will not be found."
19 win32api = None
20
Guido van Rossum75dc4961998-03-05 03:42:00 +000021
22IMPORT_NAME = dis.opname.index('IMPORT_NAME')
23IMPORT_FROM = dis.opname.index('IMPORT_FROM')
24
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000025# Modulefinder does a good job at simulating Python's, but it can not
26# handle __path__ modifications packages make at runtime. Therefore there
27# is a mechanism whereby you can register extra paths in this map for a
Thomas Wouters7e474022000-07-16 12:04:32 +000028# package, and it will be honored.
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000029
30# Note this is a mapping is lists of paths.
31packagePathMap = {}
32
33# A Public interface
34def AddPackagePath(packagename, path):
Guido van Rossume7e632a1998-09-14 16:02:28 +000035 paths = packagePathMap.get(packagename, [])
36 paths.append(path)
37 packagePathMap[packagename] = paths
Guido van Rossum75dc4961998-03-05 03:42:00 +000038
39class Module:
40
41 def __init__(self, name, file=None, path=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +000042 self.__name__ = name
43 self.__file__ = file
44 self.__path__ = path
45 self.__code__ = None
Guido van Rossum75dc4961998-03-05 03:42:00 +000046
47 def __repr__(self):
Guido van Rossum912a14c1998-03-05 04:56:37 +000048 s = "Module(%s" % `self.__name__`
49 if self.__file__ is not None:
50 s = s + ", %s" % `self.__file__`
51 if self.__path__ is not None:
52 s = s + ", %s" % `self.__path__`
53 s = s + ")"
54 return s
Guido van Rossum75dc4961998-03-05 03:42:00 +000055
56
57class ModuleFinder:
58
Guido van Rossum78fc3631998-03-20 17:37:24 +000059 def __init__(self, path=None, debug=0, excludes = []):
Guido van Rossum912a14c1998-03-05 04:56:37 +000060 if path is None:
61 path = sys.path
62 self.path = path
63 self.modules = {}
64 self.badmodules = {}
65 self.debug = debug
66 self.indent = 0
Guido van Rossum78fc3631998-03-20 17:37:24 +000067 self.excludes = excludes
Guido van Rossum75dc4961998-03-05 03:42:00 +000068
69 def msg(self, level, str, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000070 if level <= self.debug:
71 for i in range(self.indent):
72 print " ",
73 print str,
74 for arg in args:
75 print repr(arg),
76 print
Guido van Rossum75dc4961998-03-05 03:42:00 +000077
78 def msgin(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000079 level = args[0]
80 if level <= self.debug:
81 self.indent = self.indent + 1
82 apply(self.msg, args)
Guido van Rossum75dc4961998-03-05 03:42:00 +000083
84 def msgout(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000085 level = args[0]
86 if level <= self.debug:
87 self.indent = self.indent - 1
88 apply(self.msg, args)
Guido van Rossum75dc4961998-03-05 03:42:00 +000089
90 def run_script(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +000091 self.msg(2, "run_script", pathname)
92 fp = open(pathname)
93 stuff = ("", "r", imp.PY_SOURCE)
94 self.load_module('__main__', fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +000095
96 def load_file(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +000097 dir, name = os.path.split(pathname)
98 name, ext = os.path.splitext(name)
99 fp = open(pathname)
100 stuff = (ext, "r", imp.PY_SOURCE)
101 self.load_module(name, fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000102
103 def import_hook(self, name, caller=None, fromlist=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000104 self.msg(3, "import_hook", name, caller, fromlist)
105 parent = self.determine_parent(caller)
106 q, tail = self.find_head_package(parent, name)
107 m = self.load_tail(q, tail)
108 if not fromlist:
109 return q
110 if m.__path__:
111 self.ensure_fromlist(m, fromlist)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000112
113 def determine_parent(self, caller):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000114 self.msgin(4, "determine_parent", caller)
115 if not caller:
116 self.msgout(4, "determine_parent -> None")
117 return None
118 pname = caller.__name__
119 if caller.__path__:
120 parent = self.modules[pname]
121 assert caller is parent
122 self.msgout(4, "determine_parent ->", parent)
123 return parent
124 if '.' in pname:
125 i = string.rfind(pname, '.')
126 pname = pname[:i]
127 parent = self.modules[pname]
128 assert parent.__name__ == pname
129 self.msgout(4, "determine_parent ->", parent)
130 return parent
131 self.msgout(4, "determine_parent -> None")
132 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000133
134 def find_head_package(self, parent, name):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000135 self.msgin(4, "find_head_package", parent, name)
136 if '.' in name:
137 i = string.find(name, '.')
138 head = name[:i]
139 tail = name[i+1:]
140 else:
141 head = name
142 tail = ""
143 if parent:
144 qname = "%s.%s" % (parent.__name__, head)
145 else:
146 qname = head
147 q = self.import_module(head, qname, parent)
148 if q:
149 self.msgout(4, "find_head_package ->", (q, tail))
150 return q, tail
151 if parent:
152 qname = head
153 parent = None
154 q = self.import_module(head, qname, parent)
155 if q:
156 self.msgout(4, "find_head_package ->", (q, tail))
157 return q, tail
158 self.msgout(4, "raise ImportError: No module named", qname)
159 raise ImportError, "No module named " + qname
Guido van Rossum75dc4961998-03-05 03:42:00 +0000160
161 def load_tail(self, q, tail):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000162 self.msgin(4, "load_tail", q, tail)
163 m = q
164 while tail:
165 i = string.find(tail, '.')
166 if i < 0: i = len(tail)
167 head, tail = tail[:i], tail[i+1:]
168 mname = "%s.%s" % (m.__name__, head)
169 m = self.import_module(head, mname, m)
170 if not m:
171 self.msgout(4, "raise ImportError: No module named", mname)
172 raise ImportError, "No module named " + mname
173 self.msgout(4, "load_tail ->", m)
174 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000175
176 def ensure_fromlist(self, m, fromlist, recursive=0):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000177 self.msg(4, "ensure_fromlist", m, fromlist, recursive)
178 for sub in fromlist:
179 if sub == "*":
180 if not recursive:
181 all = self.find_all_submodules(m)
182 if all:
183 self.ensure_fromlist(m, all, 1)
184 elif not hasattr(m, sub):
185 subname = "%s.%s" % (m.__name__, sub)
186 submod = self.import_module(sub, subname, m)
187 if not submod:
188 raise ImportError, "No module named " + subname
Guido van Rossum75dc4961998-03-05 03:42:00 +0000189
190 def find_all_submodules(self, m):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000191 if not m.__path__:
192 return
193 modules = {}
194 suffixes = [".py", ".pyc", ".pyo"]
195 for dir in m.__path__:
196 try:
197 names = os.listdir(dir)
198 except os.error:
199 self.msg(2, "can't list directory", dir)
200 continue
201 for name in names:
202 mod = None
203 for suff in suffixes:
204 n = len(suff)
205 if name[-n:] == suff:
206 mod = name[:-n]
207 break
208 if mod and mod != "__init__":
209 modules[mod] = mod
210 return modules.keys()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000211
212 def import_module(self, partname, fqname, parent):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000213 self.msgin(3, "import_module", partname, fqname, parent)
214 try:
215 m = self.modules[fqname]
216 except KeyError:
217 pass
218 else:
219 self.msgout(3, "import_module ->", m)
220 return m
221 if self.badmodules.has_key(fqname):
222 self.msgout(3, "import_module -> None")
Guido van Rossum8b4b46e1999-11-02 15:46:44 +0000223 if parent:
224 self.badmodules[fqname][parent.__name__] = None
Guido van Rossum912a14c1998-03-05 04:56:37 +0000225 return None
226 try:
227 fp, pathname, stuff = self.find_module(partname,
228 parent and parent.__path__)
229 except ImportError:
230 self.msgout(3, "import_module ->", None)
231 return None
232 try:
233 m = self.load_module(fqname, fp, pathname, stuff)
234 finally:
235 if fp: fp.close()
236 if parent:
237 setattr(parent, partname, m)
238 self.msgout(3, "import_module ->", m)
239 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000240
241 def load_module(self, fqname, fp, pathname, (suffix, mode, type)):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000242 self.msgin(2, "load_module", fqname, fp and "fp", pathname)
243 if type == imp.PKG_DIRECTORY:
244 m = self.load_package(fqname, pathname)
245 self.msgout(2, "load_module ->", m)
246 return m
247 if type == imp.PY_SOURCE:
Guido van Rossum78fc3631998-03-20 17:37:24 +0000248 co = compile(fp.read()+'\n', pathname, 'exec')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000249 elif type == imp.PY_COMPILED:
250 if fp.read(4) != imp.get_magic():
251 self.msgout(2, "raise ImportError: Bad magic number", pathname)
Guido van Rossumce33eb32000-05-02 13:49:13 +0000252 raise ImportError, "Bad magic number in %s" % pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000253 fp.read(4)
254 co = marshal.load(fp)
255 else:
256 co = None
257 m = self.add_module(fqname)
Guido van Rossumab045f91998-03-06 19:55:10 +0000258 m.__file__ = pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000259 if co:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000260 m.__code__ = co
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000261 self.scan_code(co, m)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000262 self.msgout(2, "load_module ->", m)
263 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000264
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000265 def scan_code(self, co, m):
266 code = co.co_code
267 n = len(code)
268 i = 0
269 lastname = None
270 while i < n:
271 c = code[i]
272 i = i+1
273 op = ord(c)
274 if op >= dis.HAVE_ARGUMENT:
275 oparg = ord(code[i]) + ord(code[i+1])*256
276 i = i+2
277 if op == IMPORT_NAME:
278 name = lastname = co.co_names[oparg]
279 if not self.badmodules.has_key(lastname):
280 try:
281 self.import_hook(name, m)
282 except ImportError, msg:
283 self.msg(2, "ImportError:", str(msg))
Guido van Rossumfdd30281998-12-22 13:44:01 +0000284 if not self.badmodules.has_key(name):
285 self.badmodules[name] = {}
286 self.badmodules[name][m.__name__] = None
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000287 elif op == IMPORT_FROM:
288 name = co.co_names[oparg]
289 assert lastname is not None
290 if not self.badmodules.has_key(lastname):
291 try:
292 self.import_hook(lastname, m, [name])
293 except ImportError, msg:
294 self.msg(2, "ImportError:", str(msg))
295 fullname = lastname + "." + name
Guido van Rossumfdd30281998-12-22 13:44:01 +0000296 if not self.badmodules.has_key(fullname):
297 self.badmodules[fullname] = {}
298 self.badmodules[fullname][m.__name__] = None
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000299 else:
300 lastname = None
301 for c in co.co_consts:
302 if isinstance(c, type(co)):
303 self.scan_code(c, m)
304
Guido van Rossum75dc4961998-03-05 03:42:00 +0000305 def load_package(self, fqname, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000306 self.msgin(2, "load_package", fqname, pathname)
307 m = self.add_module(fqname)
308 m.__file__ = pathname
309 m.__path__ = [pathname]
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000310
Guido van Rossume7e632a1998-09-14 16:02:28 +0000311 # As per comment at top of file, simulate runtime __path__ additions.
312 m.__path__ = m.__path__ + packagePathMap.get(fqname, [])
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000313
Guido van Rossum912a14c1998-03-05 04:56:37 +0000314 fp, buf, stuff = self.find_module("__init__", m.__path__)
315 self.load_module(fqname, fp, buf, stuff)
316 self.msgout(2, "load_package ->", m)
317 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000318
319 def add_module(self, fqname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000320 if self.modules.has_key(fqname):
321 return self.modules[fqname]
322 self.modules[fqname] = m = Module(fqname)
323 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000324
325 def find_module(self, name, path):
Guido van Rossum78fc3631998-03-20 17:37:24 +0000326 if name in self.excludes:
327 self.msgout(3, "find_module -> Excluded")
328 raise ImportError, name
329
Guido van Rossum912a14c1998-03-05 04:56:37 +0000330 if path is None:
331 if name in sys.builtin_module_names:
332 return (None, None, ("", "", imp.C_BUILTIN))
Guido van Rossum78fc3631998-03-20 17:37:24 +0000333
334 # Emulate the Registered Module support on Windows.
335 if sys.platform=="win32" and win32api is not None:
336 HKEY_LOCAL_MACHINE = 0x80000002
337 try:
338 pathname = win32api.RegQueryValue(HKEY_LOCAL_MACHINE, "Software\\Python\\PythonCore\\%s\\Modules\\%s" % (sys.winver, name))
339 fp = open(pathname, "rb")
340 # XXX - To do - remove the hard code of C_EXTENSION.
341 stuff = "", "rb", imp.C_EXTENSION
342 return fp, pathname, stuff
343 except win32api.error:
344 pass
345
Guido van Rossum912a14c1998-03-05 04:56:37 +0000346 path = self.path
347 return imp.find_module(name, path)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000348
349 def report(self):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000350 print
351 print " %-25s %s" % ("Name", "File")
352 print " %-25s %s" % ("----", "----")
353 # Print modules found
354 keys = self.modules.keys()
355 keys.sort()
356 for key in keys:
357 m = self.modules[key]
358 if m.__path__:
359 print "P",
360 else:
361 print "m",
362 print "%-25s" % key, m.__file__ or ""
Guido van Rossum75dc4961998-03-05 03:42:00 +0000363
Guido van Rossum912a14c1998-03-05 04:56:37 +0000364 # Print missing modules
365 keys = self.badmodules.keys()
366 keys.sort()
367 for key in keys:
Thomas Wouters7e474022000-07-16 12:04:32 +0000368 # ... but not if they were explicitly excluded.
Guido van Rossumbaf06031998-08-25 14:06:55 +0000369 if key not in self.excludes:
Guido van Rossum38b92eb1998-12-15 15:35:23 +0000370 mods = self.badmodules[key].keys()
371 mods.sort()
372 print "?", key, "from", string.join(mods, ', ')
Guido van Rossum75dc4961998-03-05 03:42:00 +0000373
374
375def test():
376 # Parse command line
377 import getopt
378 try:
Guido van Rossumbaf06031998-08-25 14:06:55 +0000379 opts, args = getopt.getopt(sys.argv[1:], "dmp:qx:")
Guido van Rossum75dc4961998-03-05 03:42:00 +0000380 except getopt.error, msg:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000381 print msg
382 return
Guido van Rossum75dc4961998-03-05 03:42:00 +0000383
384 # Process options
385 debug = 1
386 domods = 0
387 addpath = []
Guido van Rossumbaf06031998-08-25 14:06:55 +0000388 exclude = []
Guido van Rossum75dc4961998-03-05 03:42:00 +0000389 for o, a in opts:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000390 if o == '-d':
391 debug = debug + 1
392 if o == '-m':
393 domods = 1
394 if o == '-p':
395 addpath = addpath + string.split(a, os.pathsep)
396 if o == '-q':
397 debug = 0
Guido van Rossumbaf06031998-08-25 14:06:55 +0000398 if o == '-x':
399 exclude.append(a)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000400
401 # Provide default arguments
402 if not args:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000403 script = "hello.py"
Guido van Rossum75dc4961998-03-05 03:42:00 +0000404 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000405 script = args[0]
Guido van Rossum75dc4961998-03-05 03:42:00 +0000406
407 # Set the path based on sys.path and the script directory
408 path = sys.path[:]
409 path[0] = os.path.dirname(script)
410 path = addpath + path
411 if debug > 1:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000412 print "path:"
413 for item in path:
414 print " ", `item`
Guido van Rossum75dc4961998-03-05 03:42:00 +0000415
416 # Create the module finder and turn its crank
Guido van Rossumbaf06031998-08-25 14:06:55 +0000417 mf = ModuleFinder(path, debug, exclude)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000418 for arg in args[1:]:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000419 if arg == '-m':
420 domods = 1
421 continue
422 if domods:
423 if arg[-2:] == '.*':
424 mf.import_hook(arg[:-2], None, ["*"])
425 else:
426 mf.import_hook(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000427 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000428 mf.load_file(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000429 mf.run_script(script)
430 mf.report()
431
432
433if __name__ == '__main__':
434 try:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000435 test()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000436 except KeyboardInterrupt:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000437 print "\n[interrupt]"