blob: 8c4b74d442f974cf7b0b69765cf7f98b618659b7 [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 Rossum75dc4961998-03-05 03:42:00 +000011IMPORT_NAME = dis.opname.index('IMPORT_NAME')
12IMPORT_FROM = dis.opname.index('IMPORT_FROM')
Guido van Rossum89990532000-09-15 16:37:42 +000013STORE_NAME = dis.opname.index('STORE_NAME')
14STORE_FAST = dis.opname.index('STORE_FAST')
15STORE_GLOBAL = dis.opname.index('STORE_GLOBAL')
16STORE_OPS = [STORE_NAME, STORE_FAST, STORE_GLOBAL]
Guido van Rossum75dc4961998-03-05 03:42:00 +000017
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000018# Modulefinder does a good job at simulating Python's, but it can not
19# handle __path__ modifications packages make at runtime. Therefore there
20# is a mechanism whereby you can register extra paths in this map for a
Thomas Wouters7e474022000-07-16 12:04:32 +000021# package, and it will be honored.
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000022
23# Note this is a mapping is lists of paths.
24packagePathMap = {}
25
26# A Public interface
27def AddPackagePath(packagename, path):
Guido van Rossume7e632a1998-09-14 16:02:28 +000028 paths = packagePathMap.get(packagename, [])
29 paths.append(path)
30 packagePathMap[packagename] = paths
Guido van Rossum75dc4961998-03-05 03:42:00 +000031
32class Module:
33
34 def __init__(self, name, file=None, path=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +000035 self.__name__ = name
36 self.__file__ = file
37 self.__path__ = path
38 self.__code__ = None
Guido van Rossum75dc4961998-03-05 03:42:00 +000039
40 def __repr__(self):
Guido van Rossum912a14c1998-03-05 04:56:37 +000041 s = "Module(%s" % `self.__name__`
42 if self.__file__ is not None:
43 s = s + ", %s" % `self.__file__`
44 if self.__path__ is not None:
45 s = s + ", %s" % `self.__path__`
46 s = s + ")"
47 return s
Guido van Rossum75dc4961998-03-05 03:42:00 +000048
49
50class ModuleFinder:
51
Guido van Rossum78fc3631998-03-20 17:37:24 +000052 def __init__(self, path=None, debug=0, excludes = []):
Guido van Rossum912a14c1998-03-05 04:56:37 +000053 if path is None:
54 path = sys.path
55 self.path = path
56 self.modules = {}
57 self.badmodules = {}
58 self.debug = debug
59 self.indent = 0
Guido van Rossum78fc3631998-03-20 17:37:24 +000060 self.excludes = excludes
Guido van Rossum75dc4961998-03-05 03:42:00 +000061
62 def msg(self, level, str, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000063 if level <= self.debug:
64 for i in range(self.indent):
65 print " ",
66 print str,
67 for arg in args:
68 print repr(arg),
69 print
Guido van Rossum75dc4961998-03-05 03:42:00 +000070
71 def msgin(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000072 level = args[0]
73 if level <= self.debug:
74 self.indent = self.indent + 1
75 apply(self.msg, args)
Guido van Rossum75dc4961998-03-05 03:42:00 +000076
77 def msgout(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000078 level = args[0]
79 if level <= self.debug:
80 self.indent = self.indent - 1
81 apply(self.msg, args)
Guido van Rossum75dc4961998-03-05 03:42:00 +000082
83 def run_script(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +000084 self.msg(2, "run_script", pathname)
85 fp = open(pathname)
86 stuff = ("", "r", imp.PY_SOURCE)
87 self.load_module('__main__', fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +000088
89 def load_file(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +000090 dir, name = os.path.split(pathname)
91 name, ext = os.path.splitext(name)
92 fp = open(pathname)
93 stuff = (ext, "r", imp.PY_SOURCE)
94 self.load_module(name, fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +000095
96 def import_hook(self, name, caller=None, fromlist=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +000097 self.msg(3, "import_hook", name, caller, fromlist)
98 parent = self.determine_parent(caller)
99 q, tail = self.find_head_package(parent, name)
100 m = self.load_tail(q, tail)
101 if not fromlist:
102 return q
103 if m.__path__:
104 self.ensure_fromlist(m, fromlist)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000105
106 def determine_parent(self, caller):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000107 self.msgin(4, "determine_parent", caller)
108 if not caller:
109 self.msgout(4, "determine_parent -> None")
110 return None
111 pname = caller.__name__
112 if caller.__path__:
113 parent = self.modules[pname]
114 assert caller is parent
115 self.msgout(4, "determine_parent ->", parent)
116 return parent
117 if '.' in pname:
118 i = string.rfind(pname, '.')
119 pname = pname[:i]
120 parent = self.modules[pname]
121 assert parent.__name__ == pname
122 self.msgout(4, "determine_parent ->", parent)
123 return parent
124 self.msgout(4, "determine_parent -> None")
125 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000126
127 def find_head_package(self, parent, name):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000128 self.msgin(4, "find_head_package", parent, name)
129 if '.' in name:
130 i = string.find(name, '.')
131 head = name[:i]
132 tail = name[i+1:]
133 else:
134 head = name
135 tail = ""
136 if parent:
137 qname = "%s.%s" % (parent.__name__, head)
138 else:
139 qname = head
140 q = self.import_module(head, qname, parent)
141 if q:
142 self.msgout(4, "find_head_package ->", (q, tail))
143 return q, tail
144 if parent:
145 qname = head
146 parent = None
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 self.msgout(4, "raise ImportError: No module named", qname)
152 raise ImportError, "No module named " + qname
Guido van Rossum75dc4961998-03-05 03:42:00 +0000153
154 def load_tail(self, q, tail):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000155 self.msgin(4, "load_tail", q, tail)
156 m = q
157 while tail:
158 i = string.find(tail, '.')
159 if i < 0: i = len(tail)
160 head, tail = tail[:i], tail[i+1:]
161 mname = "%s.%s" % (m.__name__, head)
162 m = self.import_module(head, mname, m)
163 if not m:
164 self.msgout(4, "raise ImportError: No module named", mname)
165 raise ImportError, "No module named " + mname
166 self.msgout(4, "load_tail ->", m)
167 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000168
169 def ensure_fromlist(self, m, fromlist, recursive=0):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000170 self.msg(4, "ensure_fromlist", m, fromlist, recursive)
171 for sub in fromlist:
172 if sub == "*":
173 if not recursive:
174 all = self.find_all_submodules(m)
175 if all:
176 self.ensure_fromlist(m, all, 1)
177 elif not hasattr(m, sub):
178 subname = "%s.%s" % (m.__name__, sub)
179 submod = self.import_module(sub, subname, m)
180 if not submod:
181 raise ImportError, "No module named " + subname
Guido van Rossum75dc4961998-03-05 03:42:00 +0000182
183 def find_all_submodules(self, m):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000184 if not m.__path__:
185 return
186 modules = {}
187 suffixes = [".py", ".pyc", ".pyo"]
188 for dir in m.__path__:
189 try:
190 names = os.listdir(dir)
191 except os.error:
192 self.msg(2, "can't list directory", dir)
193 continue
194 for name in names:
195 mod = None
196 for suff in suffixes:
197 n = len(suff)
198 if name[-n:] == suff:
199 mod = name[:-n]
200 break
201 if mod and mod != "__init__":
202 modules[mod] = mod
203 return modules.keys()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000204
205 def import_module(self, partname, fqname, parent):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000206 self.msgin(3, "import_module", partname, fqname, parent)
207 try:
208 m = self.modules[fqname]
209 except KeyError:
210 pass
211 else:
212 self.msgout(3, "import_module ->", m)
213 return m
214 if self.badmodules.has_key(fqname):
215 self.msgout(3, "import_module -> None")
Guido van Rossum8b4b46e1999-11-02 15:46:44 +0000216 if parent:
217 self.badmodules[fqname][parent.__name__] = None
Guido van Rossum912a14c1998-03-05 04:56:37 +0000218 return None
219 try:
220 fp, pathname, stuff = self.find_module(partname,
221 parent and parent.__path__)
222 except ImportError:
223 self.msgout(3, "import_module ->", None)
224 return None
225 try:
226 m = self.load_module(fqname, fp, pathname, stuff)
227 finally:
228 if fp: fp.close()
229 if parent:
230 setattr(parent, partname, m)
231 self.msgout(3, "import_module ->", m)
232 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000233
234 def load_module(self, fqname, fp, pathname, (suffix, mode, type)):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000235 self.msgin(2, "load_module", fqname, fp and "fp", pathname)
236 if type == imp.PKG_DIRECTORY:
237 m = self.load_package(fqname, pathname)
238 self.msgout(2, "load_module ->", m)
239 return m
240 if type == imp.PY_SOURCE:
Guido van Rossum78fc3631998-03-20 17:37:24 +0000241 co = compile(fp.read()+'\n', pathname, 'exec')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000242 elif type == imp.PY_COMPILED:
243 if fp.read(4) != imp.get_magic():
244 self.msgout(2, "raise ImportError: Bad magic number", pathname)
Guido van Rossumce33eb32000-05-02 13:49:13 +0000245 raise ImportError, "Bad magic number in %s" % pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000246 fp.read(4)
247 co = marshal.load(fp)
248 else:
249 co = None
250 m = self.add_module(fqname)
Guido van Rossumab045f91998-03-06 19:55:10 +0000251 m.__file__ = pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000252 if co:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000253 m.__code__ = co
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000254 self.scan_code(co, m)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000255 self.msgout(2, "load_module ->", m)
256 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000257
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000258 def scan_code(self, co, m):
259 code = co.co_code
260 n = len(code)
261 i = 0
262 lastname = None
263 while i < n:
264 c = code[i]
265 i = i+1
266 op = ord(c)
267 if op >= dis.HAVE_ARGUMENT:
268 oparg = ord(code[i]) + ord(code[i+1])*256
269 i = i+2
270 if op == IMPORT_NAME:
271 name = lastname = co.co_names[oparg]
272 if not self.badmodules.has_key(lastname):
273 try:
274 self.import_hook(name, m)
275 except ImportError, msg:
276 self.msg(2, "ImportError:", str(msg))
Guido van Rossumfdd30281998-12-22 13:44:01 +0000277 if not self.badmodules.has_key(name):
278 self.badmodules[name] = {}
279 self.badmodules[name][m.__name__] = None
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000280 elif op == IMPORT_FROM:
281 name = co.co_names[oparg]
282 assert lastname is not None
283 if not self.badmodules.has_key(lastname):
284 try:
285 self.import_hook(lastname, m, [name])
286 except ImportError, msg:
287 self.msg(2, "ImportError:", str(msg))
288 fullname = lastname + "." + name
Guido van Rossumfdd30281998-12-22 13:44:01 +0000289 if not self.badmodules.has_key(fullname):
290 self.badmodules[fullname] = {}
291 self.badmodules[fullname][m.__name__] = None
Guido van Rossum89990532000-09-15 16:37:42 +0000292 elif op in STORE_OPS:
293 # Skip; each IMPORT_FROM is followed by a STORE_* opcode
294 pass
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000295 else:
296 lastname = None
297 for c in co.co_consts:
298 if isinstance(c, type(co)):
299 self.scan_code(c, m)
300
Guido van Rossum75dc4961998-03-05 03:42:00 +0000301 def load_package(self, fqname, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000302 self.msgin(2, "load_package", fqname, pathname)
303 m = self.add_module(fqname)
304 m.__file__ = pathname
305 m.__path__ = [pathname]
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000306
Guido van Rossume7e632a1998-09-14 16:02:28 +0000307 # As per comment at top of file, simulate runtime __path__ additions.
308 m.__path__ = m.__path__ + packagePathMap.get(fqname, [])
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000309
Guido van Rossum912a14c1998-03-05 04:56:37 +0000310 fp, buf, stuff = self.find_module("__init__", m.__path__)
311 self.load_module(fqname, fp, buf, stuff)
312 self.msgout(2, "load_package ->", m)
313 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000314
315 def add_module(self, fqname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000316 if self.modules.has_key(fqname):
317 return self.modules[fqname]
318 self.modules[fqname] = m = Module(fqname)
319 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000320
321 def find_module(self, name, path):
Guido van Rossum78fc3631998-03-20 17:37:24 +0000322 if name in self.excludes:
323 self.msgout(3, "find_module -> Excluded")
324 raise ImportError, name
325
Guido van Rossum912a14c1998-03-05 04:56:37 +0000326 if path is None:
327 if name in sys.builtin_module_names:
328 return (None, None, ("", "", imp.C_BUILTIN))
Guido van Rossum78fc3631998-03-20 17:37:24 +0000329
330 # Emulate the Registered Module support on Windows.
Mark Hammondd3d7bb12000-11-06 02:49:27 +0000331 if sys.platform=="win32":
332 import _winreg
333 from _winreg import HKEY_LOCAL_MACHINE
Guido van Rossum78fc3631998-03-20 17:37:24 +0000334 try:
Mark Hammondd3d7bb12000-11-06 02:49:27 +0000335 pathname = _winreg.QueryValueEx(HKEY_LOCAL_MACHINE, \
336 "Software\\Python\\PythonCore\\%s\\Modules\\%s" % (sys.winver, name))
Guido van Rossum78fc3631998-03-20 17:37:24 +0000337 fp = open(pathname, "rb")
338 # XXX - To do - remove the hard code of C_EXTENSION.
339 stuff = "", "rb", imp.C_EXTENSION
340 return fp, pathname, stuff
Mark Hammondd3d7bb12000-11-06 02:49:27 +0000341 except _winreg.error:
Guido van Rossum78fc3631998-03-20 17:37:24 +0000342 pass
343
Guido van Rossum912a14c1998-03-05 04:56:37 +0000344 path = self.path
345 return imp.find_module(name, path)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000346
347 def report(self):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000348 print
349 print " %-25s %s" % ("Name", "File")
350 print " %-25s %s" % ("----", "----")
351 # Print modules found
352 keys = self.modules.keys()
353 keys.sort()
354 for key in keys:
355 m = self.modules[key]
356 if m.__path__:
357 print "P",
358 else:
359 print "m",
360 print "%-25s" % key, m.__file__ or ""
Guido van Rossum75dc4961998-03-05 03:42:00 +0000361
Guido van Rossum912a14c1998-03-05 04:56:37 +0000362 # Print missing modules
363 keys = self.badmodules.keys()
364 keys.sort()
365 for key in keys:
Thomas Wouters7e474022000-07-16 12:04:32 +0000366 # ... but not if they were explicitly excluded.
Guido van Rossumbaf06031998-08-25 14:06:55 +0000367 if key not in self.excludes:
Guido van Rossum38b92eb1998-12-15 15:35:23 +0000368 mods = self.badmodules[key].keys()
369 mods.sort()
370 print "?", key, "from", string.join(mods, ', ')
Guido van Rossum75dc4961998-03-05 03:42:00 +0000371
372
373def test():
374 # Parse command line
375 import getopt
376 try:
Guido van Rossumbaf06031998-08-25 14:06:55 +0000377 opts, args = getopt.getopt(sys.argv[1:], "dmp:qx:")
Guido van Rossum75dc4961998-03-05 03:42:00 +0000378 except getopt.error, msg:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000379 print msg
380 return
Guido van Rossum75dc4961998-03-05 03:42:00 +0000381
382 # Process options
383 debug = 1
384 domods = 0
385 addpath = []
Guido van Rossumbaf06031998-08-25 14:06:55 +0000386 exclude = []
Guido van Rossum75dc4961998-03-05 03:42:00 +0000387 for o, a in opts:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000388 if o == '-d':
389 debug = debug + 1
390 if o == '-m':
391 domods = 1
392 if o == '-p':
393 addpath = addpath + string.split(a, os.pathsep)
394 if o == '-q':
395 debug = 0
Guido van Rossumbaf06031998-08-25 14:06:55 +0000396 if o == '-x':
397 exclude.append(a)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000398
399 # Provide default arguments
400 if not args:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000401 script = "hello.py"
Guido van Rossum75dc4961998-03-05 03:42:00 +0000402 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000403 script = args[0]
Guido van Rossum75dc4961998-03-05 03:42:00 +0000404
405 # Set the path based on sys.path and the script directory
406 path = sys.path[:]
407 path[0] = os.path.dirname(script)
408 path = addpath + path
409 if debug > 1:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000410 print "path:"
411 for item in path:
412 print " ", `item`
Guido van Rossum75dc4961998-03-05 03:42:00 +0000413
414 # Create the module finder and turn its crank
Guido van Rossumbaf06031998-08-25 14:06:55 +0000415 mf = ModuleFinder(path, debug, exclude)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000416 for arg in args[1:]:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000417 if arg == '-m':
418 domods = 1
419 continue
420 if domods:
421 if arg[-2:] == '.*':
422 mf.import_hook(arg[:-2], None, ["*"])
423 else:
424 mf.import_hook(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000425 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000426 mf.load_file(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000427 mf.run_script(script)
428 mf.report()
429
430
431if __name__ == '__main__':
432 try:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000433 test()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000434 except KeyboardInterrupt:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000435 print "\n[interrupt]"