blob: e84a54128f6f932c000fb35701c47ac72a9edb34 [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')
Guido van Rossum89990532000-09-15 16:37:42 +000024STORE_NAME = dis.opname.index('STORE_NAME')
25STORE_FAST = dis.opname.index('STORE_FAST')
26STORE_GLOBAL = dis.opname.index('STORE_GLOBAL')
27STORE_OPS = [STORE_NAME, STORE_FAST, STORE_GLOBAL]
Guido van Rossum75dc4961998-03-05 03:42:00 +000028
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000029# Modulefinder does a good job at simulating Python's, but it can not
30# handle __path__ modifications packages make at runtime. Therefore there
31# is a mechanism whereby you can register extra paths in this map for a
Thomas Wouters7e474022000-07-16 12:04:32 +000032# package, and it will be honored.
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +000033
34# Note this is a mapping is lists of paths.
35packagePathMap = {}
36
37# A Public interface
38def AddPackagePath(packagename, path):
Guido van Rossume7e632a1998-09-14 16:02:28 +000039 paths = packagePathMap.get(packagename, [])
40 paths.append(path)
41 packagePathMap[packagename] = paths
Guido van Rossum75dc4961998-03-05 03:42:00 +000042
43class 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
Guido van Rossum75dc4961998-03-05 03:42:00 +000050
51 def __repr__(self):
Guido van Rossum912a14c1998-03-05 04:56:37 +000052 s = "Module(%s" % `self.__name__`
53 if self.__file__ is not None:
54 s = s + ", %s" % `self.__file__`
55 if self.__path__ is not None:
56 s = s + ", %s" % `self.__path__`
57 s = s + ")"
58 return s
Guido van Rossum75dc4961998-03-05 03:42:00 +000059
60
61class ModuleFinder:
62
Guido van Rossum78fc3631998-03-20 17:37:24 +000063 def __init__(self, path=None, debug=0, excludes = []):
Guido van Rossum912a14c1998-03-05 04:56:37 +000064 if path is None:
65 path = sys.path
66 self.path = path
67 self.modules = {}
68 self.badmodules = {}
69 self.debug = debug
70 self.indent = 0
Guido van Rossum78fc3631998-03-20 17:37:24 +000071 self.excludes = excludes
Guido van Rossum75dc4961998-03-05 03:42:00 +000072
73 def msg(self, level, str, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000074 if level <= self.debug:
75 for i in range(self.indent):
76 print " ",
77 print str,
78 for arg in args:
79 print repr(arg),
80 print
Guido van Rossum75dc4961998-03-05 03:42:00 +000081
82 def msgin(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000083 level = args[0]
84 if level <= self.debug:
85 self.indent = self.indent + 1
86 apply(self.msg, args)
Guido van Rossum75dc4961998-03-05 03:42:00 +000087
88 def msgout(self, *args):
Guido van Rossum912a14c1998-03-05 04:56:37 +000089 level = args[0]
90 if level <= self.debug:
91 self.indent = self.indent - 1
92 apply(self.msg, args)
Guido van Rossum75dc4961998-03-05 03:42:00 +000093
94 def run_script(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +000095 self.msg(2, "run_script", pathname)
96 fp = open(pathname)
97 stuff = ("", "r", imp.PY_SOURCE)
98 self.load_module('__main__', fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +000099
100 def load_file(self, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000101 dir, name = os.path.split(pathname)
102 name, ext = os.path.splitext(name)
103 fp = open(pathname)
104 stuff = (ext, "r", imp.PY_SOURCE)
105 self.load_module(name, fp, pathname, stuff)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000106
107 def import_hook(self, name, caller=None, fromlist=None):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000108 self.msg(3, "import_hook", name, caller, fromlist)
109 parent = self.determine_parent(caller)
110 q, tail = self.find_head_package(parent, name)
111 m = self.load_tail(q, tail)
112 if not fromlist:
113 return q
114 if m.__path__:
115 self.ensure_fromlist(m, fromlist)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000116
117 def determine_parent(self, caller):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000118 self.msgin(4, "determine_parent", caller)
119 if not caller:
120 self.msgout(4, "determine_parent -> None")
121 return None
122 pname = caller.__name__
123 if caller.__path__:
124 parent = self.modules[pname]
125 assert caller is parent
126 self.msgout(4, "determine_parent ->", parent)
127 return parent
128 if '.' in pname:
129 i = string.rfind(pname, '.')
130 pname = pname[:i]
131 parent = self.modules[pname]
132 assert parent.__name__ == pname
133 self.msgout(4, "determine_parent ->", parent)
134 return parent
135 self.msgout(4, "determine_parent -> None")
136 return None
Guido van Rossum75dc4961998-03-05 03:42:00 +0000137
138 def find_head_package(self, parent, name):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000139 self.msgin(4, "find_head_package", parent, name)
140 if '.' in name:
141 i = string.find(name, '.')
142 head = name[:i]
143 tail = name[i+1:]
144 else:
145 head = name
146 tail = ""
147 if parent:
148 qname = "%s.%s" % (parent.__name__, head)
149 else:
150 qname = head
151 q = self.import_module(head, qname, parent)
152 if q:
153 self.msgout(4, "find_head_package ->", (q, tail))
154 return q, tail
155 if parent:
156 qname = head
157 parent = None
158 q = self.import_module(head, qname, parent)
159 if q:
160 self.msgout(4, "find_head_package ->", (q, tail))
161 return q, tail
162 self.msgout(4, "raise ImportError: No module named", qname)
163 raise ImportError, "No module named " + qname
Guido van Rossum75dc4961998-03-05 03:42:00 +0000164
165 def load_tail(self, q, tail):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000166 self.msgin(4, "load_tail", q, tail)
167 m = q
168 while tail:
169 i = string.find(tail, '.')
170 if i < 0: i = len(tail)
171 head, tail = tail[:i], tail[i+1:]
172 mname = "%s.%s" % (m.__name__, head)
173 m = self.import_module(head, mname, m)
174 if not m:
175 self.msgout(4, "raise ImportError: No module named", mname)
176 raise ImportError, "No module named " + mname
177 self.msgout(4, "load_tail ->", m)
178 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000179
180 def ensure_fromlist(self, m, fromlist, recursive=0):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000181 self.msg(4, "ensure_fromlist", m, fromlist, recursive)
182 for sub in fromlist:
183 if sub == "*":
184 if not recursive:
185 all = self.find_all_submodules(m)
186 if all:
187 self.ensure_fromlist(m, all, 1)
188 elif not hasattr(m, sub):
189 subname = "%s.%s" % (m.__name__, sub)
190 submod = self.import_module(sub, subname, m)
191 if not submod:
192 raise ImportError, "No module named " + subname
Guido van Rossum75dc4961998-03-05 03:42:00 +0000193
194 def find_all_submodules(self, m):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000195 if not m.__path__:
196 return
197 modules = {}
198 suffixes = [".py", ".pyc", ".pyo"]
199 for dir in m.__path__:
200 try:
201 names = os.listdir(dir)
202 except os.error:
203 self.msg(2, "can't list directory", dir)
204 continue
205 for name in names:
206 mod = None
207 for suff in suffixes:
208 n = len(suff)
209 if name[-n:] == suff:
210 mod = name[:-n]
211 break
212 if mod and mod != "__init__":
213 modules[mod] = mod
214 return modules.keys()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000215
216 def import_module(self, partname, fqname, parent):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000217 self.msgin(3, "import_module", partname, fqname, parent)
218 try:
219 m = self.modules[fqname]
220 except KeyError:
221 pass
222 else:
223 self.msgout(3, "import_module ->", m)
224 return m
225 if self.badmodules.has_key(fqname):
226 self.msgout(3, "import_module -> None")
Guido van Rossum8b4b46e1999-11-02 15:46:44 +0000227 if parent:
228 self.badmodules[fqname][parent.__name__] = None
Guido van Rossum912a14c1998-03-05 04:56:37 +0000229 return None
230 try:
231 fp, pathname, stuff = self.find_module(partname,
232 parent and parent.__path__)
233 except ImportError:
234 self.msgout(3, "import_module ->", None)
235 return None
236 try:
237 m = self.load_module(fqname, fp, pathname, stuff)
238 finally:
239 if fp: fp.close()
240 if parent:
241 setattr(parent, partname, m)
242 self.msgout(3, "import_module ->", m)
243 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000244
245 def load_module(self, fqname, fp, pathname, (suffix, mode, type)):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000246 self.msgin(2, "load_module", fqname, fp and "fp", pathname)
247 if type == imp.PKG_DIRECTORY:
248 m = self.load_package(fqname, pathname)
249 self.msgout(2, "load_module ->", m)
250 return m
251 if type == imp.PY_SOURCE:
Guido van Rossum78fc3631998-03-20 17:37:24 +0000252 co = compile(fp.read()+'\n', pathname, 'exec')
Guido van Rossum912a14c1998-03-05 04:56:37 +0000253 elif type == imp.PY_COMPILED:
254 if fp.read(4) != imp.get_magic():
255 self.msgout(2, "raise ImportError: Bad magic number", pathname)
Guido van Rossumce33eb32000-05-02 13:49:13 +0000256 raise ImportError, "Bad magic number in %s" % pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000257 fp.read(4)
258 co = marshal.load(fp)
259 else:
260 co = None
261 m = self.add_module(fqname)
Guido van Rossumab045f91998-03-06 19:55:10 +0000262 m.__file__ = pathname
Guido van Rossum912a14c1998-03-05 04:56:37 +0000263 if co:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000264 m.__code__ = co
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000265 self.scan_code(co, m)
Guido van Rossum912a14c1998-03-05 04:56:37 +0000266 self.msgout(2, "load_module ->", m)
267 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000268
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000269 def scan_code(self, co, m):
270 code = co.co_code
271 n = len(code)
272 i = 0
273 lastname = None
274 while i < n:
275 c = code[i]
276 i = i+1
277 op = ord(c)
278 if op >= dis.HAVE_ARGUMENT:
279 oparg = ord(code[i]) + ord(code[i+1])*256
280 i = i+2
281 if op == IMPORT_NAME:
282 name = lastname = co.co_names[oparg]
283 if not self.badmodules.has_key(lastname):
284 try:
285 self.import_hook(name, m)
286 except ImportError, msg:
287 self.msg(2, "ImportError:", str(msg))
Guido van Rossumfdd30281998-12-22 13:44:01 +0000288 if not self.badmodules.has_key(name):
289 self.badmodules[name] = {}
290 self.badmodules[name][m.__name__] = None
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000291 elif op == IMPORT_FROM:
292 name = co.co_names[oparg]
293 assert lastname is not None
294 if not self.badmodules.has_key(lastname):
295 try:
296 self.import_hook(lastname, m, [name])
297 except ImportError, msg:
298 self.msg(2, "ImportError:", str(msg))
299 fullname = lastname + "." + name
Guido van Rossumfdd30281998-12-22 13:44:01 +0000300 if not self.badmodules.has_key(fullname):
301 self.badmodules[fullname] = {}
302 self.badmodules[fullname][m.__name__] = None
Guido van Rossum89990532000-09-15 16:37:42 +0000303 elif op in STORE_OPS:
304 # Skip; each IMPORT_FROM is followed by a STORE_* opcode
305 pass
Guido van Rossum3c51cf21998-03-05 05:15:07 +0000306 else:
307 lastname = None
308 for c in co.co_consts:
309 if isinstance(c, type(co)):
310 self.scan_code(c, m)
311
Guido van Rossum75dc4961998-03-05 03:42:00 +0000312 def load_package(self, fqname, pathname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000313 self.msgin(2, "load_package", fqname, pathname)
314 m = self.add_module(fqname)
315 m.__file__ = pathname
316 m.__path__ = [pathname]
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000317
Guido van Rossume7e632a1998-09-14 16:02:28 +0000318 # As per comment at top of file, simulate runtime __path__ additions.
319 m.__path__ = m.__path__ + packagePathMap.get(fqname, [])
Guido van Rossumf1b5a0e1998-05-18 20:21:56 +0000320
Guido van Rossum912a14c1998-03-05 04:56:37 +0000321 fp, buf, stuff = self.find_module("__init__", m.__path__)
322 self.load_module(fqname, fp, buf, stuff)
323 self.msgout(2, "load_package ->", m)
324 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000325
326 def add_module(self, fqname):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000327 if self.modules.has_key(fqname):
328 return self.modules[fqname]
329 self.modules[fqname] = m = Module(fqname)
330 return m
Guido van Rossum75dc4961998-03-05 03:42:00 +0000331
332 def find_module(self, name, path):
Guido van Rossum78fc3631998-03-20 17:37:24 +0000333 if name in self.excludes:
334 self.msgout(3, "find_module -> Excluded")
335 raise ImportError, name
336
Guido van Rossum912a14c1998-03-05 04:56:37 +0000337 if path is None:
338 if name in sys.builtin_module_names:
339 return (None, None, ("", "", imp.C_BUILTIN))
Guido van Rossum78fc3631998-03-20 17:37:24 +0000340
341 # Emulate the Registered Module support on Windows.
342 if sys.platform=="win32" and win32api is not None:
343 HKEY_LOCAL_MACHINE = 0x80000002
344 try:
345 pathname = win32api.RegQueryValue(HKEY_LOCAL_MACHINE, "Software\\Python\\PythonCore\\%s\\Modules\\%s" % (sys.winver, name))
346 fp = open(pathname, "rb")
347 # XXX - To do - remove the hard code of C_EXTENSION.
348 stuff = "", "rb", imp.C_EXTENSION
349 return fp, pathname, stuff
350 except win32api.error:
351 pass
352
Guido van Rossum912a14c1998-03-05 04:56:37 +0000353 path = self.path
354 return imp.find_module(name, path)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000355
356 def report(self):
Guido van Rossum912a14c1998-03-05 04:56:37 +0000357 print
358 print " %-25s %s" % ("Name", "File")
359 print " %-25s %s" % ("----", "----")
360 # Print modules found
361 keys = self.modules.keys()
362 keys.sort()
363 for key in keys:
364 m = self.modules[key]
365 if m.__path__:
366 print "P",
367 else:
368 print "m",
369 print "%-25s" % key, m.__file__ or ""
Guido van Rossum75dc4961998-03-05 03:42:00 +0000370
Guido van Rossum912a14c1998-03-05 04:56:37 +0000371 # Print missing modules
372 keys = self.badmodules.keys()
373 keys.sort()
374 for key in keys:
Thomas Wouters7e474022000-07-16 12:04:32 +0000375 # ... but not if they were explicitly excluded.
Guido van Rossumbaf06031998-08-25 14:06:55 +0000376 if key not in self.excludes:
Guido van Rossum38b92eb1998-12-15 15:35:23 +0000377 mods = self.badmodules[key].keys()
378 mods.sort()
379 print "?", key, "from", string.join(mods, ', ')
Guido van Rossum75dc4961998-03-05 03:42:00 +0000380
381
382def test():
383 # Parse command line
384 import getopt
385 try:
Guido van Rossumbaf06031998-08-25 14:06:55 +0000386 opts, args = getopt.getopt(sys.argv[1:], "dmp:qx:")
Guido van Rossum75dc4961998-03-05 03:42:00 +0000387 except getopt.error, msg:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000388 print msg
389 return
Guido van Rossum75dc4961998-03-05 03:42:00 +0000390
391 # Process options
392 debug = 1
393 domods = 0
394 addpath = []
Guido van Rossumbaf06031998-08-25 14:06:55 +0000395 exclude = []
Guido van Rossum75dc4961998-03-05 03:42:00 +0000396 for o, a in opts:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000397 if o == '-d':
398 debug = debug + 1
399 if o == '-m':
400 domods = 1
401 if o == '-p':
402 addpath = addpath + string.split(a, os.pathsep)
403 if o == '-q':
404 debug = 0
Guido van Rossumbaf06031998-08-25 14:06:55 +0000405 if o == '-x':
406 exclude.append(a)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000407
408 # Provide default arguments
409 if not args:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000410 script = "hello.py"
Guido van Rossum75dc4961998-03-05 03:42:00 +0000411 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000412 script = args[0]
Guido van Rossum75dc4961998-03-05 03:42:00 +0000413
414 # Set the path based on sys.path and the script directory
415 path = sys.path[:]
416 path[0] = os.path.dirname(script)
417 path = addpath + path
418 if debug > 1:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000419 print "path:"
420 for item in path:
421 print " ", `item`
Guido van Rossum75dc4961998-03-05 03:42:00 +0000422
423 # Create the module finder and turn its crank
Guido van Rossumbaf06031998-08-25 14:06:55 +0000424 mf = ModuleFinder(path, debug, exclude)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000425 for arg in args[1:]:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000426 if arg == '-m':
427 domods = 1
428 continue
429 if domods:
430 if arg[-2:] == '.*':
431 mf.import_hook(arg[:-2], None, ["*"])
432 else:
433 mf.import_hook(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000434 else:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000435 mf.load_file(arg)
Guido van Rossum75dc4961998-03-05 03:42:00 +0000436 mf.run_script(script)
437 mf.report()
438
439
440if __name__ == '__main__':
441 try:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000442 test()
Guido van Rossum75dc4961998-03-05 03:42:00 +0000443 except KeyboardInterrupt:
Guido van Rossum912a14c1998-03-05 04:56:37 +0000444 print "\n[interrupt]"