Lib/modulefinder.py

   1 """Find modules used by a script, using introspection."""
   2 # This module should be kept compatible with Python 2.2, see PEP 291.
   3
   4 from __future__ import generators
   5 import dis
   6 import imp
   7 import marshal
   8 import os
   9 import sys
  10 import types
  11 import struct
  12
  13 if hasattr(sys.__stdout__, "newlines"):
  14     READ_MODE = "U"  # universal line endings
  15 else:
  16     # remain compatible with Python  < 2.3
  17     READ_MODE = "r"
  18
  19 LOAD_CONST = chr(dis.opname.index('LOAD_CONST'))
  20 IMPORT_NAME = chr(dis.opname.index('IMPORT_NAME'))
  21 STORE_NAME = chr(dis.opname.index('STORE_NAME'))
  22 STORE_GLOBAL = chr(dis.opname.index('STORE_GLOBAL'))
  23 STORE_OPS = [STORE_NAME, STORE_GLOBAL]
  24 HAVE_ARGUMENT = chr(dis.HAVE_ARGUMENT)
  25
  26 # Modulefinder does a good job at simulating Python's, but it can not
  27 # handle __path__ modifications packages make at runtime.  Therefore there
  28 # is a mechanism whereby you can register extra paths in this map for a
  29 # package, and it will be honored.
  30
  31 # Note this is a mapping is lists of paths.
  32 packagePathMap = {}
  33
  34 # A Public interface
  35 def AddPackagePath(packagename, path):
  36     paths = packagePathMap.get(packagename, [])
  37     paths.append(path)
  38     packagePathMap[packagename] = paths
  39
  40 replacePackageMap = {}
  41
  42 # This ReplacePackage mechanism allows modulefinder to work around the
  43 # way the _xmlplus package injects itself under the name "xml" into
  44 # sys.modules at runtime by calling ReplacePackage("_xmlplus", "xml")
  45 # before running ModuleFinder.
  46
  47 def ReplacePackage(oldname, newname):
  48     replacePackageMap[oldname] = newname
  49
  50
  51 class Module:
  52
  53     def __init__(self, name, file=None, path=None):
  54         self.__name__ = name
  55         self.__file__ = file
  56         self.__path__ = path
  57         self.__code__ = None
  58         # The set of global names that are assigned to in the module.
  59         # This includes those names imported through starimports of
  60         # Python modules.
  61         self.globalnames = {}
  62         # The set of starimports this module did that could not be
  63         # resolved, ie. a starimport from a non-Python module.
  64         self.starimports = {}
  65
  66     def __repr__(self):
  67         s = "Module(%r" % (self.__name__,)
  68         if self.__file__ is not None:
  69             s = s + ", %r" % (self.__file__,)
  70         if self.__path__ is not None:
  71             s = s + ", %r" % (self.__path__,)
  72         s = s + ")"
  73         return s
  74
  75 class ModuleFinder:
  76
  77     def __init__(self, path=None, debug=0, excludes=[], replace_paths=[]):
  78         if path is None:
  79             path = sys.path
  80         self.path = path
  81         self.modules = {}
  82         self.badmodules = {}
  83         self.debug = debug
  84         self.indent = 0
  85         self.excludes = excludes
  86         self.replace_paths = replace_paths
  87         self.processed_paths = []   # Used in debugging only
  88
  89     def msg(self, level, str, *args):
  90         if level <= self.debug:
  91             for i in range(self.indent):
  92                 print "   ",
  93             print str,
  94             for arg in args:
  95                 print repr(arg),
  96             print
  97
  98     def msgin(self, *args):
  99         level = args[0]
 100         if level <= self.debug:
 101             self.indent = self.indent + 1
 102             self.msg(*args)
 103
 104     def msgout(self, *args):
 105         level = args[0]
 106         if level <= self.debug:
 107             self.indent = self.indent - 1
 108             self.msg(*args)
 109
 110     def run_script(self, pathname):
 111         self.msg(2, "run_script", pathname)
 112         fp = open(pathname, READ_MODE)
 113         stuff = ("", "r", imp.PY_SOURCE)
 114         self.load_module('__main__', fp, pathname, stuff)
 115
 116     def load_file(self, pathname):
 117         dir, name = os.path.split(pathname)
 118         name, ext = os.path.splitext(name)
 119         fp = open(pathname, READ_MODE)
 120         stuff = (ext, "r", imp.PY_SOURCE)
 121         self.load_module(name, fp, pathname, stuff)
 122
 123     def import_hook(self, name, caller=None, fromlist=None, level=-1):
 124         self.msg(3, "import_hook", name, caller, fromlist, level)
 125         parent = self.determine_parent(caller, level=level)
 126         q, tail = self.find_head_package(parent, name)
 127         m = self.load_tail(q, tail)
 128         if not fromlist:
 129             return q
 130         if m.__path__:
 131             self.ensure_fromlist(m, fromlist)
 132         return None
 133
 134     def determine_parent(self, caller, level=-1):
 135         self.msgin(4, "determine_parent", caller, level)
 136         if not caller or level == 0:
 137             self.msgout(4, "determine_parent -> None")
 138             return None
 139         pname = caller.__name__
 140         if level >= 1: # relative import
 141             if caller.__path__:
 142                 level -= 1
 143             if level == 0:
 144                 parent = self.modules[pname]
 145                 assert parent is caller
 146                 self.msgout(4, "determine_parent ->", parent)
 147                 return parent
 148             if pname.count(".") < level:
 149                 raise ImportError, "relative importpath too deep"
 150             pname = ".".join(pname.split(".")[:-level])
 151             parent = self.modules[pname]
 152             self.msgout(4, "determine_parent ->", parent)
 153             return parent
 154         if caller.__path__:
 155             parent = self.modules[pname]
 156             assert caller is parent
 157             self.msgout(4, "determine_parent ->", parent)
 158             return parent
 159         if '.' in pname:
 160             i = pname.rfind('.')
 161             pname = pname[:i]
 162             parent = self.modules[pname]
 163             assert parent.__name__ == pname
 164             self.msgout(4, "determine_parent ->", parent)
 165             return parent
 166         self.msgout(4, "determine_parent -> None")
 167         return None
 168
 169     def find_head_package(self, parent, name):
 170         self.msgin(4, "find_head_package", parent, name)
 171         if '.' in name:
 172             i = name.find('.')
 173             head = name[:i]
 174             tail = name[i+1:]
 175         else:
 176             head = name
 177             tail = ""
 178         if parent:
 179             qname = "%s.%s" % (parent.__name__, head)
 180         else:
 181             qname = head
 182         q = self.import_module(head, qname, parent)
 183         if q:
 184             self.msgout(4, "find_head_package ->", (q, tail))
 185             return q, tail
 186         if parent:
 187             qname = head
 188             parent = None
 189             q = self.import_module(head, qname, parent)
 190             if q:
 191                 self.msgout(4, "find_head_package ->", (q, tail))
 192                 return q, tail
 193         self.msgout(4, "raise ImportError: No module named", qname)
 194         raise ImportError, "No module named " + qname
 195
 196     def load_tail(self, q, tail):
 197         self.msgin(4, "load_tail", q, tail)
 198         m = q
 199         while tail:
 200             i = tail.find('.')
 201             if i < 0: i = len(tail)
 202             head, tail = tail[:i], tail[i+1:]
 203             mname = "%s.%s" % (m.__name__, head)
 204             m = self.import_module(head, mname, m)
 205             if not m:
 206                 self.msgout(4, "raise ImportError: No module named", mname)
 207                 raise ImportError, "No module named " + mname
 208         self.msgout(4, "load_tail ->", m)
 209         return m
 210
 211     def ensure_fromlist(self, m, fromlist, recursive=0):
 212         self.msg(4, "ensure_fromlist", m, fromlist, recursive)
 213         for sub in fromlist:
 214             if sub == "*":
 215                 if not recursive:
 216                     all = self.find_all_submodules(m)
 217                     if all:
 218                         self.ensure_fromlist(m, all, 1)
 219             elif not hasattr(m, sub):
 220                 subname = "%s.%s" % (m.__name__, sub)
 221                 submod = self.import_module(sub, subname, m)
 222                 if not submod:
 223                     raise ImportError, "No module named " + subname
 224
 225     def find_all_submodules(self, m):
 226         if not m.__path__:
 227             return
 228         modules = {}
 229         # 'suffixes' used to be a list hardcoded to [".py", ".pyc", ".pyo"].
 230         # But we must also collect Python extension modules - although
 231         # we cannot separate normal dlls from Python extensions.
 232         suffixes = []
 233         for triple in imp.get_suffixes():
 234             suffixes.append(triple[0])
 235         for dir in m.__path__:
 236             try:
 237                 names = os.listdir(dir)
 238             except os.error:
 239                 self.msg(2, "can't list directory", dir)
 240                 continue
 241             for name in names:
 242                 mod = None
 243                 for suff in suffixes:
 244                     n = len(suff)
 245                     if name[-n:] == suff:
 246                         mod = name[:-n]
 247                         break
 248                 if mod and mod != "__init__":
 249                     modules[mod] = mod
 250         return modules.keys()
 251
 252     def import_module(self, partname, fqname, parent):
 253         self.msgin(3, "import_module", partname, fqname, parent)
 254         try:
 255             m = self.modules[fqname]
 256         except KeyError:
 257             pass
 258         else:
 259             self.msgout(3, "import_module ->", m)
 260             return m
 261         if fqname in self.badmodules:
 262             self.msgout(3, "import_module -> None")
 263             return None
 264         if parent and parent.__path__ is None:
 265             self.msgout(3, "import_module -> None")
 266             return None
 267         try:
 268             fp, pathname, stuff = self.find_module(partname,
 269                                                    parent and parent.__path__, parent)
 270         except ImportError:
 271             self.msgout(3, "import_module ->", None)
 272             return None
 273         try:
 274             m = self.load_module(fqname, fp, pathname, stuff)
 275         finally:
 276             if fp: fp.close()
 277         if parent:
 278             setattr(parent, partname, m)
 279         self.msgout(3, "import_module ->", m)
 280         return m
 281
 282     def load_module(self, fqname, fp, pathname, file_info):
 283         suffix, mode, type = file_info
 284         self.msgin(2, "load_module", fqname, fp and "fp", pathname)
 285         if type == imp.PKG_DIRECTORY:
 286             m = self.load_package(fqname, pathname)
 287             self.msgout(2, "load_module ->", m)
 288             return m
 289         if type == imp.PY_SOURCE:
 290             co = compile(fp.read()+'\n', pathname, 'exec')
 291         elif type == imp.PY_COMPILED:
 292             if fp.read(4) != imp.get_magic():
 293                 self.msgout(2, "raise ImportError: Bad magic number", pathname)
 294                 raise ImportError, "Bad magic number in %s" % pathname
 295             fp.read(4)
 296             co = marshal.load(fp)
 297         else:
 298             co = None
 299         m = self.add_module(fqname)
 300         m.__file__ = pathname
 301         if co:
 302             if self.replace_paths:
 303                 co = self.replace_paths_in_code(co)
 304             m.__code__ = co
 305             self.scan_code(co, m)
 306         self.msgout(2, "load_module ->", m)
 307         return m
 308
 309     def _add_badmodule(self, name, caller):
 310         if name not in self.badmodules:
 311             self.badmodules[name] = {}
 312         if caller:
 313             self.badmodules[name][caller.__name__] = 1
 314         else:
 315             self.badmodules[name]["-"] = 1
 316
 317     def _safe_import_hook(self, name, caller, fromlist, level=-1):
 318         # wrapper for self.import_hook() that won't raise ImportError
 319         if name in self.badmodules:
 320             self._add_badmodule(name, caller)
 321             return
 322         try:
 323             self.import_hook(name, caller, level=level)
 324         except ImportError, msg:
 325             self.msg(2, "ImportError:", str(msg))
 326             self._add_badmodule(name, caller)
 327         else:
 328             if fromlist:
 329                 for sub in fromlist:
 330                     if sub in self.badmodules:
 331                         self._add_badmodule(sub, caller)
 332                         continue
 333                     try:
 334                         self.import_hook(name, caller, [sub], level=level)
 335                     except ImportError, msg:
 336                         self.msg(2, "ImportError:", str(msg))
 337                         fullname = name + "." + sub
 338                         self._add_badmodule(fullname, caller)
 339
 340     def scan_opcodes(self, co,
 341                      unpack = struct.unpack):
 342         # Scan the code, and yield 'interesting' opcode combinations
 343         # Version for Python 2.4 and older
 344         code = co.co_code
 345         names = co.co_names
 346         consts = co.co_consts
 347         while code:
 348             c = code[0]
 349             if c in STORE_OPS:
 350                 oparg, = unpack('<H', code[1:3])
 351                 yield "store", (names[oparg],)
 352                 code = code[3:]
 353                 continue
 354             if c == LOAD_CONST and code[3] == IMPORT_NAME:
 355                 oparg_1, oparg_2 = unpack('<xHxH', code[:6])
 356                 yield "import", (consts[oparg_1], names[oparg_2])
 357                 code = code[6:]
 358                 continue
 359             if c >= HAVE_ARGUMENT:
 360                 code = code[3:]
 361             else:
 362                 code = code[1:]
 363
 364     def scan_opcodes_25(self, co,
 365                      unpack = struct.unpack):
 366         # Scan the code, and yield 'interesting' opcode combinations
 367         # Python 2.5 version (has absolute and relative imports)
 368         code = co.co_code
 369         names = co.co_names
 370         consts = co.co_consts
 371         LOAD_LOAD_AND_IMPORT = LOAD_CONST + LOAD_CONST + IMPORT_NAME
 372         while code:
 373             c = code[0]
 374             if c in STORE_OPS:
 375                 oparg, = unpack('<H', code[1:3])
 376                 yield "store", (names[oparg],)
 377                 code = code[3:]
 378                 continue
 379             if code[:9:3] == LOAD_LOAD_AND_IMPORT:
 380                 oparg_1, oparg_2, oparg_3 = unpack('<xHxHxH', code[:9])
 381                 level = consts[oparg_1]
 382                 if level == -1: # normal import
 383                     yield "import", (consts[oparg_2], names[oparg_3])
 384                 elif level == 0: # absolute import
 385                     yield "absolute_import", (consts[oparg_2], names[oparg_3])
 386                 else: # relative import
 387                     yield "relative_import", (level, consts[oparg_2], names[oparg_3])
 388                 code = code[9:]
 389                 continue
 390             if c >= HAVE_ARGUMENT:
 391                 code = code[3:]
 392             else:
 393                 code = code[1:]
 394
 395     def scan_code(self, co, m):
 396         code = co.co_code
 397         if sys.version_info >= (2, 5):
 398             scanner = self.scan_opcodes_25
 399         else:
 400             scanner = self.scan_opcodes
 401         for what, args in scanner(co):
 402             if what == "store":
 403                 name, = args
 404                 m.globalnames[name] = 1
 405             elif what in ("import", "absolute_import"):
 406                 fromlist, name = args
 407                 have_star = 0
 408                 if fromlist is not None:
 409                     if "*" in fromlist:
 410                         have_star = 1
 411                     fromlist = [f for f in fromlist if f != "*"]
 412                 if what == "absolute_import": level = 0
 413                 else: level = -1
 414                 self._safe_import_hook(name, m, fromlist, level=level)
 415                 if have_star:
 416                     # We've encountered an "import *". If it is a Python module,
 417                     # the code has already been parsed and we can suck out the
 418                     # global names.
 419                     mm = None
 420                     if m.__path__:
 421                         # At this point we don't know whether 'name' is a
 422                         # submodule of 'm' or a global module. Let's just try
 423                         # the full name first.
 424                         mm = self.modules.get(m.__name__ + "." + name)
 425                     if mm is None:
 426                         mm = self.modules.get(name)
 427                     if mm is not None:
 428                         m.globalnames.update(mm.globalnames)
 429                         m.starimports.update(mm.starimports)
 430                         if mm.__code__ is None:
 431                             m.starimports[name] = 1
 432                     else:
 433                         m.starimports[name] = 1
 434             elif what == "relative_import":
 435                 level, fromlist, name = args
 436                 if name:
 437                     self._safe_import_hook(name, m, fromlist, level=level)
 438                 else:
 439                     parent = self.determine_parent(m, level=level)
 440                     self._safe_import_hook(parent.__name__, None, fromlist, level=0)
 441             else:
 442                 # We don't expect anything else from the generator.
 443                 raise RuntimeError(what)
 444
 445         for c in co.co_consts:
 446             if isinstance(c, type(co)):
 447                 self.scan_code(c, m)
 448
 449     def load_package(self, fqname, pathname):
 450         self.msgin(2, "load_package", fqname, pathname)
 451         newname = replacePackageMap.get(fqname)
 452         if newname:
 453             fqname = newname
 454         m = self.add_module(fqname)
 455         m.__file__ = pathname
 456         m.__path__ = [pathname]
 457
 458         # As per comment at top of file, simulate runtime __path__ additions.
 459         m.__path__ = m.__path__ + packagePathMap.get(fqname, [])
 460
 461         fp, buf, stuff = self.find_module("__init__", m.__path__)
 462         self.load_module(fqname, fp, buf, stuff)
 463         self.msgout(2, "load_package ->", m)
 464         return m
 465
 466     def add_module(self, fqname):
 467         if fqname in self.modules:
 468             return self.modules[fqname]
 469         self.modules[fqname] = m = Module(fqname)
 470         return m
 471
 472     def find_module(self, name, path, parent=None):
 473         if parent is not None:
 474             # assert path is not None
 475             fullname = parent.__name__+'.'+name
 476         else:
 477             fullname = name
 478         if fullname in self.excludes:
 479             self.msgout(3, "find_module -> Excluded", fullname)
 480             raise ImportError, name
 481
 482         if path is None:
 483             if name in sys.builtin_module_names:
 484                 return (None, None, ("", "", imp.C_BUILTIN))
 485
 486             path = self.path
 487         return imp.find_module(name, path)
 488
 489     def report(self):
 490         """Print a report to stdout, listing the found modules with their
 491         paths, as well as modules that are missing, or seem to be missing.
 492         """
 493         print
 494         print "  %-25s %s" % ("Name", "File")
 495         print "  %-25s %s" % ("----", "----")
 496         # Print modules found
 497         keys = self.modules.keys()
 498         keys.sort()
 499         for key in keys:
 500             m = self.modules[key]
 501             if m.__path__:
 502                 print "P",
 503             else:
 504                 print "m",
 505             print "%-25s" % key, m.__file__ or ""
 506
 507         # Print missing modules
 508         missing, maybe = self.any_missing_maybe()
 509         if missing:
 510             print
 511             print "Missing modules:"
 512             for name in missing:
 513                 mods = self.badmodules[name].keys()
 514                 mods.sort()
 515                 print "?", name, "imported from", ', '.join(mods)
 516         # Print modules that may be missing, but then again, maybe not...
 517         if maybe:
 518             print
 519             print "Submodules thay appear to be missing, but could also be",
 520             print "global names in the parent package:"
 521             for name in maybe:
 522                 mods = self.badmodules[name].keys()
 523                 mods.sort()
 524                 print "?", name, "imported from", ', '.join(mods)
 525
 526     def any_missing(self):
 527         """Return a list of modules that appear to be missing. Use
 528         any_missing_maybe() if you want to know which modules are
 529         certain to be missing, and which *may* be missing.
 530         """
 531         missing, maybe = self.any_missing_maybe()
 532         return missing + maybe
 533
 534     def any_missing_maybe(self):
 535         """Return two lists, one with modules that are certainly missing
 536         and one with modules that *may* be missing. The latter names could
 537         either be submodules *or* just global names in the package.
 538
 539         The reason it can't always be determined is that it's impossible to
 540         tell which names are imported when "from module import *" is done
 541         with an extension module, short of actually importing it.
 542         """
 543         missing = []
 544         maybe = []
 545         for name in self.badmodules:
 546             if name in self.excludes:
 547                 continue
 548             i = name.rfind(".")
 549             if i < 0:
 550                 missing.append(name)
 551                 continue
 552             subname = name[i+1:]
 553             pkgname = name[:i]
 554             pkg = self.modules.get(pkgname)
 555             if pkg is not None:
 556                 if pkgname in self.badmodules[name]:
 557                     # The package tried to import this module itself and
 558                     # failed. It's definitely missing.
 559                     missing.append(name)
 560                 elif subname in pkg.globalnames:
 561                     # It's a global in the package: definitely not missing.
 562                     pass
 563                 elif pkg.starimports:
 564                     # It could be missing, but the package did an "import *"
 565                     # from a non-Python module, so we simply can't be sure.
 566                     maybe.append(name)
 567                 else:
 568                     # It's not a global in the package, the package didn't
 569                     # do funny star imports, it's very likely to be missing.
 570                     # The symbol could be inserted into the package from the
 571                     # outside, but since that's not good style we simply list
 572                     # it missing.
 573                     missing.append(name)
 574             else:
 575                 missing.append(name)
 576         missing.sort()
 577         maybe.sort()
 578         return missing, maybe
 579
 580     def replace_paths_in_code(self, co):
 581         new_filename = original_filename = os.path.normpath(co.co_filename)
 582         for f, r in self.replace_paths:
 583             if original_filename.startswith(f):
 584                 new_filename = r + original_filename[len(f):]
 585                 break
 586
 587         if self.debug and original_filename not in self.processed_paths:
 588             if new_filename != original_filename:
 589                 self.msgout(2, "co_filename %r changed to %r" \
 590                                     % (original_filename,new_filename,))
 591             else:
 592                 self.msgout(2, "co_filename %r remains unchanged" \
 593                                     % (original_filename,))
 594             self.processed_paths.append(original_filename)
 595
 596         consts = list(co.co_consts)
 597         for i in range(len(consts)):
 598             if isinstance(consts[i], type(co)):
 599                 consts[i] = self.replace_paths_in_code(consts[i])
 600
 601         return types.CodeType(co.co_argcount, co.co_nlocals, co.co_stacksize,
 602                          co.co_flags, co.co_code, tuple(consts), co.co_names,
 603                          co.co_varnames, new_filename, co.co_name,
 604                          co.co_firstlineno, co.co_lnotab,
 605                          co.co_freevars, co.co_cellvars)
 606
 607
 608 def test():
 609     # Parse command line
 610     import getopt
 611     try:
 612         opts, args = getopt.getopt(sys.argv[1:], "dmp:qx:")
 613     except getopt.error, msg:
 614         print msg
 615         return
 616
 617     # Process options
 618     debug = 1
 619     domods = 0
 620     addpath = []
 621     exclude = []
 622     for o, a in opts:
 623         if o == '-d':
 624             debug = debug + 1
 625         if o == '-m':
 626             domods = 1
 627         if o == '-p':
 628             addpath = addpath + a.split(os.pathsep)
 629         if o == '-q':
 630             debug = 0
 631         if o == '-x':
 632             exclude.append(a)
 633
 634     # Provide default arguments
 635     if not args:
 636         script = "hello.py"
 637     else:
 638         script = args[0]
 639
 640     # Set the path based on sys.path and the script directory
 641     path = sys.path[:]
 642     path[0] = os.path.dirname(script)
 643     path = addpath + path
 644     if debug > 1:
 645         print "path:"
 646         for item in path:
 647             print "   ", repr(item)
 648
 649     # Create the module finder and turn its crank
 650     mf = ModuleFinder(path, debug, exclude)
 651     for arg in args[1:]:
 652         if arg == '-m':
 653             domods = 1
 654             continue
 655         if domods:
 656             if arg[-2:] == '.*':
 657                 mf.import_hook(arg[:-2], None, ["*"])
 658             else:
 659                 mf.import_hook(arg)
 660         else:
 661             mf.load_file(arg)
 662     mf.run_script(script)
 663     mf.report()
 664     return mf  # for -i debugging
 665
 666
 667 if __name__ == '__main__':
 668     try:
 669         mf = test()
 670     except KeyboardInterrupt:
 671         print "\n[interrupt]"