Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 1 | """Find modules used by a script, using introspection.""" |
| 2 | |
| 3 | import dis |
| 4 | import imp |
| 5 | import marshal |
| 6 | import os |
| 7 | import re |
| 8 | import string |
| 9 | import sys |
| 10 | |
| 11 | |
| 12 | IMPORT_NAME = dis.opname.index('IMPORT_NAME') |
| 13 | IMPORT_FROM = dis.opname.index('IMPORT_FROM') |
| 14 | |
| 15 | |
| 16 | class Module: |
| 17 | |
| 18 | def __init__(self, name, file=None, path=None): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 19 | self.__name__ = name |
| 20 | self.__file__ = file |
| 21 | self.__path__ = path |
| 22 | self.__code__ = None |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 23 | |
| 24 | def __repr__(self): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 25 | s = "Module(%s" % `self.__name__` |
| 26 | if self.__file__ is not None: |
| 27 | s = s + ", %s" % `self.__file__` |
| 28 | if self.__path__ is not None: |
| 29 | s = s + ", %s" % `self.__path__` |
| 30 | s = s + ")" |
| 31 | return s |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 32 | |
| 33 | |
| 34 | class ModuleFinder: |
| 35 | |
| 36 | def __init__(self, path=None, debug=0): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 37 | if path is None: |
| 38 | path = sys.path |
| 39 | self.path = path |
| 40 | self.modules = {} |
| 41 | self.badmodules = {} |
| 42 | self.debug = debug |
| 43 | self.indent = 0 |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 44 | |
| 45 | def msg(self, level, str, *args): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 46 | if level <= self.debug: |
| 47 | for i in range(self.indent): |
| 48 | print " ", |
| 49 | print str, |
| 50 | for arg in args: |
| 51 | print repr(arg), |
| 52 | print |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 53 | |
| 54 | def msgin(self, *args): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 55 | level = args[0] |
| 56 | if level <= self.debug: |
| 57 | self.indent = self.indent + 1 |
| 58 | apply(self.msg, args) |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 59 | |
| 60 | def msgout(self, *args): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 61 | level = args[0] |
| 62 | if level <= self.debug: |
| 63 | self.indent = self.indent - 1 |
| 64 | apply(self.msg, args) |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 65 | |
| 66 | def run_script(self, pathname): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 67 | self.msg(2, "run_script", pathname) |
| 68 | fp = open(pathname) |
| 69 | stuff = ("", "r", imp.PY_SOURCE) |
| 70 | self.load_module('__main__', fp, pathname, stuff) |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 71 | |
| 72 | def load_file(self, pathname): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 73 | dir, name = os.path.split(pathname) |
| 74 | name, ext = os.path.splitext(name) |
| 75 | fp = open(pathname) |
| 76 | stuff = (ext, "r", imp.PY_SOURCE) |
| 77 | self.load_module(name, fp, pathname, stuff) |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 78 | |
| 79 | def import_hook(self, name, caller=None, fromlist=None): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 80 | self.msg(3, "import_hook", name, caller, fromlist) |
| 81 | parent = self.determine_parent(caller) |
| 82 | q, tail = self.find_head_package(parent, name) |
| 83 | m = self.load_tail(q, tail) |
| 84 | if not fromlist: |
| 85 | return q |
| 86 | if m.__path__: |
| 87 | self.ensure_fromlist(m, fromlist) |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 88 | |
| 89 | def determine_parent(self, caller): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 90 | self.msgin(4, "determine_parent", caller) |
| 91 | if not caller: |
| 92 | self.msgout(4, "determine_parent -> None") |
| 93 | return None |
| 94 | pname = caller.__name__ |
| 95 | if caller.__path__: |
| 96 | parent = self.modules[pname] |
| 97 | assert caller is parent |
| 98 | self.msgout(4, "determine_parent ->", parent) |
| 99 | return parent |
| 100 | if '.' in pname: |
| 101 | i = string.rfind(pname, '.') |
| 102 | pname = pname[:i] |
| 103 | parent = self.modules[pname] |
| 104 | assert parent.__name__ == pname |
| 105 | self.msgout(4, "determine_parent ->", parent) |
| 106 | return parent |
| 107 | self.msgout(4, "determine_parent -> None") |
| 108 | return None |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 109 | |
| 110 | def find_head_package(self, parent, name): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 111 | self.msgin(4, "find_head_package", parent, name) |
| 112 | if '.' in name: |
| 113 | i = string.find(name, '.') |
| 114 | head = name[:i] |
| 115 | tail = name[i+1:] |
| 116 | else: |
| 117 | head = name |
| 118 | tail = "" |
| 119 | if parent: |
| 120 | qname = "%s.%s" % (parent.__name__, head) |
| 121 | else: |
| 122 | qname = head |
| 123 | q = self.import_module(head, qname, parent) |
| 124 | if q: |
| 125 | self.msgout(4, "find_head_package ->", (q, tail)) |
| 126 | return q, tail |
| 127 | if parent: |
| 128 | qname = head |
| 129 | parent = None |
| 130 | q = self.import_module(head, qname, parent) |
| 131 | if q: |
| 132 | self.msgout(4, "find_head_package ->", (q, tail)) |
| 133 | return q, tail |
| 134 | self.msgout(4, "raise ImportError: No module named", qname) |
| 135 | raise ImportError, "No module named " + qname |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 136 | |
| 137 | def load_tail(self, q, tail): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 138 | self.msgin(4, "load_tail", q, tail) |
| 139 | m = q |
| 140 | while tail: |
| 141 | i = string.find(tail, '.') |
| 142 | if i < 0: i = len(tail) |
| 143 | head, tail = tail[:i], tail[i+1:] |
| 144 | mname = "%s.%s" % (m.__name__, head) |
| 145 | m = self.import_module(head, mname, m) |
| 146 | if not m: |
| 147 | self.msgout(4, "raise ImportError: No module named", mname) |
| 148 | raise ImportError, "No module named " + mname |
| 149 | self.msgout(4, "load_tail ->", m) |
| 150 | return m |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 151 | |
| 152 | def ensure_fromlist(self, m, fromlist, recursive=0): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 153 | self.msg(4, "ensure_fromlist", m, fromlist, recursive) |
| 154 | for sub in fromlist: |
| 155 | if sub == "*": |
| 156 | if not recursive: |
| 157 | all = self.find_all_submodules(m) |
| 158 | if all: |
| 159 | self.ensure_fromlist(m, all, 1) |
| 160 | elif not hasattr(m, sub): |
| 161 | subname = "%s.%s" % (m.__name__, sub) |
| 162 | submod = self.import_module(sub, subname, m) |
| 163 | if not submod: |
| 164 | raise ImportError, "No module named " + subname |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 165 | |
| 166 | def find_all_submodules(self, m): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 167 | if not m.__path__: |
| 168 | return |
| 169 | modules = {} |
| 170 | suffixes = [".py", ".pyc", ".pyo"] |
| 171 | for dir in m.__path__: |
| 172 | try: |
| 173 | names = os.listdir(dir) |
| 174 | except os.error: |
| 175 | self.msg(2, "can't list directory", dir) |
| 176 | continue |
| 177 | for name in names: |
| 178 | mod = None |
| 179 | for suff in suffixes: |
| 180 | n = len(suff) |
| 181 | if name[-n:] == suff: |
| 182 | mod = name[:-n] |
| 183 | break |
| 184 | if mod and mod != "__init__": |
| 185 | modules[mod] = mod |
| 186 | return modules.keys() |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 187 | |
| 188 | def import_module(self, partname, fqname, parent): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 189 | self.msgin(3, "import_module", partname, fqname, parent) |
| 190 | try: |
| 191 | m = self.modules[fqname] |
| 192 | except KeyError: |
| 193 | pass |
| 194 | else: |
| 195 | self.msgout(3, "import_module ->", m) |
| 196 | return m |
| 197 | if self.badmodules.has_key(fqname): |
| 198 | self.msgout(3, "import_module -> None") |
| 199 | return None |
| 200 | try: |
| 201 | fp, pathname, stuff = self.find_module(partname, |
| 202 | parent and parent.__path__) |
| 203 | except ImportError: |
| 204 | self.msgout(3, "import_module ->", None) |
| 205 | return None |
| 206 | try: |
| 207 | m = self.load_module(fqname, fp, pathname, stuff) |
| 208 | finally: |
| 209 | if fp: fp.close() |
| 210 | if parent: |
| 211 | setattr(parent, partname, m) |
| 212 | self.msgout(3, "import_module ->", m) |
| 213 | return m |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 214 | |
| 215 | def load_module(self, fqname, fp, pathname, (suffix, mode, type)): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 216 | self.msgin(2, "load_module", fqname, fp and "fp", pathname) |
| 217 | if type == imp.PKG_DIRECTORY: |
| 218 | m = self.load_package(fqname, pathname) |
| 219 | self.msgout(2, "load_module ->", m) |
| 220 | return m |
| 221 | if type == imp.PY_SOURCE: |
| 222 | co = compile(fp.read(), pathname, 'exec') |
| 223 | elif type == imp.PY_COMPILED: |
| 224 | if fp.read(4) != imp.get_magic(): |
| 225 | self.msgout(2, "raise ImportError: Bad magic number", pathname) |
| 226 | raise ImportError, "Bad magic number in %s", pathname |
| 227 | fp.read(4) |
| 228 | co = marshal.load(fp) |
| 229 | else: |
| 230 | co = None |
| 231 | m = self.add_module(fqname) |
| 232 | if co: |
| 233 | m.__file__ = pathname |
| 234 | m.__code__ = co |
| 235 | code = co.co_code |
| 236 | n = len(code) |
| 237 | i = 0 |
| 238 | lastname = None |
| 239 | while i < n: |
| 240 | c = code[i] |
| 241 | i = i+1 |
| 242 | op = ord(c) |
| 243 | if op >= dis.HAVE_ARGUMENT: |
| 244 | oparg = ord(code[i]) + ord(code[i+1])*256 |
| 245 | i = i+2 |
| 246 | if op == IMPORT_NAME: |
| 247 | name = lastname = co.co_names[oparg] |
| 248 | if not self.badmodules.has_key(lastname): |
| 249 | try: |
| 250 | self.import_hook(name, m) |
| 251 | except ImportError, msg: |
| 252 | self.msg(2, "ImportError:", str(msg)) |
| 253 | self.badmodules[name] = None |
| 254 | elif op == IMPORT_FROM: |
| 255 | name = co.co_names[oparg] |
| 256 | assert lastname is not None |
| 257 | if not self.badmodules.has_key(lastname): |
| 258 | try: |
| 259 | self.import_hook(lastname, m, [name]) |
| 260 | except ImportError, msg: |
| 261 | self.msg(2, "ImportError:", str(msg)) |
| 262 | fullname = lastname + "." + name |
| 263 | self.badmodules[fullname] = None |
| 264 | else: |
| 265 | lastname = None |
| 266 | self.msgout(2, "load_module ->", m) |
| 267 | return m |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 268 | |
| 269 | def load_package(self, fqname, pathname): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 270 | self.msgin(2, "load_package", fqname, pathname) |
| 271 | m = self.add_module(fqname) |
| 272 | m.__file__ = pathname |
| 273 | m.__path__ = [pathname] |
| 274 | fp, buf, stuff = self.find_module("__init__", m.__path__) |
| 275 | self.load_module(fqname, fp, buf, stuff) |
| 276 | self.msgout(2, "load_package ->", m) |
| 277 | return m |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 278 | |
| 279 | def add_module(self, fqname): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 280 | if self.modules.has_key(fqname): |
| 281 | return self.modules[fqname] |
| 282 | self.modules[fqname] = m = Module(fqname) |
| 283 | return m |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 284 | |
| 285 | def find_module(self, name, path): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 286 | if path is None: |
| 287 | if name in sys.builtin_module_names: |
| 288 | return (None, None, ("", "", imp.C_BUILTIN)) |
| 289 | path = self.path |
| 290 | return imp.find_module(name, path) |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 291 | |
| 292 | def report(self): |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 293 | print |
| 294 | print " %-25s %s" % ("Name", "File") |
| 295 | print " %-25s %s" % ("----", "----") |
| 296 | # Print modules found |
| 297 | keys = self.modules.keys() |
| 298 | keys.sort() |
| 299 | for key in keys: |
| 300 | m = self.modules[key] |
| 301 | if m.__path__: |
| 302 | print "P", |
| 303 | else: |
| 304 | print "m", |
| 305 | print "%-25s" % key, m.__file__ or "" |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 306 | |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 307 | # Print missing modules |
| 308 | keys = self.badmodules.keys() |
| 309 | keys.sort() |
| 310 | for key in keys: |
| 311 | print "?", key |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 312 | |
| 313 | |
| 314 | def test(): |
| 315 | # Parse command line |
| 316 | import getopt |
| 317 | try: |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 318 | opts, args = getopt.getopt(sys.argv[1:], "dmp:q") |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 319 | except getopt.error, msg: |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 320 | print msg |
| 321 | return |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 322 | |
| 323 | # Process options |
| 324 | debug = 1 |
| 325 | domods = 0 |
| 326 | addpath = [] |
| 327 | for o, a in opts: |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 328 | if o == '-d': |
| 329 | debug = debug + 1 |
| 330 | if o == '-m': |
| 331 | domods = 1 |
| 332 | if o == '-p': |
| 333 | addpath = addpath + string.split(a, os.pathsep) |
| 334 | if o == '-q': |
| 335 | debug = 0 |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 336 | |
| 337 | # Provide default arguments |
| 338 | if not args: |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 339 | script = "hello.py" |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 340 | else: |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 341 | script = args[0] |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 342 | |
| 343 | # Set the path based on sys.path and the script directory |
| 344 | path = sys.path[:] |
| 345 | path[0] = os.path.dirname(script) |
| 346 | path = addpath + path |
| 347 | if debug > 1: |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 348 | print "path:" |
| 349 | for item in path: |
| 350 | print " ", `item` |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 351 | |
| 352 | # Create the module finder and turn its crank |
| 353 | mf = ModuleFinder(path, debug) |
| 354 | for arg in args[1:]: |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 355 | if arg == '-m': |
| 356 | domods = 1 |
| 357 | continue |
| 358 | if domods: |
| 359 | if arg[-2:] == '.*': |
| 360 | mf.import_hook(arg[:-2], None, ["*"]) |
| 361 | else: |
| 362 | mf.import_hook(arg) |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 363 | else: |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 364 | mf.load_file(arg) |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 365 | mf.run_script(script) |
| 366 | mf.report() |
| 367 | |
| 368 | |
| 369 | if __name__ == '__main__': |
| 370 | try: |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 371 | test() |
Guido van Rossum | 75dc496 | 1998-03-05 03:42:00 +0000 | [diff] [blame] | 372 | except KeyboardInterrupt: |
Guido van Rossum | 912a14c | 1998-03-05 04:56:37 +0000 | [diff] [blame^] | 373 | print "\n[interrupt]" |
| 374 | |
| 375 | # Local Variables: |
| 376 | # indent-tabs-mode: nil |
| 377 | # End: |