_ Git - youtube-dl/blob - youtube_dl/swfinterp.py

   1 from __future__ import unicode_literals
   2
   3 import collections
   4 import io
   5 import zlib
   6
   7 from .utils import (
   8     compat_str,
   9     ExtractorError,
  10     struct_unpack,
  11 )
  12
  13
  14 def _extract_tags(file_contents):
  15     if file_contents[1:3] != b'WS':
  16         raise ExtractorError(
  17             'Not an SWF file; header is %r' % file_contents[:3])
  18     if file_contents[:1] == b'C':
  19         content = zlib.decompress(file_contents[8:])
  20     else:
  21         raise NotImplementedError(
  22             'Unsupported compression format %r' %
  23             file_contents[:1])
  24
  25     # Determine number of bits in framesize rectangle
  26     framesize_nbits = struct_unpack('!B', content[:1])[0] >> 3
  27     framesize_len = (5 + 4 * framesize_nbits + 7) // 8
  28
  29     pos = framesize_len + 2 + 2
  30     while pos < len(content):
  31         header16 = struct_unpack('<H', content[pos:pos + 2])[0]
  32         pos += 2
  33         tag_code = header16 >> 6
  34         tag_len = header16 & 0x3f
  35         if tag_len == 0x3f:
  36             tag_len = struct_unpack('<I', content[pos:pos + 4])[0]
  37             pos += 4
  38         assert pos + tag_len <= len(content), \
  39             ('Tag %d ends at %d+%d - that\'s longer than the file (%d)'
  40                 % (tag_code, pos, tag_len, len(content)))
  41         yield (tag_code, content[pos:pos + tag_len])
  42         pos += tag_len
  43
  44
  45 class _AVMClass_Object(object):
  46     def __init__(self, avm_class):
  47         self.avm_class = avm_class
  48
  49     def __repr__(self):
  50         return '%s#%x' % (self.avm_class.name, id(self))
  51
  52
  53 class _ScopeDict(dict):
  54     def __init__(self, avm_class):
  55         super(_ScopeDict, self).__init__()
  56         self.avm_class = avm_class
  57
  58     def __repr__(self):
  59         return '%s__Scope(%s)' % (
  60             self.avm_class.name,
  61             super(_ScopeDict, self).__repr__())
  62
  63
  64 class _AVMClass(object):
  65     def __init__(self, name_idx, name):
  66         self.name_idx = name_idx
  67         self.name = name
  68         self.method_names = {}
  69         self.method_idxs = {}
  70         self.methods = {}
  71         self.method_pyfunctions = {}
  72
  73         self.variables = _ScopeDict(self)
  74
  75     def make_object(self):
  76         return _AVMClass_Object(self)
  77
  78     def __repr__(self):
  79         return '_AVMClass(%s)' % (self.name)
  80
  81     def register_methods(self, methods):
  82         self.method_names.update(methods.items())
  83         self.method_idxs.update(dict(
  84             (idx, name)
  85             for name, idx in methods.items()))
  86
  87
  88 class _Multiname(object):
  89     def __init__(self, kind):
  90         self.kind = kind
  91
  92     def __repr__(self):
  93         return '[MULTINAME kind: 0x%x]' % self.kind
  94
  95
  96 def _read_int(reader):
  97     res = 0
  98     shift = 0
  99     for _ in range(5):
 100         buf = reader.read(1)
 101         assert len(buf) == 1
 102         b = struct_unpack('<B', buf)[0]
 103         res = res | ((b & 0x7f) << shift)
 104         if b & 0x80 == 0:
 105             break
 106         shift += 7
 107     return res
 108
 109
 110 def _u30(reader):
 111     res = _read_int(reader)
 112     assert res & 0xf0000000 == 0
 113     return res
 114 _u32 = _read_int
 115
 116
 117 def _s32(reader):
 118     v = _read_int(reader)
 119     if v & 0x80000000 != 0:
 120         v = - ((v ^ 0xffffffff) + 1)
 121     return v
 122
 123
 124 def _s24(reader):
 125     bs = reader.read(3)
 126     assert len(bs) == 3
 127     last_byte = b'\xff' if (ord(bs[2:3]) >= 0x80) else b'\x00'
 128     return struct_unpack('<i', bs + last_byte)[0]
 129
 130
 131 def _read_string(reader):
 132     slen = _u30(reader)
 133     resb = reader.read(slen)
 134     assert len(resb) == slen
 135     return resb.decode('utf-8')
 136
 137
 138 def _read_bytes(count, reader):
 139     assert count >= 0
 140     resb = reader.read(count)
 141     assert len(resb) == count
 142     return resb
 143
 144
 145 def _read_byte(reader):
 146     resb = _read_bytes(1, reader=reader)
 147     res = struct_unpack('<B', resb)[0]
 148     return res
 149
 150
 151 class SWFInterpreter(object):
 152     def __init__(self, file_contents):
 153         self._patched_functions = {}
 154         code_tag = next(tag
 155                         for tag_code, tag in _extract_tags(file_contents)
 156                         if tag_code == 82)
 157         p = code_tag.index(b'\0', 4) + 1
 158         code_reader = io.BytesIO(code_tag[p:])
 159
 160         # Parse ABC (AVM2 ByteCode)
 161
 162         # Define a couple convenience methods
 163         u30 = lambda *args: _u30(*args, reader=code_reader)
 164         s32 = lambda *args: _s32(*args, reader=code_reader)
 165         u32 = lambda *args: _u32(*args, reader=code_reader)
 166         read_bytes = lambda *args: _read_bytes(*args, reader=code_reader)
 167         read_byte = lambda *args: _read_byte(*args, reader=code_reader)
 168
 169         # minor_version + major_version
 170         read_bytes(2 + 2)
 171
 172         # Constant pool
 173         int_count = u30()
 174         for _c in range(1, int_count):
 175             s32()
 176         uint_count = u30()
 177         for _c in range(1, uint_count):
 178             u32()
 179         double_count = u30()
 180         read_bytes(max(0, (double_count - 1)) * 8)
 181         string_count = u30()
 182         self.constant_strings = ['']
 183         for _c in range(1, string_count):
 184             s = _read_string(code_reader)
 185             self.constant_strings.append(s)
 186         namespace_count = u30()
 187         for _c in range(1, namespace_count):
 188             read_bytes(1)  # kind
 189             u30()  # name
 190         ns_set_count = u30()
 191         for _c in range(1, ns_set_count):
 192             count = u30()
 193             for _c2 in range(count):
 194                 u30()
 195         multiname_count = u30()
 196         MULTINAME_SIZES = {
 197             0x07: 2,  # QName
 198             0x0d: 2,  # QNameA
 199             0x0f: 1,  # RTQName
 200             0x10: 1,  # RTQNameA
 201             0x11: 0,  # RTQNameL
 202             0x12: 0,  # RTQNameLA
 203             0x09: 2,  # Multiname
 204             0x0e: 2,  # MultinameA
 205             0x1b: 1,  # MultinameL
 206             0x1c: 1,  # MultinameLA
 207         }
 208         self.multinames = ['']
 209         for _c in range(1, multiname_count):
 210             kind = u30()
 211             assert kind in MULTINAME_SIZES, 'Invalid multiname kind %r' % kind
 212             if kind == 0x07:
 213                 u30()  # namespace_idx
 214                 name_idx = u30()
 215                 self.multinames.append(self.constant_strings[name_idx])
 216             elif kind == 0x09:
 217                 name_idx = u30()
 218                 u30()
 219                 self.multinames.append(self.constant_strings[name_idx])
 220             else:
 221                 self.multinames.append(_Multiname(kind))
 222                 for _c2 in range(MULTINAME_SIZES[kind]):
 223                     u30()
 224
 225         # Methods
 226         method_count = u30()
 227         MethodInfo = collections.namedtuple(
 228             'MethodInfo',
 229             ['NEED_ARGUMENTS', 'NEED_REST'])
 230         method_infos = []
 231         for method_id in range(method_count):
 232             param_count = u30()
 233             u30()  # return type
 234             for _ in range(param_count):
 235                 u30()  # param type
 236             u30()  # name index (always 0 for youtube)
 237             flags = read_byte()
 238             if flags & 0x08 != 0:
 239                 # Options present
 240                 option_count = u30()
 241                 for c in range(option_count):
 242                     u30()  # val
 243                     read_bytes(1)  # kind
 244             if flags & 0x80 != 0:
 245                 # Param names present
 246                 for _ in range(param_count):
 247                     u30()  # param name
 248             mi = MethodInfo(flags & 0x01 != 0, flags & 0x04 != 0)
 249             method_infos.append(mi)
 250
 251         # Metadata
 252         metadata_count = u30()
 253         for _c in range(metadata_count):
 254             u30()  # name
 255             item_count = u30()
 256             for _c2 in range(item_count):
 257                 u30()  # key
 258                 u30()  # value
 259
 260         def parse_traits_info():
 261             trait_name_idx = u30()
 262             kind_full = read_byte()
 263             kind = kind_full & 0x0f
 264             attrs = kind_full >> 4
 265             methods = {}
 266             if kind in [0x00, 0x06]:  # Slot or Const
 267                 u30()  # Slot id
 268                 u30()  # type_name_idx
 269                 vindex = u30()
 270                 if vindex != 0:
 271                     read_byte()  # vkind
 272             elif kind in [0x01, 0x02, 0x03]:  # Method / Getter / Setter
 273                 u30()  # disp_id
 274                 method_idx = u30()
 275                 methods[self.multinames[trait_name_idx]] = method_idx
 276             elif kind == 0x04:  # Class
 277                 u30()  # slot_id
 278                 u30()  # classi
 279             elif kind == 0x05:  # Function
 280                 u30()  # slot_id
 281                 function_idx = u30()
 282                 methods[function_idx] = self.multinames[trait_name_idx]
 283             else:
 284                 raise ExtractorError('Unsupported trait kind %d' % kind)
 285
 286             if attrs & 0x4 != 0:  # Metadata present
 287                 metadata_count = u30()
 288                 for _c3 in range(metadata_count):
 289                     u30()  # metadata index
 290
 291             return methods
 292
 293         # Classes
 294         class_count = u30()
 295         classes = []
 296         for class_id in range(class_count):
 297             name_idx = u30()
 298
 299             cname = self.multinames[name_idx]
 300             avm_class = _AVMClass(name_idx, cname)
 301             classes.append(avm_class)
 302
 303             u30()  # super_name idx
 304             flags = read_byte()
 305             if flags & 0x08 != 0:  # Protected namespace is present
 306                 u30()  # protected_ns_idx
 307             intrf_count = u30()
 308             for _c2 in range(intrf_count):
 309                 u30()
 310             u30()  # iinit
 311             trait_count = u30()
 312             for _c2 in range(trait_count):
 313                 trait_methods = parse_traits_info()
 314                 avm_class.register_methods(trait_methods)
 315
 316         assert len(classes) == class_count
 317         self._classes_by_name = dict((c.name, c) for c in classes)
 318
 319         for avm_class in classes:
 320             u30()  # cinit
 321             trait_count = u30()
 322             for _c2 in range(trait_count):
 323                 trait_methods = parse_traits_info()
 324                 avm_class.register_methods(trait_methods)
 325
 326         # Scripts
 327         script_count = u30()
 328         for _c in range(script_count):
 329             u30()  # init
 330             trait_count = u30()
 331             for _c2 in range(trait_count):
 332                 parse_traits_info()
 333
 334         # Method bodies
 335         method_body_count = u30()
 336         Method = collections.namedtuple('Method', ['code', 'local_count'])
 337         for _c in range(method_body_count):
 338             method_idx = u30()
 339             u30()  # max_stack
 340             local_count = u30()
 341             u30()  # init_scope_depth
 342             u30()  # max_scope_depth
 343             code_length = u30()
 344             code = read_bytes(code_length)
 345             for avm_class in classes:
 346                 if method_idx in avm_class.method_idxs:
 347                     m = Method(code, local_count)
 348                     avm_class.methods[avm_class.method_idxs[method_idx]] = m
 349             exception_count = u30()
 350             for _c2 in range(exception_count):
 351                 u30()  # from
 352                 u30()  # to
 353                 u30()  # target
 354                 u30()  # exc_type
 355                 u30()  # var_name
 356             trait_count = u30()
 357             for _c2 in range(trait_count):
 358                 parse_traits_info()
 359
 360         assert p + code_reader.tell() == len(code_tag)
 361
 362     def patch_function(self, avm_class, func_name, f):
 363         self._patched_functions[(avm_class, func_name)] = f
 364
 365     def extract_class(self, class_name):
 366         try:
 367             return self._classes_by_name[class_name]
 368         except KeyError:
 369             raise ExtractorError('Class %r not found' % class_name)
 370
 371     def extract_function(self, avm_class, func_name):
 372         p = self._patched_functions.get((avm_class, func_name))
 373         if p:
 374             return p
 375         if func_name in avm_class.method_pyfunctions:
 376             return avm_class.method_pyfunctions[func_name]
 377         if func_name in self._classes_by_name:
 378             return self._classes_by_name[func_name].make_object()
 379         if func_name not in avm_class.methods:
 380             raise ExtractorError('Cannot find function %s.%s' % (
 381                 avm_class.name, func_name))
 382         m = avm_class.methods[func_name]
 383
 384         def resfunc(args):
 385             # Helper functions
 386             coder = io.BytesIO(m.code)
 387             s24 = lambda: _s24(coder)
 388             u30 = lambda: _u30(coder)
 389
 390             registers = [avm_class.variables] + list(args) + [None] * m.local_count
 391             stack = []
 392             scopes = collections.deque([
 393                 self._classes_by_name, avm_class.variables])
 394             while True:
 395                 opcode = _read_byte(coder)
 396                 if opcode == 17:  # iftrue
 397                     offset = s24()
 398                     value = stack.pop()
 399                     if value:
 400                         coder.seek(coder.tell() + offset)
 401                 elif opcode == 18:  # iffalse
 402                     offset = s24()
 403                     value = stack.pop()
 404                     if not value:
 405                         coder.seek(coder.tell() + offset)
 406                 elif opcode == 36:  # pushbyte
 407                     v = _read_byte(coder)
 408                     stack.append(v)
 409                 elif opcode == 42:  # dup
 410                     value = stack[-1]
 411                     stack.append(value)
 412                 elif opcode == 44:  # pushstring
 413                     idx = u30()
 414                     stack.append(self.constant_strings[idx])
 415                 elif opcode == 48:  # pushscope
 416                     new_scope = stack.pop()
 417                     scopes.append(new_scope)
 418                 elif opcode == 66:  # construct
 419                     arg_count = u30()
 420                     args = list(reversed(
 421                         [stack.pop() for _ in range(arg_count)]))
 422                     obj = stack.pop()
 423                     res = obj.avm_class.make_object()
 424                     stack.append(res)
 425                 elif opcode == 70:  # callproperty
 426                     index = u30()
 427                     mname = self.multinames[index]
 428                     arg_count = u30()
 429                     args = list(reversed(
 430                         [stack.pop() for _ in range(arg_count)]))
 431                     obj = stack.pop()
 432
 433                     if isinstance(obj, _AVMClass_Object):
 434                         func = self.extract_function(obj.avm_class, mname)
 435                         res = func(args)
 436                         stack.append(res)
 437                         continue
 438                     elif isinstance(obj, _ScopeDict):
 439                         if mname in obj.avm_class.method_names:
 440                             func = self.extract_function(obj.avm_class, mname)
 441                             res = func(args)
 442                         else:
 443                             res = obj[mname]
 444                         stack.append(res)
 445                         continue
 446                     elif isinstance(obj, compat_str):
 447                         if mname == 'split':
 448                             assert len(args) == 1
 449                             assert isinstance(args[0], compat_str)
 450                             if args[0] == '':
 451                                 res = list(obj)
 452                             else:
 453                                 res = obj.split(args[0])
 454                             stack.append(res)
 455                             continue
 456                     elif isinstance(obj, list):
 457                         if mname == 'slice':
 458                             assert len(args) == 1
 459                             assert isinstance(args[0], int)
 460                             res = obj[args[0]:]
 461                             stack.append(res)
 462                             continue
 463                         elif mname == 'join':
 464                             assert len(args) == 1
 465                             assert isinstance(args[0], compat_str)
 466                             res = args[0].join(obj)
 467                             stack.append(res)
 468                             continue
 469                     raise NotImplementedError(
 470                         'Unsupported property %r on %r'
 471                         % (mname, obj))
 472                 elif opcode == 72:  # returnvalue
 473                     res = stack.pop()
 474                     return res
 475                 elif opcode == 74:  # constructproperty
 476                     index = u30()
 477                     arg_count = u30()
 478                     args = list(reversed(
 479                         [stack.pop() for _ in range(arg_count)]))
 480                     obj = stack.pop()
 481
 482                     mname = self.multinames[index]
 483                     assert isinstance(obj, _AVMClass)
 484
 485                     # We do not actually call the constructor for now;
 486                     # we just pretend it does nothing
 487                     stack.append(obj.make_object())
 488                 elif opcode == 79:  # callpropvoid
 489                     index = u30()
 490                     mname = self.multinames[index]
 491                     arg_count = u30()
 492                     args = list(reversed(
 493                         [stack.pop() for _ in range(arg_count)]))
 494                     obj = stack.pop()
 495                     if mname == 'reverse':
 496                         assert isinstance(obj, list)
 497                         obj.reverse()
 498                     else:
 499                         raise NotImplementedError(
 500                             'Unsupported (void) property %r on %r'
 501                             % (mname, obj))
 502                 elif opcode == 86:  # newarray
 503                     arg_count = u30()
 504                     arr = []
 505                     for i in range(arg_count):
 506                         arr.append(stack.pop())
 507                     arr = arr[::-1]
 508                     stack.append(arr)
 509                 elif opcode == 93:  # findpropstrict
 510                     index = u30()
 511                     mname = self.multinames[index]
 512                     for s in reversed(scopes):
 513                         if mname in s:
 514                             res = s
 515                             break
 516                     else:
 517                         res = scopes[0]
 518                     stack.append(res[mname])
 519                 elif opcode == 94:  # findproperty
 520                     index = u30()
 521                     mname = self.multinames[index]
 522                     for s in reversed(scopes):
 523                         if mname in s:
 524                             res = s
 525                             break
 526                     else:
 527                         res = avm_class.variables
 528                     stack.append(res)
 529                 elif opcode == 96:  # getlex
 530                     index = u30()
 531                     mname = self.multinames[index]
 532                     for s in reversed(scopes):
 533                         if mname in s:
 534                             scope = s
 535                             break
 536                     else:
 537                         scope = avm_class.variables
 538                     # I cannot find where static variables are initialized
 539                     # so let's just return None
 540                     res = scope.get(mname)
 541                     stack.append(res)
 542                 elif opcode == 97:  # setproperty
 543                     index = u30()
 544                     value = stack.pop()
 545                     idx = self.multinames[index]
 546                     if isinstance(idx, _Multiname):
 547                         idx = stack.pop()
 548                     obj = stack.pop()
 549                     obj[idx] = value
 550                 elif opcode == 98:  # getlocal
 551                     index = u30()
 552                     stack.append(registers[index])
 553                 elif opcode == 99:  # setlocal
 554                     index = u30()
 555                     value = stack.pop()
 556                     registers[index] = value
 557                 elif opcode == 102:  # getproperty
 558                     index = u30()
 559                     pname = self.multinames[index]
 560                     if pname == 'length':
 561                         obj = stack.pop()
 562                         assert isinstance(obj, list)
 563                         stack.append(len(obj))
 564                     elif isinstance(pname, compat_str):  # Member access
 565                         obj = stack.pop()
 566                         assert isinstance(obj, (dict, _ScopeDict)), \
 567                             'Accessing member on %r' % obj
 568                         stack.append(obj[pname])
 569                     else:  # Assume attribute access
 570                         idx = stack.pop()
 571                         assert isinstance(idx, int)
 572                         obj = stack.pop()
 573                         assert isinstance(obj, list)
 574                         stack.append(obj[idx])
 575                 elif opcode == 115:  # convert_
 576                     value = stack.pop()
 577                     intvalue = int(value)
 578                     stack.append(intvalue)
 579                 elif opcode == 128:  # coerce
 580                     u30()
 581                 elif opcode == 133:  # coerce_s
 582                     assert isinstance(stack[-1], (type(None), compat_str))
 583                 elif opcode == 160:  # add
 584                     value2 = stack.pop()
 585                     value1 = stack.pop()
 586                     res = value1 + value2
 587                     stack.append(res)
 588                 elif opcode == 161:  # subtract
 589                     value2 = stack.pop()
 590                     value1 = stack.pop()
 591                     res = value1 - value2
 592                     stack.append(res)
 593                 elif opcode == 164:  # modulo
 594                     value2 = stack.pop()
 595                     value1 = stack.pop()
 596                     res = value1 % value2
 597                     stack.append(res)
 598                 elif opcode == 175:  # greaterequals
 599                     value2 = stack.pop()
 600                     value1 = stack.pop()
 601                     result = value1 >= value2
 602                     stack.append(result)
 603                 elif opcode == 208:  # getlocal_0
 604                     stack.append(registers[0])
 605                 elif opcode == 209:  # getlocal_1
 606                     stack.append(registers[1])
 607                 elif opcode == 210:  # getlocal_2
 608                     stack.append(registers[2])
 609                 elif opcode == 211:  # getlocal_3
 610                     stack.append(registers[3])
 611                 elif opcode == 212:  # setlocal_0
 612                     registers[0] = stack.pop()
 613                 elif opcode == 213:  # setlocal_1
 614                     registers[1] = stack.pop()
 615                 elif opcode == 214:  # setlocal_2
 616                     registers[2] = stack.pop()
 617                 elif opcode == 215:  # setlocal_3
 618                     registers[3] = stack.pop()
 619                 else:
 620                     raise NotImplementedError(
 621                         'Unsupported opcode %d' % opcode)
 622
 623         avm_class.method_pyfunctions[func_name] = resfunc
 624         return resfunc
 625