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         code_tag = next(tag
 154                         for tag_code, tag in _extract_tags(file_contents)
 155                         if tag_code == 82)
 156         p = code_tag.index(b'\0', 4) + 1
 157         code_reader = io.BytesIO(code_tag[p:])
 158
 159         # Parse ABC (AVM2 ByteCode)
 160
 161         # Define a couple convenience methods
 162         u30 = lambda *args: _u30(*args, reader=code_reader)
 163         s32 = lambda *args: _s32(*args, reader=code_reader)
 164         u32 = lambda *args: _u32(*args, reader=code_reader)
 165         read_bytes = lambda *args: _read_bytes(*args, reader=code_reader)
 166         read_byte = lambda *args: _read_byte(*args, reader=code_reader)
 167
 168         # minor_version + major_version
 169         read_bytes(2 + 2)
 170
 171         # Constant pool
 172         int_count = u30()
 173         for _c in range(1, int_count):
 174             s32()
 175         uint_count = u30()
 176         for _c in range(1, uint_count):
 177             u32()
 178         double_count = u30()
 179         read_bytes(max(0, (double_count - 1)) * 8)
 180         string_count = u30()
 181         self.constant_strings = ['']
 182         for _c in range(1, string_count):
 183             s = _read_string(code_reader)
 184             self.constant_strings.append(s)
 185         namespace_count = u30()
 186         for _c in range(1, namespace_count):
 187             read_bytes(1)  # kind
 188             u30()  # name
 189         ns_set_count = u30()
 190         for _c in range(1, ns_set_count):
 191             count = u30()
 192             for _c2 in range(count):
 193                 u30()
 194         multiname_count = u30()
 195         MULTINAME_SIZES = {
 196             0x07: 2,  # QName
 197             0x0d: 2,  # QNameA
 198             0x0f: 1,  # RTQName
 199             0x10: 1,  # RTQNameA
 200             0x11: 0,  # RTQNameL
 201             0x12: 0,  # RTQNameLA
 202             0x09: 2,  # Multiname
 203             0x0e: 2,  # MultinameA
 204             0x1b: 1,  # MultinameL
 205             0x1c: 1,  # MultinameLA
 206         }
 207         self.multinames = ['']
 208         for _c in range(1, multiname_count):
 209             kind = u30()
 210             assert kind in MULTINAME_SIZES, 'Invalid multiname kind %r' % kind
 211             if kind == 0x07:
 212                 u30()  # namespace_idx
 213                 name_idx = u30()
 214                 self.multinames.append(self.constant_strings[name_idx])
 215             else:
 216                 self.multinames.append(_Multiname(kind))
 217                 for _c2 in range(MULTINAME_SIZES[kind]):
 218                     u30()
 219
 220         # Methods
 221         method_count = u30()
 222         MethodInfo = collections.namedtuple(
 223             'MethodInfo',
 224             ['NEED_ARGUMENTS', 'NEED_REST'])
 225         method_infos = []
 226         for method_id in range(method_count):
 227             param_count = u30()
 228             u30()  # return type
 229             for _ in range(param_count):
 230                 u30()  # param type
 231             u30()  # name index (always 0 for youtube)
 232             flags = read_byte()
 233             if flags & 0x08 != 0:
 234                 # Options present
 235                 option_count = u30()
 236                 for c in range(option_count):
 237                     u30()  # val
 238                     read_bytes(1)  # kind
 239             if flags & 0x80 != 0:
 240                 # Param names present
 241                 for _ in range(param_count):
 242                     u30()  # param name
 243             mi = MethodInfo(flags & 0x01 != 0, flags & 0x04 != 0)
 244             method_infos.append(mi)
 245
 246         # Metadata
 247         metadata_count = u30()
 248         for _c in range(metadata_count):
 249             u30()  # name
 250             item_count = u30()
 251             for _c2 in range(item_count):
 252                 u30()  # key
 253                 u30()  # value
 254
 255         def parse_traits_info():
 256             trait_name_idx = u30()
 257             kind_full = read_byte()
 258             kind = kind_full & 0x0f
 259             attrs = kind_full >> 4
 260             methods = {}
 261             if kind in [0x00, 0x06]:  # Slot or Const
 262                 u30()  # Slot id
 263                 u30()  # type_name_idx
 264                 vindex = u30()
 265                 if vindex != 0:
 266                     read_byte()  # vkind
 267             elif kind in [0x01, 0x02, 0x03]:  # Method / Getter / Setter
 268                 u30()  # disp_id
 269                 method_idx = u30()
 270                 methods[self.multinames[trait_name_idx]] = method_idx
 271             elif kind == 0x04:  # Class
 272                 u30()  # slot_id
 273                 u30()  # classi
 274             elif kind == 0x05:  # Function
 275                 u30()  # slot_id
 276                 function_idx = u30()
 277                 methods[function_idx] = self.multinames[trait_name_idx]
 278             else:
 279                 raise ExtractorError('Unsupported trait kind %d' % kind)
 280
 281             if attrs & 0x4 != 0:  # Metadata present
 282                 metadata_count = u30()
 283                 for _c3 in range(metadata_count):
 284                     u30()  # metadata index
 285
 286             return methods
 287
 288         # Classes
 289         class_count = u30()
 290         classes = []
 291         for class_id in range(class_count):
 292             name_idx = u30()
 293
 294             cname = self.multinames[name_idx]
 295             avm_class = _AVMClass(name_idx, cname)
 296             classes.append(avm_class)
 297
 298             u30()  # super_name idx
 299             flags = read_byte()
 300             if flags & 0x08 != 0:  # Protected namespace is present
 301                 u30()  # protected_ns_idx
 302             intrf_count = u30()
 303             for _c2 in range(intrf_count):
 304                 u30()
 305             u30()  # iinit
 306             trait_count = u30()
 307             for _c2 in range(trait_count):
 308                 trait_methods = parse_traits_info()
 309                 avm_class.register_methods(trait_methods)
 310
 311         assert len(classes) == class_count
 312         self._classes_by_name = dict((c.name, c) for c in classes)
 313
 314         for avm_class in classes:
 315             u30()  # cinit
 316             trait_count = u30()
 317             for _c2 in range(trait_count):
 318                 trait_methods = parse_traits_info()
 319                 avm_class.register_methods(trait_methods)
 320
 321         # Scripts
 322         script_count = u30()
 323         for _c in range(script_count):
 324             u30()  # init
 325             trait_count = u30()
 326             for _c2 in range(trait_count):
 327                 parse_traits_info()
 328
 329         # Method bodies
 330         method_body_count = u30()
 331         Method = collections.namedtuple('Method', ['code', 'local_count'])
 332         for _c in range(method_body_count):
 333             method_idx = u30()
 334             u30()  # max_stack
 335             local_count = u30()
 336             u30()  # init_scope_depth
 337             u30()  # max_scope_depth
 338             code_length = u30()
 339             code = read_bytes(code_length)
 340             for avm_class in classes:
 341                 if method_idx in avm_class.method_idxs:
 342                     m = Method(code, local_count)
 343                     avm_class.methods[avm_class.method_idxs[method_idx]] = m
 344             exception_count = u30()
 345             for _c2 in range(exception_count):
 346                 u30()  # from
 347                 u30()  # to
 348                 u30()  # target
 349                 u30()  # exc_type
 350                 u30()  # var_name
 351             trait_count = u30()
 352             for _c2 in range(trait_count):
 353                 parse_traits_info()
 354
 355         assert p + code_reader.tell() == len(code_tag)
 356
 357     def extract_class(self, class_name):
 358         try:
 359             return self._classes_by_name[class_name]
 360         except KeyError:
 361             raise ExtractorError('Class %r not found' % class_name)
 362
 363     def extract_function(self, avm_class, func_name):
 364         if func_name in avm_class.method_pyfunctions:
 365             return avm_class.method_pyfunctions[func_name]
 366         if func_name in self._classes_by_name:
 367             return self._classes_by_name[func_name].make_object()
 368         if func_name not in avm_class.methods:
 369             raise ExtractorError('Cannot find function %s.%s' % (
 370                 avm_class.name, func_name))
 371         m = avm_class.methods[func_name]
 372
 373         def resfunc(args):
 374             # Helper functions
 375             coder = io.BytesIO(m.code)
 376             s24 = lambda: _s24(coder)
 377             u30 = lambda: _u30(coder)
 378
 379             registers = [avm_class.variables] + list(args) + [None] * m.local_count
 380             stack = []
 381             scopes = collections.deque([
 382                 self._classes_by_name, avm_class.variables])
 383             while True:
 384                 opcode = _read_byte(coder)
 385                 if opcode == 17:  # iftrue
 386                     offset = s24()
 387                     value = stack.pop()
 388                     if value:
 389                         coder.seek(coder.tell() + offset)
 390                 elif opcode == 18:  # iffalse
 391                     offset = s24()
 392                     value = stack.pop()
 393                     if not value:
 394                         coder.seek(coder.tell() + offset)
 395                 elif opcode == 36:  # pushbyte
 396                     v = _read_byte(coder)
 397                     stack.append(v)
 398                 elif opcode == 42:  # dup
 399                     value = stack[-1]
 400                     stack.append(value)
 401                 elif opcode == 44:  # pushstring
 402                     idx = u30()
 403                     stack.append(self.constant_strings[idx])
 404                 elif opcode == 48:  # pushscope
 405                     new_scope = stack.pop()
 406                     scopes.append(new_scope)
 407                 elif opcode == 66:  # construct
 408                     arg_count = u30()
 409                     args = list(reversed(
 410                         [stack.pop() for _ in range(arg_count)]))
 411                     obj = stack.pop()
 412                     res = obj.avm_class.make_object()
 413                     stack.append(res)
 414                 elif opcode == 70:  # callproperty
 415                     index = u30()
 416                     mname = self.multinames[index]
 417                     arg_count = u30()
 418                     args = list(reversed(
 419                         [stack.pop() for _ in range(arg_count)]))
 420                     obj = stack.pop()
 421
 422                     if isinstance(obj, _AVMClass_Object):
 423                         func = self.extract_function(obj.avm_class, mname)
 424                         res = func(args)
 425                         stack.append(res)
 426                         continue
 427                     elif isinstance(obj, _ScopeDict):
 428                         if mname in obj.avm_class.method_names:
 429                             func = self.extract_function(obj.avm_class, mname)
 430                             res = func(args)
 431                         else:
 432                             res = obj[mname]
 433                         stack.append(res)
 434                         continue
 435                     elif isinstance(obj, compat_str):
 436                         if mname == 'split':
 437                             assert len(args) == 1
 438                             assert isinstance(args[0], compat_str)
 439                             if args[0] == '':
 440                                 res = list(obj)
 441                             else:
 442                                 res = obj.split(args[0])
 443                             stack.append(res)
 444                             continue
 445                     elif isinstance(obj, list):
 446                         if mname == 'slice':
 447                             assert len(args) == 1
 448                             assert isinstance(args[0], int)
 449                             res = obj[args[0]:]
 450                             stack.append(res)
 451                             continue
 452                         elif mname == 'join':
 453                             assert len(args) == 1
 454                             assert isinstance(args[0], compat_str)
 455                             res = args[0].join(obj)
 456                             stack.append(res)
 457                             continue
 458                     raise NotImplementedError(
 459                         'Unsupported property %r on %r'
 460                         % (mname, obj))
 461                 elif opcode == 72:  # returnvalue
 462                     res = stack.pop()
 463                     return res
 464                 elif opcode == 74:  # constructproperty
 465                     index = u30()
 466                     arg_count = u30()
 467                     args = list(reversed(
 468                         [stack.pop() for _ in range(arg_count)]))
 469                     obj = stack.pop()
 470
 471                     mname = self.multinames[index]
 472                     assert isinstance(obj, _AVMClass)
 473
 474                     # We do not actually call the constructor for now;
 475                     # we just pretend it does nothing
 476                     stack.append(obj.make_object())
 477                 elif opcode == 79:  # callpropvoid
 478                     index = u30()
 479                     mname = self.multinames[index]
 480                     arg_count = u30()
 481                     args = list(reversed(
 482                         [stack.pop() for _ in range(arg_count)]))
 483                     obj = stack.pop()
 484                     if mname == 'reverse':
 485                         assert isinstance(obj, list)
 486                         obj.reverse()
 487                     else:
 488                         raise NotImplementedError(
 489                             'Unsupported (void) property %r on %r'
 490                             % (mname, obj))
 491                 elif opcode == 86:  # newarray
 492                     arg_count = u30()
 493                     arr = []
 494                     for i in range(arg_count):
 495                         arr.append(stack.pop())
 496                     arr = arr[::-1]
 497                     stack.append(arr)
 498                 elif opcode == 93:  # findpropstrict
 499                     index = u30()
 500                     mname = self.multinames[index]
 501                     for s in reversed(scopes):
 502                         if mname in s:
 503                             res = s
 504                             break
 505                     else:
 506                         res = scopes[0]
 507                     stack.append(res[mname])
 508                 elif opcode == 94:  # findproperty
 509                     index = u30()
 510                     mname = self.multinames[index]
 511                     for s in reversed(scopes):
 512                         if mname in s:
 513                             res = s
 514                             break
 515                     else:
 516                         res = avm_class.variables
 517                     stack.append(res)
 518                 elif opcode == 96:  # getlex
 519                     index = u30()
 520                     mname = self.multinames[index]
 521                     for s in reversed(scopes):
 522                         if mname in s:
 523                             scope = s
 524                             break
 525                     else:
 526                         scope = avm_class.variables
 527                     # I cannot find where static variables are initialized
 528                     # so let's just return None
 529                     res = scope.get(mname)
 530                     stack.append(res)
 531                 elif opcode == 97:  # setproperty
 532                     index = u30()
 533                     value = stack.pop()
 534                     idx = self.multinames[index]
 535                     if isinstance(idx, _Multiname):
 536                         idx = stack.pop()
 537                     obj = stack.pop()
 538                     obj[idx] = value
 539                 elif opcode == 98:  # getlocal
 540                     index = u30()
 541                     stack.append(registers[index])
 542                 elif opcode == 99:  # setlocal
 543                     index = u30()
 544                     value = stack.pop()
 545                     registers[index] = value
 546                 elif opcode == 102:  # getproperty
 547                     index = u30()
 548                     pname = self.multinames[index]
 549                     if pname == 'length':
 550                         obj = stack.pop()
 551                         assert isinstance(obj, list)
 552                         stack.append(len(obj))
 553                     else:  # Assume attribute access
 554                         idx = stack.pop()
 555                         assert isinstance(idx, int)
 556                         obj = stack.pop()
 557                         assert isinstance(obj, list)
 558                         stack.append(obj[idx])
 559                 elif opcode == 115:  # convert_
 560                     value = stack.pop()
 561                     intvalue = int(value)
 562                     stack.append(intvalue)
 563                 elif opcode == 128:  # coerce
 564                     u30()
 565                 elif opcode == 133:  # coerce_s
 566                     assert isinstance(stack[-1], (type(None), compat_str))
 567                 elif opcode == 160:  # add
 568                     value2 = stack.pop()
 569                     value1 = stack.pop()
 570                     res = value1 + value2
 571                     stack.append(res)
 572                 elif opcode == 161:  # subtract
 573                     value2 = stack.pop()
 574                     value1 = stack.pop()
 575                     res = value1 - value2
 576                     stack.append(res)
 577                 elif opcode == 164:  # modulo
 578                     value2 = stack.pop()
 579                     value1 = stack.pop()
 580                     res = value1 % value2
 581                     stack.append(res)
 582                 elif opcode == 175:  # greaterequals
 583                     value2 = stack.pop()
 584                     value1 = stack.pop()
 585                     result = value1 >= value2
 586                     stack.append(result)
 587                 elif opcode == 208:  # getlocal_0
 588                     stack.append(registers[0])
 589                 elif opcode == 209:  # getlocal_1
 590                     stack.append(registers[1])
 591                 elif opcode == 210:  # getlocal_2
 592                     stack.append(registers[2])
 593                 elif opcode == 211:  # getlocal_3
 594                     stack.append(registers[3])
 595                 elif opcode == 212:  # setlocal_0
 596                     registers[0] = stack.pop()
 597                 elif opcode == 213:  # setlocal_1
 598                     registers[1] = stack.pop()
 599                 elif opcode == 214:  # setlocal_2
 600                     registers[2] = stack.pop()
 601                 elif opcode == 215:  # setlocal_3
 602                     registers[3] = stack.pop()
 603                 else:
 604                     raise NotImplementedError(
 605                         'Unsupported opcode %d' % opcode)
 606
 607         avm_class.method_pyfunctions[func_name] = resfunc
 608         return resfunc
 609