_ 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             else:
 217                 self.multinames.append(_Multiname(kind))
 218                 for _c2 in range(MULTINAME_SIZES[kind]):
 219                     u30()
 220
 221         # Methods
 222         method_count = u30()
 223         MethodInfo = collections.namedtuple(
 224             'MethodInfo',
 225             ['NEED_ARGUMENTS', 'NEED_REST'])
 226         method_infos = []
 227         for method_id in range(method_count):
 228             param_count = u30()
 229             u30()  # return type
 230             for _ in range(param_count):
 231                 u30()  # param type
 232             u30()  # name index (always 0 for youtube)
 233             flags = read_byte()
 234             if flags & 0x08 != 0:
 235                 # Options present
 236                 option_count = u30()
 237                 for c in range(option_count):
 238                     u30()  # val
 239                     read_bytes(1)  # kind
 240             if flags & 0x80 != 0:
 241                 # Param names present
 242                 for _ in range(param_count):
 243                     u30()  # param name
 244             mi = MethodInfo(flags & 0x01 != 0, flags & 0x04 != 0)
 245             method_infos.append(mi)
 246
 247         # Metadata
 248         metadata_count = u30()
 249         for _c in range(metadata_count):
 250             u30()  # name
 251             item_count = u30()
 252             for _c2 in range(item_count):
 253                 u30()  # key
 254                 u30()  # value
 255
 256         def parse_traits_info():
 257             trait_name_idx = u30()
 258             kind_full = read_byte()
 259             kind = kind_full & 0x0f
 260             attrs = kind_full >> 4
 261             methods = {}
 262             if kind in [0x00, 0x06]:  # Slot or Const
 263                 u30()  # Slot id
 264                 u30()  # type_name_idx
 265                 vindex = u30()
 266                 if vindex != 0:
 267                     read_byte()  # vkind
 268             elif kind in [0x01, 0x02, 0x03]:  # Method / Getter / Setter
 269                 u30()  # disp_id
 270                 method_idx = u30()
 271                 methods[self.multinames[trait_name_idx]] = method_idx
 272             elif kind == 0x04:  # Class
 273                 u30()  # slot_id
 274                 u30()  # classi
 275             elif kind == 0x05:  # Function
 276                 u30()  # slot_id
 277                 function_idx = u30()
 278                 methods[function_idx] = self.multinames[trait_name_idx]
 279             else:
 280                 raise ExtractorError('Unsupported trait kind %d' % kind)
 281
 282             if attrs & 0x4 != 0:  # Metadata present
 283                 metadata_count = u30()
 284                 for _c3 in range(metadata_count):
 285                     u30()  # metadata index
 286
 287             return methods
 288
 289         # Classes
 290         class_count = u30()
 291         classes = []
 292         for class_id in range(class_count):
 293             name_idx = u30()
 294
 295             cname = self.multinames[name_idx]
 296             avm_class = _AVMClass(name_idx, cname)
 297             classes.append(avm_class)
 298
 299             u30()  # super_name idx
 300             flags = read_byte()
 301             if flags & 0x08 != 0:  # Protected namespace is present
 302                 u30()  # protected_ns_idx
 303             intrf_count = u30()
 304             for _c2 in range(intrf_count):
 305                 u30()
 306             u30()  # iinit
 307             trait_count = u30()
 308             for _c2 in range(trait_count):
 309                 trait_methods = parse_traits_info()
 310                 avm_class.register_methods(trait_methods)
 311
 312         assert len(classes) == class_count
 313         self._classes_by_name = dict((c.name, c) for c in classes)
 314
 315         for avm_class in classes:
 316             u30()  # cinit
 317             trait_count = u30()
 318             for _c2 in range(trait_count):
 319                 trait_methods = parse_traits_info()
 320                 avm_class.register_methods(trait_methods)
 321
 322         # Scripts
 323         script_count = u30()
 324         for _c in range(script_count):
 325             u30()  # init
 326             trait_count = u30()
 327             for _c2 in range(trait_count):
 328                 parse_traits_info()
 329
 330         # Method bodies
 331         method_body_count = u30()
 332         Method = collections.namedtuple('Method', ['code', 'local_count'])
 333         for _c in range(method_body_count):
 334             method_idx = u30()
 335             u30()  # max_stack
 336             local_count = u30()
 337             u30()  # init_scope_depth
 338             u30()  # max_scope_depth
 339             code_length = u30()
 340             code = read_bytes(code_length)
 341             for avm_class in classes:
 342                 if method_idx in avm_class.method_idxs:
 343                     m = Method(code, local_count)
 344                     avm_class.methods[avm_class.method_idxs[method_idx]] = m
 345             exception_count = u30()
 346             for _c2 in range(exception_count):
 347                 u30()  # from
 348                 u30()  # to
 349                 u30()  # target
 350                 u30()  # exc_type
 351                 u30()  # var_name
 352             trait_count = u30()
 353             for _c2 in range(trait_count):
 354                 parse_traits_info()
 355
 356         assert p + code_reader.tell() == len(code_tag)
 357
 358     def patch_function(self, avm_class, func_name, f):
 359         self._patched_functions[(avm_class, func_name)] = f
 360
 361     def extract_class(self, class_name):
 362         try:
 363             return self._classes_by_name[class_name]
 364         except KeyError:
 365             raise ExtractorError('Class %r not found' % class_name)
 366
 367     def extract_function(self, avm_class, func_name):
 368         p = self._patched_functions.get((avm_class, func_name))
 369         if p:
 370             return p
 371         if func_name in avm_class.method_pyfunctions:
 372             return avm_class.method_pyfunctions[func_name]
 373         if func_name in self._classes_by_name:
 374             return self._classes_by_name[func_name].make_object()
 375         if func_name not in avm_class.methods:
 376             raise ExtractorError('Cannot find function %s.%s' % (
 377                 avm_class.name, func_name))
 378         m = avm_class.methods[func_name]
 379
 380         def resfunc(args):
 381             # Helper functions
 382             coder = io.BytesIO(m.code)
 383             s24 = lambda: _s24(coder)
 384             u30 = lambda: _u30(coder)
 385
 386             registers = [avm_class.variables] + list(args) + [None] * m.local_count
 387             stack = []
 388             scopes = collections.deque([
 389                 self._classes_by_name, avm_class.variables])
 390             while True:
 391                 opcode = _read_byte(coder)
 392                 if opcode == 17:  # iftrue
 393                     offset = s24()
 394                     value = stack.pop()
 395                     if value:
 396                         coder.seek(coder.tell() + offset)
 397                 elif opcode == 18:  # iffalse
 398                     offset = s24()
 399                     value = stack.pop()
 400                     if not value:
 401                         coder.seek(coder.tell() + offset)
 402                 elif opcode == 36:  # pushbyte
 403                     v = _read_byte(coder)
 404                     stack.append(v)
 405                 elif opcode == 42:  # dup
 406                     value = stack[-1]
 407                     stack.append(value)
 408                 elif opcode == 44:  # pushstring
 409                     idx = u30()
 410                     stack.append(self.constant_strings[idx])
 411                 elif opcode == 48:  # pushscope
 412                     new_scope = stack.pop()
 413                     scopes.append(new_scope)
 414                 elif opcode == 66:  # construct
 415                     arg_count = u30()
 416                     args = list(reversed(
 417                         [stack.pop() for _ in range(arg_count)]))
 418                     obj = stack.pop()
 419                     res = obj.avm_class.make_object()
 420                     stack.append(res)
 421                 elif opcode == 70:  # callproperty
 422                     index = u30()
 423                     mname = self.multinames[index]
 424                     arg_count = u30()
 425                     args = list(reversed(
 426                         [stack.pop() for _ in range(arg_count)]))
 427                     obj = stack.pop()
 428
 429                     if isinstance(obj, _AVMClass_Object):
 430                         func = self.extract_function(obj.avm_class, mname)
 431                         res = func(args)
 432                         stack.append(res)
 433                         continue
 434                     elif isinstance(obj, _ScopeDict):
 435                         if mname in obj.avm_class.method_names:
 436                             func = self.extract_function(obj.avm_class, mname)
 437                             res = func(args)
 438                         else:
 439                             res = obj[mname]
 440                         stack.append(res)
 441                         continue
 442                     elif isinstance(obj, compat_str):
 443                         if mname == 'split':
 444                             assert len(args) == 1
 445                             assert isinstance(args[0], compat_str)
 446                             if args[0] == '':
 447                                 res = list(obj)
 448                             else:
 449                                 res = obj.split(args[0])
 450                             stack.append(res)
 451                             continue
 452                     elif isinstance(obj, list):
 453                         if mname == 'slice':
 454                             assert len(args) == 1
 455                             assert isinstance(args[0], int)
 456                             res = obj[args[0]:]
 457                             stack.append(res)
 458                             continue
 459                         elif mname == 'join':
 460                             assert len(args) == 1
 461                             assert isinstance(args[0], compat_str)
 462                             res = args[0].join(obj)
 463                             stack.append(res)
 464                             continue
 465                     raise NotImplementedError(
 466                         'Unsupported property %r on %r'
 467                         % (mname, obj))
 468                 elif opcode == 72:  # returnvalue
 469                     res = stack.pop()
 470                     return res
 471                 elif opcode == 74:  # constructproperty
 472                     index = u30()
 473                     arg_count = u30()
 474                     args = list(reversed(
 475                         [stack.pop() for _ in range(arg_count)]))
 476                     obj = stack.pop()
 477
 478                     mname = self.multinames[index]
 479                     assert isinstance(obj, _AVMClass)
 480
 481                     # We do not actually call the constructor for now;
 482                     # we just pretend it does nothing
 483                     stack.append(obj.make_object())
 484                 elif opcode == 79:  # callpropvoid
 485                     index = u30()
 486                     mname = self.multinames[index]
 487                     arg_count = u30()
 488                     args = list(reversed(
 489                         [stack.pop() for _ in range(arg_count)]))
 490                     obj = stack.pop()
 491                     if mname == 'reverse':
 492                         assert isinstance(obj, list)
 493                         obj.reverse()
 494                     else:
 495                         raise NotImplementedError(
 496                             'Unsupported (void) property %r on %r'
 497                             % (mname, obj))
 498                 elif opcode == 86:  # newarray
 499                     arg_count = u30()
 500                     arr = []
 501                     for i in range(arg_count):
 502                         arr.append(stack.pop())
 503                     arr = arr[::-1]
 504                     stack.append(arr)
 505                 elif opcode == 93:  # findpropstrict
 506                     index = u30()
 507                     mname = self.multinames[index]
 508                     for s in reversed(scopes):
 509                         if mname in s:
 510                             res = s
 511                             break
 512                     else:
 513                         res = scopes[0]
 514                     stack.append(res[mname])
 515                 elif opcode == 94:  # findproperty
 516                     index = u30()
 517                     mname = self.multinames[index]
 518                     for s in reversed(scopes):
 519                         if mname in s:
 520                             res = s
 521                             break
 522                     else:
 523                         res = avm_class.variables
 524                     stack.append(res)
 525                 elif opcode == 96:  # getlex
 526                     index = u30()
 527                     mname = self.multinames[index]
 528                     for s in reversed(scopes):
 529                         if mname in s:
 530                             scope = s
 531                             break
 532                     else:
 533                         scope = avm_class.variables
 534                     # I cannot find where static variables are initialized
 535                     # so let's just return None
 536                     res = scope.get(mname)
 537                     stack.append(res)
 538                 elif opcode == 97:  # setproperty
 539                     index = u30()
 540                     value = stack.pop()
 541                     idx = self.multinames[index]
 542                     if isinstance(idx, _Multiname):
 543                         idx = stack.pop()
 544                     obj = stack.pop()
 545                     obj[idx] = value
 546                 elif opcode == 98:  # getlocal
 547                     index = u30()
 548                     stack.append(registers[index])
 549                 elif opcode == 99:  # setlocal
 550                     index = u30()
 551                     value = stack.pop()
 552                     registers[index] = value
 553                 elif opcode == 102:  # getproperty
 554                     index = u30()
 555                     pname = self.multinames[index]
 556                     if pname == 'length':
 557                         obj = stack.pop()
 558                         assert isinstance(obj, list)
 559                         stack.append(len(obj))
 560                     else:  # Assume attribute access
 561                         idx = stack.pop()
 562                         assert isinstance(idx, int)
 563                         obj = stack.pop()
 564                         assert isinstance(obj, list)
 565                         stack.append(obj[idx])
 566                 elif opcode == 115:  # convert_
 567                     value = stack.pop()
 568                     intvalue = int(value)
 569                     stack.append(intvalue)
 570                 elif opcode == 128:  # coerce
 571                     u30()
 572                 elif opcode == 133:  # coerce_s
 573                     assert isinstance(stack[-1], (type(None), compat_str))
 574                 elif opcode == 160:  # add
 575                     value2 = stack.pop()
 576                     value1 = stack.pop()
 577                     res = value1 + value2
 578                     stack.append(res)
 579                 elif opcode == 161:  # subtract
 580                     value2 = stack.pop()
 581                     value1 = stack.pop()
 582                     res = value1 - value2
 583                     stack.append(res)
 584                 elif opcode == 164:  # modulo
 585                     value2 = stack.pop()
 586                     value1 = stack.pop()
 587                     res = value1 % value2
 588                     stack.append(res)
 589                 elif opcode == 175:  # greaterequals
 590                     value2 = stack.pop()
 591                     value1 = stack.pop()
 592                     result = value1 >= value2
 593                     stack.append(result)
 594                 elif opcode == 208:  # getlocal_0
 595                     stack.append(registers[0])
 596                 elif opcode == 209:  # getlocal_1
 597                     stack.append(registers[1])
 598                 elif opcode == 210:  # getlocal_2
 599                     stack.append(registers[2])
 600                 elif opcode == 211:  # getlocal_3
 601                     stack.append(registers[3])
 602                 elif opcode == 212:  # setlocal_0
 603                     registers[0] = stack.pop()
 604                 elif opcode == 213:  # setlocal_1
 605                     registers[1] = stack.pop()
 606                 elif opcode == 214:  # setlocal_2
 607                     registers[2] = stack.pop()
 608                 elif opcode == 215:  # setlocal_3
 609                     registers[3] = stack.pop()
 610                 else:
 611                     raise NotImplementedError(
 612                         'Unsupported opcode %d' % opcode)
 613
 614         avm_class.method_pyfunctions[func_name] = resfunc
 615         return resfunc
 616