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

   1 from __future__ import unicode_literals
   2
   3 import collections
   4 import io
   5 import struct
   6 import zlib
   7
   8 from .utils import (
   9     compat_str,
  10     ExtractorError,
  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 _AVM_Object(object):
  46     def __init__(self, value=None, name_hint=None):
  47         self.value = value
  48         self.name_hint = name_hint
  49
  50     def __repr__(self):
  51         nh = '' if self.name_hint is None else (' %s' % self.name_hint)
  52         return 'AVMObject%s(%r)' % (nh, self.value)
  53
  54
  55 class _AVMClass_Object(object):
  56     def __init__(self, avm_class):
  57         self.avm_class = avm_class
  58
  59     def __repr__(self):
  60         return '%s#%x' % (self.avm_class.name, id(self))
  61
  62
  63 class _AVMClass(object):
  64     def __init__(self, name_idx, name):
  65         self.name_idx = name_idx
  66         self.name = name
  67         self.method_names = {}
  68         self.method_idxs = {}
  69         self.methods = {}
  70         self.method_pyfunctions = {}
  71
  72         class ScopeDict(dict):
  73             def __init__(self, avm_class):
  74                 super(ScopeDict, self).__init__()
  75                 self.avm_class = avm_class
  76
  77             def __getitem__(self, k):
  78                 print('getting %r' % k)
  79                 return super(ScopeDict, self).__getitem__(k)
  80
  81             def __contains__(self, k):
  82                 print('contains %r' % k)
  83                 return super(ScopeDict, self).__contains__(k)
  84
  85             def __repr__(self):
  86                 return '%s__Scope(%s)' % (
  87                     self.avm_class.name,
  88                     super(ScopeDict, self).__repr__())
  89
  90         self.variables = ScopeDict(self)
  91
  92     def make_object(self):
  93         return _AVMClass_Object(self)
  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: %d]' % 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             classes.append(_AVMClass(name_idx, self.multinames[name_idx]))
 294             u30()  # super_name idx
 295             flags = read_byte()
 296             if flags & 0x08 != 0:  # Protected namespace is present
 297                 u30()  # protected_ns_idx
 298             intrf_count = u30()
 299             for _c2 in range(intrf_count):
 300                 u30()
 301             u30()  # iinit
 302             trait_count = u30()
 303             for _c2 in range(trait_count):
 304                 parse_traits_info()
 305         assert len(classes) == class_count
 306         self._classes_by_name = dict((c.name, c) for c in classes)
 307
 308         for avm_class in classes:
 309             u30()  # cinit
 310             trait_count = u30()
 311             for _c2 in range(trait_count):
 312                 trait_methods = parse_traits_info()
 313                 avm_class.method_names.update(trait_methods.items())
 314                 avm_class.method_idxs.update(dict(
 315                     (idx, name)
 316                     for name, idx in trait_methods.items()))
 317
 318         # Scripts
 319         script_count = u30()
 320         for _c in range(script_count):
 321             u30()  # init
 322             trait_count = u30()
 323             for _c2 in range(trait_count):
 324                 parse_traits_info()
 325
 326         # Method bodies
 327         method_body_count = u30()
 328         Method = collections.namedtuple('Method', ['code', 'local_count'])
 329         for _c in range(method_body_count):
 330             method_idx = u30()
 331             u30()  # max_stack
 332             local_count = u30()
 333             u30()  # init_scope_depth
 334             u30()  # max_scope_depth
 335             code_length = u30()
 336             code = read_bytes(code_length)
 337             for avm_class in classes:
 338                 if method_idx in avm_class.method_idxs:
 339                     m = Method(code, local_count)
 340                     avm_class.methods[avm_class.method_idxs[method_idx]] = m
 341             exception_count = u30()
 342             for _c2 in range(exception_count):
 343                 u30()  # from
 344                 u30()  # to
 345                 u30()  # target
 346                 u30()  # exc_type
 347                 u30()  # var_name
 348             trait_count = u30()
 349             for _c2 in range(trait_count):
 350                 parse_traits_info()
 351
 352         assert p + code_reader.tell() == len(code_tag)
 353
 354     def extract_class(self, class_name):
 355         try:
 356             return self._classes_by_name[class_name]
 357         except KeyError:
 358             raise ExtractorError('Class %r not found' % class_name)
 359
 360     def extract_function(self, avm_class, func_name):
 361         if func_name in avm_class.method_pyfunctions:
 362             return avm_class.method_pyfunctions[func_name]
 363         if func_name in self._classes_by_name:
 364             return self._classes_by_name[func_name].make_object()
 365         if func_name not in avm_class.methods:
 366             raise ExtractorError('Cannot find function %r' % func_name)
 367         m = avm_class.methods[func_name]
 368
 369         def resfunc(args):
 370             # Helper functions
 371             coder = io.BytesIO(m.code)
 372             s24 = lambda: _s24(coder)
 373             u30 = lambda: _u30(coder)
 374
 375             print('Invoking %s.%s(%r)' % (avm_class.name, func_name, tuple(args)))
 376             registers = [avm_class.variables] + list(args) + [None] * m.local_count
 377             stack = []
 378             scopes = collections.deque([avm_class.variables])
 379             while True:
 380                 opcode = _read_byte(coder)
 381                 print('opcode: %r, stack(%d): %r' % (opcode, len(stack), stack))
 382                 if opcode == 17:  # iftrue
 383                     offset = s24()
 384                     value = stack.pop()
 385                     if value:
 386                         coder.seek(coder.tell() + offset)
 387                 elif opcode == 18:  # iffalse
 388                     offset = s24()
 389                     value = stack.pop()
 390                     if not value:
 391                         coder.seek(coder.tell() + offset)
 392                 elif opcode == 36:  # pushbyte
 393                     v = _read_byte(coder)
 394                     stack.append(v)
 395                 elif opcode == 42:  # dup
 396                     value = stack[-1]
 397                     stack.append(value)
 398                 elif opcode == 44:  # pushstring
 399                     idx = u30()
 400                     stack.append(self.constant_strings[idx])
 401                 elif opcode == 48:  # pushscope
 402                     new_scope = stack.pop()
 403                     scopes.append(new_scope)
 404                 elif opcode == 70:  # callproperty
 405                     index = u30()
 406                     mname = self.multinames[index]
 407                     arg_count = u30()
 408                     args = list(reversed(
 409                         [stack.pop() for _ in range(arg_count)]))
 410                     obj = stack.pop()
 411                     if mname == 'split':
 412                         assert len(args) == 1
 413                         assert isinstance(args[0], compat_str)
 414                         assert isinstance(obj, compat_str)
 415                         if args[0] == '':
 416                             res = list(obj)
 417                         else:
 418                             res = obj.split(args[0])
 419                         stack.append(res)
 420                     elif mname == 'slice':
 421                         assert len(args) == 1
 422                         assert isinstance(args[0], int)
 423                         assert isinstance(obj, list)
 424                         res = obj[args[0]:]
 425                         stack.append(res)
 426                     elif mname == 'join':
 427                         assert len(args) == 1
 428                         assert isinstance(args[0], compat_str)
 429                         assert isinstance(obj, list)
 430                         res = args[0].join(obj)
 431                         stack.append(res)
 432                     elif mname in avm_class.method_pyfunctions:
 433                         stack.append(avm_class.method_pyfunctions[mname](args))
 434                     else:
 435                         raise NotImplementedError(
 436                             'Unsupported property %r on %r'
 437                             % (mname, obj))
 438                 elif opcode == 72:  # returnvalue
 439                     res = stack.pop()
 440                     return res
 441                 elif opcode == 74:  # constructproperty
 442                     index = u30()
 443                     arg_count = u30()
 444                     args = list(reversed(
 445                         [stack.pop() for _ in range(arg_count)]))
 446                     obj = stack.pop()
 447
 448                     mname = self.multinames[index]
 449                     construct_method = self.extract_function(
 450                         obj.avm_class, mname)
 451                     # We do not actually call the constructor for now;
 452                     # we just pretend it does nothing
 453                     stack.append(obj)
 454                 elif opcode == 79:  # callpropvoid
 455                     index = u30()
 456                     mname = self.multinames[index]
 457                     arg_count = u30()
 458                     args = list(reversed(
 459                         [stack.pop() for _ in range(arg_count)]))
 460                     obj = stack.pop()
 461                     if mname == 'reverse':
 462                         assert isinstance(obj, list)
 463                         obj.reverse()
 464                     else:
 465                         raise NotImplementedError(
 466                             'Unsupported (void) property %r on %r'
 467                             % (mname, obj))
 468                 elif opcode == 86:  # newarray
 469                     arg_count = u30()
 470                     arr = []
 471                     for i in range(arg_count):
 472                         arr.append(stack.pop())
 473                     arr = arr[::-1]
 474                     stack.append(arr)
 475                 elif opcode == 93:  # findpropstrict
 476                     index = u30()
 477                     mname = self.multinames[index]
 478                     for s in reversed(scopes):
 479                         if mname in s:
 480                             res = s
 481                             break
 482                     else:
 483                         res = scopes[0]
 484                     stack.append(res)
 485                 elif opcode == 94:  # findproperty
 486                     index = u30()
 487                     mname = self.multinames[index]
 488                     for s in reversed(scopes):
 489                         if mname in s:
 490                             res = s
 491                             break
 492                     else:
 493                         res = scopes[0]
 494                     stack.append(res)
 495                 elif opcode == 96:  # getlex
 496                     index = u30()
 497                     mname = self.multinames[index]
 498                     for s in reversed(scopes):
 499                         if mname in s:
 500                             scope = s
 501                             break
 502                     else:
 503                         scope = scopes[0]
 504                     # I cannot find where static variables are initialized
 505                     # so let's just return None
 506                     res = scope.get(mname)
 507                     stack.append(res)
 508                 elif opcode == 97:  # setproperty
 509                     index = u30()
 510                     value = stack.pop()
 511                     idx = self.multinames[index]
 512                     obj = stack.pop()
 513                     obj[idx] = value
 514                 elif opcode == 98:  # getlocal
 515                     index = u30()
 516                     stack.append(registers[index])
 517                 elif opcode == 99:  # setlocal
 518                     index = u30()
 519                     value = stack.pop()
 520                     registers[index] = value
 521                 elif opcode == 102:  # getproperty
 522                     index = u30()
 523                     pname = self.multinames[index]
 524                     if pname == 'length':
 525                         obj = stack.pop()
 526                         assert isinstance(obj, list)
 527                         stack.append(len(obj))
 528                     else:  # Assume attribute access
 529                         idx = stack.pop()
 530                         assert isinstance(idx, int)
 531                         obj = stack.pop()
 532                         assert isinstance(obj, list)
 533                         stack.append(obj[idx])
 534                 elif opcode == 115:  # convert_
 535                     value = stack.pop()
 536                     intvalue = int(value)
 537                     stack.append(intvalue)
 538                 elif opcode == 128:  # coerce
 539                     u30()
 540                 elif opcode == 133:  # coerce_s
 541                     assert isinstance(stack[-1], (type(None), compat_str))
 542                 elif opcode == 160:  # add
 543                     value2 = stack.pop()
 544                     value1 = stack.pop()
 545                     res = value1 + value2
 546                     stack.append(res)
 547                 elif opcode == 161:  # subtract
 548                     value2 = stack.pop()
 549                     value1 = stack.pop()
 550                     res = value1 - value2
 551                     stack.append(res)
 552                 elif opcode == 164:  # modulo
 553                     value2 = stack.pop()
 554                     value1 = stack.pop()
 555                     res = value1 % value2
 556                     stack.append(res)
 557                 elif opcode == 175:  # greaterequals
 558                     value2 = stack.pop()
 559                     value1 = stack.pop()
 560                     result = value1 >= value2
 561                     stack.append(result)
 562                 elif opcode == 208:  # getlocal_0
 563                     stack.append(registers[0])
 564                 elif opcode == 209:  # getlocal_1
 565                     stack.append(registers[1])
 566                 elif opcode == 210:  # getlocal_2
 567                     stack.append(registers[2])
 568                 elif opcode == 211:  # getlocal_3
 569                     stack.append(registers[3])
 570                 elif opcode == 212:  # setlocal_0
 571                     registers[0] = stack.pop()
 572                 elif opcode == 213:  # setlocal_1
 573                     registers[1] = stack.pop()
 574                 elif opcode == 214:  # setlocal_2
 575                     registers[2] = stack.pop()
 576                 elif opcode == 215:  # setlocal_3
 577                     registers[3] = stack.pop()
 578                 else:
 579                     raise NotImplementedError(
 580                         'Unsupported opcode %d' % opcode)
 581
 582         avm_class.method_pyfunctions[func_name] = resfunc
 583         return resfunc
 584