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