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