1 from __future__ import unicode_literals
14 def _extract_tags(file_contents):
15 if file_contents[1:3] != b'WS':
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:])
21 raise NotImplementedError(
22 'Unsupported compression format %r' %
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
29 pos = framesize_len + 2 + 2
30 while pos < len(content):
31 header16 = struct.unpack('<H', content[pos:pos + 2])[0]
33 tag_code = header16 >> 6
34 tag_len = header16 & 0x3f
36 tag_len = struct.unpack('<I', content[pos:pos + 4])[0]
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])
45 class _AVMClass_Object(object):
46 def __init__(self, avm_class):
47 self.avm_class = avm_class
50 return '%s#%x' % (self.avm_class.name, id(self))
53 class _ScopeDict(dict):
54 def __init__(self, avm_class):
55 super(_ScopeDict, self).__init__()
56 self.avm_class = avm_class
59 return '%s__Scope(%s)' % (
61 super(_ScopeDict, self).__repr__())
64 class _AVMClass(object):
65 def __init__(self, name_idx, name):
66 self.name_idx = name_idx
68 self.method_names = {}
71 self.method_pyfunctions = {}
73 self.variables = _ScopeDict(self)
75 def make_object(self):
76 return _AVMClass_Object(self)
79 return '_AVMClass(%s)' % (self.name)
81 def register_methods(self, methods):
82 self.method_names.update(methods.items())
83 self.method_idxs.update(dict(
85 for name, idx in methods.items()))
88 class _Multiname(object):
89 def __init__(self, kind):
93 return '[MULTINAME kind: 0x%x]' % self.kind
96 def _read_int(reader):
102 b = struct.unpack('<B', buf)[0]
103 res = res | ((b & 0x7f) << shift)
111 res = _read_int(reader)
112 assert res & 0xf0000000 == 0
118 v = _read_int(reader)
119 if v & 0x80000000 != 0:
120 v = - ((v ^ 0xffffffff) + 1)
127 last_byte = b'\xff' if (ord(bs[2:3]) >= 0x80) else b'\x00'
128 return struct.unpack('<i', bs + last_byte)[0]
131 def _read_string(reader):
133 resb = reader.read(slen)
134 assert len(resb) == slen
135 return resb.decode('utf-8')
138 def _read_bytes(count, reader):
140 resb = reader.read(count)
141 assert len(resb) == count
145 def _read_byte(reader):
146 resb = _read_bytes(1, reader=reader)
147 res = struct.unpack('<B', resb)[0]
151 class SWFInterpreter(object):
152 def __init__(self, file_contents):
154 for tag_code, tag in _extract_tags(file_contents)
156 p = code_tag.index(b'\0', 4) + 1
157 code_reader = io.BytesIO(code_tag[p:])
159 # Parse ABC (AVM2 ByteCode)
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)
168 # minor_version + major_version
173 for _c in range(1, int_count):
176 for _c in range(1, uint_count):
179 read_bytes(max(0, (double_count - 1)) * 8)
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):
190 for _c in range(1, ns_set_count):
192 for _c2 in range(count):
194 multiname_count = u30()
203 0x0e: 2, # MultinameA
204 0x1b: 1, # MultinameL
205 0x1c: 1, # MultinameLA
207 self.multinames = ['']
208 for _c in range(1, multiname_count):
210 assert kind in MULTINAME_SIZES, 'Invalid multiname kind %r' % kind
212 u30() # namespace_idx
214 self.multinames.append(self.constant_strings[name_idx])
216 self.multinames.append(_Multiname(kind))
217 for _c2 in range(MULTINAME_SIZES[kind]):
222 MethodInfo = collections.namedtuple(
224 ['NEED_ARGUMENTS', 'NEED_REST'])
226 for method_id in range(method_count):
229 for _ in range(param_count):
231 u30() # name index (always 0 for youtube)
233 if flags & 0x08 != 0:
236 for c in range(option_count):
239 if flags & 0x80 != 0:
240 # Param names present
241 for _ in range(param_count):
243 mi = MethodInfo(flags & 0x01 != 0, flags & 0x04 != 0)
244 method_infos.append(mi)
247 metadata_count = u30()
248 for _c in range(metadata_count):
251 for _c2 in range(item_count):
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
261 if kind in [0x00, 0x06]: # Slot or Const
263 u30() # type_name_idx
267 elif kind in [0x01, 0x02, 0x03]: # Method / Getter / Setter
270 methods[self.multinames[trait_name_idx]] = method_idx
271 elif kind == 0x04: # Class
274 elif kind == 0x05: # Function
277 methods[function_idx] = self.multinames[trait_name_idx]
279 raise ExtractorError('Unsupported trait kind %d' % kind)
281 if attrs & 0x4 != 0: # Metadata present
282 metadata_count = u30()
283 for _c3 in range(metadata_count):
284 u30() # metadata index
291 for class_id in range(class_count):
294 cname = self.multinames[name_idx]
295 avm_class = _AVMClass(name_idx, cname)
296 classes.append(avm_class)
298 u30() # super_name idx
300 if flags & 0x08 != 0: # Protected namespace is present
301 u30() # protected_ns_idx
303 for _c2 in range(intrf_count):
307 for _c2 in range(trait_count):
308 trait_methods = parse_traits_info()
309 avm_class.register_methods(trait_methods)
311 assert len(classes) == class_count
312 self._classes_by_name = dict((c.name, c) for c in classes)
314 for avm_class in classes:
317 for _c2 in range(trait_count):
318 trait_methods = parse_traits_info()
319 avm_class.register_methods(trait_methods)
323 for _c in range(script_count):
326 for _c2 in range(trait_count):
330 method_body_count = u30()
331 Method = collections.namedtuple('Method', ['code', 'local_count'])
332 for _c in range(method_body_count):
336 u30() # init_scope_depth
337 u30() # max_scope_depth
339 code = read_bytes(code_length)
340 for avm_class in classes:
341 if method_idx in avm_class.method_idxs:
342 m = Method(code, local_count)
343 avm_class.methods[avm_class.method_idxs[method_idx]] = m
344 exception_count = u30()
345 for _c2 in range(exception_count):
352 for _c2 in range(trait_count):
355 assert p + code_reader.tell() == len(code_tag)
357 def extract_class(self, class_name):
359 return self._classes_by_name[class_name]
361 raise ExtractorError('Class %r not found' % class_name)
363 def extract_function(self, avm_class, func_name):
364 print('Extracting %s.%s' % (avm_class.name, func_name))
365 if func_name in avm_class.method_pyfunctions:
366 return avm_class.method_pyfunctions[func_name]
367 if func_name in self._classes_by_name:
368 return self._classes_by_name[func_name].make_object()
369 if func_name not in avm_class.methods:
370 raise ExtractorError('Cannot find function %s.%s' % (
371 avm_class.name, func_name))
372 m = avm_class.methods[func_name]
376 coder = io.BytesIO(m.code)
377 s24 = lambda: _s24(coder)
378 u30 = lambda: _u30(coder)
380 print('Invoking %s.%s(%r)' % (avm_class.name, func_name, tuple(args)))
381 registers = [avm_class.variables] + list(args) + [None] * m.local_count
383 scopes = collections.deque([
384 self._classes_by_name, avm_class.variables])
386 opcode = _read_byte(coder)
387 print('opcode: %r, stack(%d): %r' % (opcode, len(stack), stack))
388 if opcode == 17: # iftrue
392 coder.seek(coder.tell() + offset)
393 elif opcode == 18: # iffalse
397 coder.seek(coder.tell() + offset)
398 elif opcode == 36: # pushbyte
399 v = _read_byte(coder)
401 elif opcode == 42: # dup
404 elif opcode == 44: # pushstring
406 stack.append(self.constant_strings[idx])
407 elif opcode == 48: # pushscope
408 new_scope = stack.pop()
409 scopes.append(new_scope)
410 elif opcode == 66: # construct
412 args = list(reversed(
413 [stack.pop() for _ in range(arg_count)]))
415 res = obj.avm_class.make_object()
417 elif opcode == 70: # callproperty
419 mname = self.multinames[index]
421 args = list(reversed(
422 [stack.pop() for _ in range(arg_count)]))
425 if isinstance(obj, _AVMClass_Object):
426 func = self.extract_function(obj.avm_class, mname)
430 elif isinstance(obj, _ScopeDict):
431 if mname in obj.avm_class.method_names:
432 func = self.extract_function(obj.avm_class, mname)
438 elif isinstance(obj, compat_str):
440 assert len(args) == 1
441 assert isinstance(args[0], compat_str)
445 res = obj.split(args[0])
448 elif isinstance(obj, list):
450 assert len(args) == 1
451 assert isinstance(args[0], int)
455 elif mname == 'join':
456 assert len(args) == 1
457 assert isinstance(args[0], compat_str)
458 res = args[0].join(obj)
461 raise NotImplementedError(
462 'Unsupported property %r on %r'
464 elif opcode == 72: # returnvalue
467 elif opcode == 74: # constructproperty
470 args = list(reversed(
471 [stack.pop() for _ in range(arg_count)]))
474 mname = self.multinames[index]
475 assert isinstance(obj, _AVMClass)
476 construct_method = self.extract_function(
478 # We do not actually call the constructor for now;
479 # we just pretend it does nothing
480 stack.append(obj.make_object())
481 elif opcode == 79: # callpropvoid
483 mname = self.multinames[index]
485 args = list(reversed(
486 [stack.pop() for _ in range(arg_count)]))
488 if mname == 'reverse':
489 assert isinstance(obj, list)
492 raise NotImplementedError(
493 'Unsupported (void) property %r on %r'
495 elif opcode == 86: # newarray
498 for i in range(arg_count):
499 arr.append(stack.pop())
502 elif opcode == 93: # findpropstrict
504 mname = self.multinames[index]
505 for s in reversed(scopes):
511 stack.append(res[mname])
512 elif opcode == 94: # findproperty
514 mname = self.multinames[index]
515 for s in reversed(scopes):
520 res = avm_class.variables
522 elif opcode == 96: # getlex
524 mname = self.multinames[index]
525 for s in reversed(scopes):
530 scope = avm_class.variables
531 # I cannot find where static variables are initialized
532 # so let's just return None
533 res = scope.get(mname)
535 elif opcode == 97: # setproperty
538 idx = self.multinames[index]
539 if isinstance(idx, _Multiname):
542 print('Setting %r.%r = %r' % (obj, idx, value))
544 elif opcode == 98: # getlocal
546 stack.append(registers[index])
547 elif opcode == 99: # setlocal
550 registers[index] = value
551 elif opcode == 102: # getproperty
553 pname = self.multinames[index]
554 if pname == 'length':
556 assert isinstance(obj, list)
557 stack.append(len(obj))
558 else: # Assume attribute access
560 assert isinstance(idx, int)
562 assert isinstance(obj, list)
563 stack.append(obj[idx])
564 elif opcode == 115: # convert_
566 intvalue = int(value)
567 stack.append(intvalue)
568 elif opcode == 128: # coerce
570 elif opcode == 133: # coerce_s
571 assert isinstance(stack[-1], (type(None), compat_str))
572 elif opcode == 160: # add
575 res = value1 + value2
577 elif opcode == 161: # subtract
580 res = value1 - value2
582 elif opcode == 164: # modulo
585 res = value1 % value2
587 elif opcode == 175: # greaterequals
590 result = value1 >= value2
592 elif opcode == 208: # getlocal_0
593 stack.append(registers[0])
594 elif opcode == 209: # getlocal_1
595 stack.append(registers[1])
596 elif opcode == 210: # getlocal_2
597 stack.append(registers[2])
598 elif opcode == 211: # getlocal_3
599 stack.append(registers[3])
600 elif opcode == 212: # setlocal_0
601 registers[0] = stack.pop()
602 elif opcode == 213: # setlocal_1
603 registers[1] = stack.pop()
604 elif opcode == 214: # setlocal_2
605 registers[2] = stack.pop()
606 elif opcode == 215: # setlocal_3
607 registers[3] = stack.pop()
609 raise NotImplementedError(
610 'Unsupported opcode %d' % opcode)
612 avm_class.method_pyfunctions[func_name] = resfunc