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):
153 self._patched_functions = {}
155 for tag_code, tag in _extract_tags(file_contents)
157 p = code_tag.index(b'\0', 4) + 1
158 code_reader = io.BytesIO(code_tag[p:])
160 # Parse ABC (AVM2 ByteCode)
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)
169 # minor_version + major_version
174 for _c in range(1, int_count):
177 for _c in range(1, uint_count):
180 read_bytes(max(0, (double_count - 1)) * 8)
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):
191 for _c in range(1, ns_set_count):
193 for _c2 in range(count):
195 multiname_count = u30()
204 0x0e: 2, # MultinameA
205 0x1b: 1, # MultinameL
206 0x1c: 1, # MultinameLA
208 self.multinames = ['']
209 for _c in range(1, multiname_count):
211 assert kind in MULTINAME_SIZES, 'Invalid multiname kind %r' % kind
213 u30() # namespace_idx
215 self.multinames.append(self.constant_strings[name_idx])
217 self.multinames.append(_Multiname(kind))
218 for _c2 in range(MULTINAME_SIZES[kind]):
223 MethodInfo = collections.namedtuple(
225 ['NEED_ARGUMENTS', 'NEED_REST'])
227 for method_id in range(method_count):
230 for _ in range(param_count):
232 u30() # name index (always 0 for youtube)
234 if flags & 0x08 != 0:
237 for c in range(option_count):
240 if flags & 0x80 != 0:
241 # Param names present
242 for _ in range(param_count):
244 mi = MethodInfo(flags & 0x01 != 0, flags & 0x04 != 0)
245 method_infos.append(mi)
248 metadata_count = u30()
249 for _c in range(metadata_count):
252 for _c2 in range(item_count):
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
262 if kind in [0x00, 0x06]: # Slot or Const
264 u30() # type_name_idx
268 elif kind in [0x01, 0x02, 0x03]: # Method / Getter / Setter
271 methods[self.multinames[trait_name_idx]] = method_idx
272 elif kind == 0x04: # Class
275 elif kind == 0x05: # Function
278 methods[function_idx] = self.multinames[trait_name_idx]
280 raise ExtractorError('Unsupported trait kind %d' % kind)
282 if attrs & 0x4 != 0: # Metadata present
283 metadata_count = u30()
284 for _c3 in range(metadata_count):
285 u30() # metadata index
292 for class_id in range(class_count):
295 cname = self.multinames[name_idx]
296 avm_class = _AVMClass(name_idx, cname)
297 classes.append(avm_class)
299 u30() # super_name idx
301 if flags & 0x08 != 0: # Protected namespace is present
302 u30() # protected_ns_idx
304 for _c2 in range(intrf_count):
308 for _c2 in range(trait_count):
309 trait_methods = parse_traits_info()
310 avm_class.register_methods(trait_methods)
312 assert len(classes) == class_count
313 self._classes_by_name = dict((c.name, c) for c in classes)
315 for avm_class in classes:
318 for _c2 in range(trait_count):
319 trait_methods = parse_traits_info()
320 avm_class.register_methods(trait_methods)
324 for _c in range(script_count):
327 for _c2 in range(trait_count):
331 method_body_count = u30()
332 Method = collections.namedtuple('Method', ['code', 'local_count'])
333 for _c in range(method_body_count):
337 u30() # init_scope_depth
338 u30() # max_scope_depth
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):
353 for _c2 in range(trait_count):
356 assert p + code_reader.tell() == len(code_tag)
358 def patch_function(self, avm_class, func_name, f):
359 self._patched_functions[(avm_class, func_name)] = f
361 def extract_class(self, class_name):
363 return self._classes_by_name[class_name]
365 raise ExtractorError('Class %r not found' % class_name)
367 def extract_function(self, avm_class, func_name):
368 p = self._patched_functions.get((avm_class, func_name))
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]
382 coder = io.BytesIO(m.code)
383 s24 = lambda: _s24(coder)
384 u30 = lambda: _u30(coder)
386 registers = [avm_class.variables] + list(args) + [None] * m.local_count
388 scopes = collections.deque([
389 self._classes_by_name, avm_class.variables])
391 opcode = _read_byte(coder)
392 if opcode == 17: # iftrue
396 coder.seek(coder.tell() + offset)
397 elif opcode == 18: # iffalse
401 coder.seek(coder.tell() + offset)
402 elif opcode == 36: # pushbyte
403 v = _read_byte(coder)
405 elif opcode == 42: # dup
408 elif opcode == 44: # pushstring
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
416 args = list(reversed(
417 [stack.pop() for _ in range(arg_count)]))
419 res = obj.avm_class.make_object()
421 elif opcode == 70: # callproperty
423 mname = self.multinames[index]
425 args = list(reversed(
426 [stack.pop() for _ in range(arg_count)]))
429 if isinstance(obj, _AVMClass_Object):
430 func = self.extract_function(obj.avm_class, mname)
434 elif isinstance(obj, _ScopeDict):
435 if mname in obj.avm_class.method_names:
436 func = self.extract_function(obj.avm_class, mname)
442 elif isinstance(obj, compat_str):
444 assert len(args) == 1
445 assert isinstance(args[0], compat_str)
449 res = obj.split(args[0])
452 elif isinstance(obj, list):
454 assert len(args) == 1
455 assert isinstance(args[0], int)
459 elif mname == 'join':
460 assert len(args) == 1
461 assert isinstance(args[0], compat_str)
462 res = args[0].join(obj)
465 raise NotImplementedError(
466 'Unsupported property %r on %r'
468 elif opcode == 72: # returnvalue
471 elif opcode == 74: # constructproperty
474 args = list(reversed(
475 [stack.pop() for _ in range(arg_count)]))
478 mname = self.multinames[index]
479 assert isinstance(obj, _AVMClass)
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
486 mname = self.multinames[index]
488 args = list(reversed(
489 [stack.pop() for _ in range(arg_count)]))
491 if mname == 'reverse':
492 assert isinstance(obj, list)
495 raise NotImplementedError(
496 'Unsupported (void) property %r on %r'
498 elif opcode == 86: # newarray
501 for i in range(arg_count):
502 arr.append(stack.pop())
505 elif opcode == 93: # findpropstrict
507 mname = self.multinames[index]
508 for s in reversed(scopes):
514 stack.append(res[mname])
515 elif opcode == 94: # findproperty
517 mname = self.multinames[index]
518 for s in reversed(scopes):
523 res = avm_class.variables
525 elif opcode == 96: # getlex
527 mname = self.multinames[index]
528 for s in reversed(scopes):
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)
538 elif opcode == 97: # setproperty
541 idx = self.multinames[index]
542 if isinstance(idx, _Multiname):
546 elif opcode == 98: # getlocal
548 stack.append(registers[index])
549 elif opcode == 99: # setlocal
552 registers[index] = value
553 elif opcode == 102: # getproperty
555 pname = self.multinames[index]
556 if pname == 'length':
558 assert isinstance(obj, list)
559 stack.append(len(obj))
560 else: # Assume attribute access
562 assert isinstance(idx, int)
564 assert isinstance(obj, list)
565 stack.append(obj[idx])
566 elif opcode == 115: # convert_
568 intvalue = int(value)
569 stack.append(intvalue)
570 elif opcode == 128: # coerce
572 elif opcode == 133: # coerce_s
573 assert isinstance(stack[-1], (type(None), compat_str))
574 elif opcode == 160: # add
577 res = value1 + value2
579 elif opcode == 161: # subtract
582 res = value1 - value2
584 elif opcode == 164: # modulo
587 res = value1 % value2
589 elif opcode == 175: # greaterequals
592 result = value1 >= value2
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()
611 raise NotImplementedError(
612 'Unsupported opcode %d' % opcode)
614 avm_class.method_pyfunctions[func_name] = resfunc