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, static_properties=None):
66 self.name_idx = name_idx
68 self.method_names = {}
71 self.method_pyfunctions = {}
72 self.static_properties = static_properties if static_properties else {}
74 self.variables = _ScopeDict(self)
77 def make_object(self):
78 return _AVMClass_Object(self)
81 return '_AVMClass(%s)' % (self.name)
83 def register_methods(self, methods):
84 self.method_names.update(methods.items())
85 self.method_idxs.update(dict(
87 for name, idx in methods.items()))
90 class _Multiname(object):
91 def __init__(self, kind):
95 return '[MULTINAME kind: 0x%x]' % self.kind
98 def _read_int(reader):
104 b = struct_unpack('<B', buf)[0]
105 res = res | ((b & 0x7f) << shift)
113 res = _read_int(reader)
114 assert res & 0xf0000000 == 0
120 v = _read_int(reader)
121 if v & 0x80000000 != 0:
122 v = - ((v ^ 0xffffffff) + 1)
129 last_byte = b'\xff' if (ord(bs[2:3]) >= 0x80) else b'\x00'
130 return struct_unpack('<i', bs + last_byte)[0]
133 def _read_string(reader):
135 resb = reader.read(slen)
136 assert len(resb) == slen
137 return resb.decode('utf-8')
140 def _read_bytes(count, reader):
142 resb = reader.read(count)
143 assert len(resb) == count
147 def _read_byte(reader):
148 resb = _read_bytes(1, reader=reader)
149 res = struct_unpack('<B', resb)[0]
153 StringClass = _AVMClass('(no name idx)', 'String')
154 ByteArrayClass = _AVMClass('(no name idx)', 'ByteArray')
155 TimerClass = _AVMClass('(no name idx)', 'Timer')
156 TimerEventClass = _AVMClass('(no name idx)', 'TimerEvent', {'TIMER': 'timer'})
158 StringClass.name: StringClass,
159 ByteArrayClass.name: ByteArrayClass,
160 TimerClass.name: TimerClass,
161 TimerEventClass.name: TimerEventClass,
165 class _Undefined(object):
168 __nonzero__ = __bool__
177 undefined = _Undefined()
180 class SWFInterpreter(object):
181 def __init__(self, file_contents):
182 self._patched_functions = {
183 (TimerClass, 'addEventListener'): lambda params: undefined,
186 for tag_code, tag in _extract_tags(file_contents)
188 p = code_tag.index(b'\0', 4) + 1
189 code_reader = io.BytesIO(code_tag[p:])
191 # Parse ABC (AVM2 ByteCode)
193 # Define a couple convenience methods
194 u30 = lambda *args: _u30(*args, reader=code_reader)
195 s32 = lambda *args: _s32(*args, reader=code_reader)
196 u32 = lambda *args: _u32(*args, reader=code_reader)
197 read_bytes = lambda *args: _read_bytes(*args, reader=code_reader)
198 read_byte = lambda *args: _read_byte(*args, reader=code_reader)
200 # minor_version + major_version
205 self.constant_ints = [0]
206 for _c in range(1, int_count):
207 self.constant_ints.append(s32())
208 self.constant_uints = [0]
210 for _c in range(1, uint_count):
211 self.constant_uints.append(u32())
213 read_bytes(max(0, (double_count - 1)) * 8)
215 self.constant_strings = ['']
216 for _c in range(1, string_count):
217 s = _read_string(code_reader)
218 self.constant_strings.append(s)
219 namespace_count = u30()
220 for _c in range(1, namespace_count):
224 for _c in range(1, ns_set_count):
226 for _c2 in range(count):
228 multiname_count = u30()
237 0x0e: 2, # MultinameA
238 0x1b: 1, # MultinameL
239 0x1c: 1, # MultinameLA
241 self.multinames = ['']
242 for _c in range(1, multiname_count):
244 assert kind in MULTINAME_SIZES, 'Invalid multiname kind %r' % kind
246 u30() # namespace_idx
248 self.multinames.append(self.constant_strings[name_idx])
252 self.multinames.append(self.constant_strings[name_idx])
254 self.multinames.append(_Multiname(kind))
255 for _c2 in range(MULTINAME_SIZES[kind]):
260 MethodInfo = collections.namedtuple(
262 ['NEED_ARGUMENTS', 'NEED_REST'])
264 for method_id in range(method_count):
267 for _ in range(param_count):
269 u30() # name index (always 0 for youtube)
271 if flags & 0x08 != 0:
274 for c in range(option_count):
277 if flags & 0x80 != 0:
278 # Param names present
279 for _ in range(param_count):
281 mi = MethodInfo(flags & 0x01 != 0, flags & 0x04 != 0)
282 method_infos.append(mi)
285 metadata_count = u30()
286 for _c in range(metadata_count):
289 for _c2 in range(item_count):
293 def parse_traits_info():
294 trait_name_idx = u30()
295 kind_full = read_byte()
296 kind = kind_full & 0x0f
297 attrs = kind_full >> 4
300 if kind == 0x00: # Slot
302 u30() # type_name_idx
306 elif kind == 0x06: # Const
308 u30() # type_name_idx
313 if vkind == 0x03: # Constant_Int
314 value = self.constant_ints[vindex]
315 elif vkind == 0x04: # Constant_UInt
316 value = self.constant_uints[vindex]
318 return {}, None # Ignore silently for now
319 constants = {self.multinames[trait_name_idx]: value}
320 elif kind in (0x01, 0x02, 0x03): # Method / Getter / Setter
323 methods[self.multinames[trait_name_idx]] = method_idx
324 elif kind == 0x04: # Class
327 elif kind == 0x05: # Function
330 methods[function_idx] = self.multinames[trait_name_idx]
332 raise ExtractorError('Unsupported trait kind %d' % kind)
334 if attrs & 0x4 != 0: # Metadata present
335 metadata_count = u30()
336 for _c3 in range(metadata_count):
337 u30() # metadata index
339 return methods, constants
344 for class_id in range(class_count):
347 cname = self.multinames[name_idx]
348 avm_class = _AVMClass(name_idx, cname)
349 classes.append(avm_class)
351 u30() # super_name idx
353 if flags & 0x08 != 0: # Protected namespace is present
354 u30() # protected_ns_idx
356 for _c2 in range(intrf_count):
360 for _c2 in range(trait_count):
361 trait_methods, trait_constants = parse_traits_info()
362 avm_class.register_methods(trait_methods)
364 avm_class.constants.update(trait_constants)
366 assert len(classes) == class_count
367 self._classes_by_name = dict((c.name, c) for c in classes)
369 for avm_class in classes:
370 avm_class.cinit_idx = u30()
372 for _c2 in range(trait_count):
373 trait_methods, trait_constants = parse_traits_info()
374 avm_class.register_methods(trait_methods)
376 avm_class.constants.update(trait_constants)
380 for _c in range(script_count):
383 for _c2 in range(trait_count):
387 method_body_count = u30()
388 Method = collections.namedtuple('Method', ['code', 'local_count'])
389 self._all_methods = []
390 for _c in range(method_body_count):
394 u30() # init_scope_depth
395 u30() # max_scope_depth
397 code = read_bytes(code_length)
398 m = Method(code, local_count)
399 self._all_methods.append(m)
400 for avm_class in classes:
401 if method_idx in avm_class.method_idxs:
402 avm_class.methods[avm_class.method_idxs[method_idx]] = m
403 exception_count = u30()
404 for _c2 in range(exception_count):
411 for _c2 in range(trait_count):
414 assert p + code_reader.tell() == len(code_tag)
416 def patch_function(self, avm_class, func_name, f):
417 self._patched_functions[(avm_class, func_name)] = f
419 def extract_class(self, class_name, call_cinit=True):
421 res = self._classes_by_name[class_name]
423 raise ExtractorError('Class %r not found' % class_name)
425 if call_cinit and hasattr(res, 'cinit_idx'):
426 res.register_methods({'$cinit': res.cinit_idx})
427 res.methods['$cinit'] = self._all_methods[res.cinit_idx]
428 cinit = self.extract_function(res, '$cinit')
433 def extract_function(self, avm_class, func_name):
434 p = self._patched_functions.get((avm_class, func_name))
437 if func_name in avm_class.method_pyfunctions:
438 return avm_class.method_pyfunctions[func_name]
439 if func_name in self._classes_by_name:
440 return self._classes_by_name[func_name].make_object()
441 if func_name not in avm_class.methods:
442 raise ExtractorError('Cannot find function %s.%s' % (
443 avm_class.name, func_name))
444 m = avm_class.methods[func_name]
448 coder = io.BytesIO(m.code)
449 s24 = lambda: _s24(coder)
450 u30 = lambda: _u30(coder)
452 registers = [avm_class.variables] + list(args) + [None] * m.local_count
454 scopes = collections.deque([
455 self._classes_by_name, avm_class.constants, avm_class.variables])
457 opcode = _read_byte(coder)
458 if opcode == 9: # label
459 pass # Spec says: "Do nothing."
460 elif opcode == 16: # jump
462 coder.seek(coder.tell() + offset)
463 elif opcode == 17: # iftrue
467 coder.seek(coder.tell() + offset)
468 elif opcode == 18: # iffalse
472 coder.seek(coder.tell() + offset)
473 elif opcode == 19: # ifeq
478 coder.seek(coder.tell() + offset)
479 elif opcode == 20: # ifne
484 coder.seek(coder.tell() + offset)
485 elif opcode == 21: # iflt
490 coder.seek(coder.tell() + offset)
491 elif opcode == 32: # pushnull
493 elif opcode == 33: # pushundefined
494 stack.append(undefined)
495 elif opcode == 36: # pushbyte
496 v = _read_byte(coder)
498 elif opcode == 37: # pushshort
501 elif opcode == 38: # pushtrue
503 elif opcode == 39: # pushfalse
505 elif opcode == 40: # pushnan
506 stack.append(float('NaN'))
507 elif opcode == 42: # dup
510 elif opcode == 44: # pushstring
512 stack.append(self.constant_strings[idx])
513 elif opcode == 48: # pushscope
514 new_scope = stack.pop()
515 scopes.append(new_scope)
516 elif opcode == 66: # construct
518 args = list(reversed(
519 [stack.pop() for _ in range(arg_count)]))
521 res = obj.avm_class.make_object()
523 elif opcode == 70: # callproperty
525 mname = self.multinames[index]
527 args = list(reversed(
528 [stack.pop() for _ in range(arg_count)]))
531 if obj == StringClass:
532 if mname == 'String':
533 assert len(args) == 1
534 assert isinstance(args[0], (
535 int, compat_str, _Undefined))
536 if args[0] == undefined:
539 res = compat_str(args[0])
543 raise NotImplementedError(
544 'Function String.%s is not yet implemented'
546 elif isinstance(obj, _AVMClass_Object):
547 func = self.extract_function(obj.avm_class, mname)
551 elif isinstance(obj, _AVMClass):
552 func = self.extract_function(obj, mname)
556 elif isinstance(obj, _ScopeDict):
557 if mname in obj.avm_class.method_names:
558 func = self.extract_function(obj.avm_class, mname)
564 elif isinstance(obj, compat_str):
566 assert len(args) == 1
567 assert isinstance(args[0], compat_str)
571 res = obj.split(args[0])
574 elif mname == 'charCodeAt':
575 assert len(args) <= 1
576 idx = 0 if len(args) == 0 else args[0]
577 assert isinstance(idx, int)
581 elif isinstance(obj, list):
583 assert len(args) == 1
584 assert isinstance(args[0], int)
588 elif mname == 'join':
589 assert len(args) == 1
590 assert isinstance(args[0], compat_str)
591 res = args[0].join(obj)
594 raise NotImplementedError(
595 'Unsupported property %r on %r'
597 elif opcode == 71: # returnvoid
600 elif opcode == 72: # returnvalue
603 elif opcode == 73: # constructsuper
604 # Not yet implemented, just hope it works without it
606 args = list(reversed(
607 [stack.pop() for _ in range(arg_count)]))
609 elif opcode == 74: # constructproperty
612 args = list(reversed(
613 [stack.pop() for _ in range(arg_count)]))
616 mname = self.multinames[index]
617 assert isinstance(obj, _AVMClass)
619 # We do not actually call the constructor for now;
620 # we just pretend it does nothing
621 stack.append(obj.make_object())
622 elif opcode == 79: # callpropvoid
624 mname = self.multinames[index]
626 args = list(reversed(
627 [stack.pop() for _ in range(arg_count)]))
629 if isinstance(obj, _AVMClass_Object):
630 func = self.extract_function(obj.avm_class, mname)
632 assert res is undefined
634 if isinstance(obj, _ScopeDict):
635 assert mname in obj.avm_class.method_names
636 func = self.extract_function(obj.avm_class, mname)
638 assert res is undefined
640 if mname == 'reverse':
641 assert isinstance(obj, list)
644 raise NotImplementedError(
645 'Unsupported (void) property %r on %r'
647 elif opcode == 86: # newarray
650 for i in range(arg_count):
651 arr.append(stack.pop())
654 elif opcode == 93: # findpropstrict
656 mname = self.multinames[index]
657 for s in reversed(scopes):
663 if mname not in res and mname in _builtin_classes:
664 stack.append(_builtin_classes[mname])
666 stack.append(res[mname])
667 elif opcode == 94: # findproperty
669 mname = self.multinames[index]
670 for s in reversed(scopes):
675 res = avm_class.variables
677 elif opcode == 96: # getlex
679 mname = self.multinames[index]
680 for s in reversed(scopes):
685 scope = avm_class.variables
689 elif mname in _builtin_classes:
690 res = _builtin_classes[mname]
696 elif opcode == 97: # setproperty
699 idx = self.multinames[index]
700 if isinstance(idx, _Multiname):
704 elif opcode == 98: # getlocal
706 stack.append(registers[index])
707 elif opcode == 99: # setlocal
710 registers[index] = value
711 elif opcode == 102: # getproperty
713 pname = self.multinames[index]
714 if pname == 'length':
716 assert isinstance(obj, (compat_str, list))
717 stack.append(len(obj))
718 elif isinstance(pname, compat_str): # Member access
720 if isinstance(obj, _AVMClass):
721 res = obj.static_properties[pname]
725 assert isinstance(obj, (dict, _ScopeDict)),\
726 'Accessing member %r on %r' % (pname, obj)
727 res = obj.get(pname, undefined)
729 else: # Assume attribute access
731 assert isinstance(idx, int)
733 assert isinstance(obj, list)
734 stack.append(obj[idx])
735 elif opcode == 104: # initproperty
738 idx = self.multinames[index]
739 if isinstance(idx, _Multiname):
743 elif opcode == 115: # convert_
745 intvalue = int(value)
746 stack.append(intvalue)
747 elif opcode == 128: # coerce
749 elif opcode == 130: # coerce_a
751 # um, yes, it's any value
753 elif opcode == 133: # coerce_s
754 assert isinstance(stack[-1], (type(None), compat_str))
755 elif opcode == 147: # decrement
757 assert isinstance(value, int)
758 stack.append(value - 1)
759 elif opcode == 149: # typeof
762 _Undefined: 'undefined',
763 compat_str: 'String',
767 elif opcode == 160: # add
770 res = value1 + value2
772 elif opcode == 161: # subtract
775 res = value1 - value2
777 elif opcode == 162: # multiply
780 res = value1 * value2
782 elif opcode == 164: # modulo
785 res = value1 % value2
787 elif opcode == 168: # bitand
790 assert isinstance(value1, int)
791 assert isinstance(value2, int)
792 res = value1 & value2
794 elif opcode == 171: # equals
797 result = value1 == value2
799 elif opcode == 175: # greaterequals
802 result = value1 >= value2
804 elif opcode == 192: # increment_i
806 assert isinstance(value, int)
807 stack.append(value + 1)
808 elif opcode == 208: # getlocal_0
809 stack.append(registers[0])
810 elif opcode == 209: # getlocal_1
811 stack.append(registers[1])
812 elif opcode == 210: # getlocal_2
813 stack.append(registers[2])
814 elif opcode == 211: # getlocal_3
815 stack.append(registers[3])
816 elif opcode == 212: # setlocal_0
817 registers[0] = stack.pop()
818 elif opcode == 213: # setlocal_1
819 registers[1] = stack.pop()
820 elif opcode == 214: # setlocal_2
821 registers[2] = stack.pop()
822 elif opcode == 215: # setlocal_3
823 registers[3] = stack.pop()
825 raise NotImplementedError(
826 'Unsupported opcode %d' % opcode)
828 avm_class.method_pyfunctions[func_name] = resfunc