[ChangeLog] Actualize
[youtube-dl] / youtube_dl / swfinterp.py
index 5d17fb1ac39c6546b98e6a5121e306cf434f2ef6..0c71585753134e93fba8d8de5cee003d31f050c9 100644 (file)
@@ -4,10 +4,12 @@ import collections
 import io
 import zlib
 
-from .utils import (
+from .compat import (
     compat_str,
+    compat_struct_unpack,
+)
+from .utils import (
     ExtractorError,
-    struct_unpack,
 )
 
 
@@ -23,17 +25,17 @@ def _extract_tags(file_contents):
             file_contents[:1])
 
     # Determine number of bits in framesize rectangle
-    framesize_nbits = struct_unpack('!B', content[:1])[0] >> 3
+    framesize_nbits = compat_struct_unpack('!B', content[:1])[0] >> 3
     framesize_len = (5 + 4 * framesize_nbits + 7) // 8
 
     pos = framesize_len + 2 + 2
     while pos < len(content):
-        header16 = struct_unpack('<H', content[pos:pos + 2])[0]
+        header16 = compat_struct_unpack('<H', content[pos:pos + 2])[0]
         pos += 2
         tag_code = header16 >> 6
         tag_len = header16 & 0x3f
         if tag_len == 0x3f:
-            tag_len = struct_unpack('<I', content[pos:pos + 4])[0]
+            tag_len = compat_struct_unpack('<I', content[pos:pos + 4])[0]
             pos += 4
         assert pos + tag_len <= len(content), \
             ('Tag %d ends at %d+%d - that\'s longer than the file (%d)'
@@ -62,13 +64,14 @@ class _ScopeDict(dict):
 
 
 class _AVMClass(object):
-    def __init__(self, name_idx, name):
+    def __init__(self, name_idx, name, static_properties=None):
         self.name_idx = name_idx
         self.name = name
         self.method_names = {}
         self.method_idxs = {}
         self.methods = {}
         self.method_pyfunctions = {}
+        self.static_properties = static_properties if static_properties else {}
 
         self.variables = _ScopeDict(self)
         self.constants = {}
@@ -100,7 +103,7 @@ def _read_int(reader):
     for _ in range(5):
         buf = reader.read(1)
         assert len(buf) == 1
-        b = struct_unpack('<B', buf)[0]
+        b = compat_struct_unpack('<B', buf)[0]
         res = res | ((b & 0x7f) << shift)
         if b & 0x80 == 0:
             break
@@ -112,6 +115,8 @@ def _u30(reader):
     res = _read_int(reader)
     assert res & 0xf0000000 == 0
     return res
+
+
 _u32 = _read_int
 
 
@@ -126,7 +131,7 @@ def _s24(reader):
     bs = reader.read(3)
     assert len(bs) == 3
     last_byte = b'\xff' if (ord(bs[2:3]) >= 0x80) else b'\x00'
-    return struct_unpack('<i', bs + last_byte)[0]
+    return compat_struct_unpack('<i', bs + last_byte)[0]
 
 
 def _read_string(reader):
@@ -145,31 +150,43 @@ def _read_bytes(count, reader):
 
 def _read_byte(reader):
     resb = _read_bytes(1, reader=reader)
-    res = struct_unpack('<B', resb)[0]
+    res = compat_struct_unpack('<B', resb)[0]
     return res
 
 
 StringClass = _AVMClass('(no name idx)', 'String')
 ByteArrayClass = _AVMClass('(no name idx)', 'ByteArray')
+TimerClass = _AVMClass('(no name idx)', 'Timer')
+TimerEventClass = _AVMClass('(no name idx)', 'TimerEvent', {'TIMER': 'timer'})
 _builtin_classes = {
     StringClass.name: StringClass,
     ByteArrayClass.name: ByteArrayClass,
+    TimerClass.name: TimerClass,
+    TimerEventClass.name: TimerEventClass,
 }
 
 
 class _Undefined(object):
-    def __boolean__(self):
+    def __bool__(self):
         return False
+    __nonzero__ = __bool__
 
     def __hash__(self):
         return 0
 
+    def __str__(self):
+        return 'undefined'
+    __repr__ = __str__
+
+
 undefined = _Undefined()
 
 
 class SWFInterpreter(object):
     def __init__(self, file_contents):
-        self._patched_functions = {}
+        self._patched_functions = {
+            (TimerClass, 'addEventListener'): lambda params: undefined,
+        }
         code_tag = next(tag
                         for tag_code, tag in _extract_tags(file_contents)
                         if tag_code == 82)
@@ -346,15 +363,16 @@ class SWFInterpreter(object):
             u30()  # iinit
             trait_count = u30()
             for _c2 in range(trait_count):
-                trait_methods, constants = parse_traits_info()
+                trait_methods, trait_constants = parse_traits_info()
                 avm_class.register_methods(trait_methods)
-                assert constants is None
+                if trait_constants:
+                    avm_class.constants.update(trait_constants)
 
         assert len(classes) == class_count
         self._classes_by_name = dict((c.name, c) for c in classes)
 
         for avm_class in classes:
-            u30()  # cinit
+            avm_class.cinit_idx = u30()
             trait_count = u30()
             for _c2 in range(trait_count):
                 trait_methods, trait_constants = parse_traits_info()
@@ -373,6 +391,7 @@ class SWFInterpreter(object):
         # Method bodies
         method_body_count = u30()
         Method = collections.namedtuple('Method', ['code', 'local_count'])
+        self._all_methods = []
         for _c in range(method_body_count):
             method_idx = u30()
             u30()  # max_stack
@@ -381,9 +400,10 @@ class SWFInterpreter(object):
             u30()  # max_scope_depth
             code_length = u30()
             code = read_bytes(code_length)
+            m = Method(code, local_count)
+            self._all_methods.append(m)
             for avm_class in classes:
                 if method_idx in avm_class.method_idxs:
-                    m = Method(code, local_count)
                     avm_class.methods[avm_class.method_idxs[method_idx]] = m
             exception_count = u30()
             for _c2 in range(exception_count):
@@ -401,12 +421,20 @@ class SWFInterpreter(object):
     def patch_function(self, avm_class, func_name, f):
         self._patched_functions[(avm_class, func_name)] = f
 
-    def extract_class(self, class_name):
+    def extract_class(self, class_name, call_cinit=True):
         try:
-            return self._classes_by_name[class_name]
+            res = self._classes_by_name[class_name]
         except KeyError:
             raise ExtractorError('Class %r not found' % class_name)
 
+        if call_cinit and hasattr(res, 'cinit_idx'):
+            res.register_methods({'$cinit': res.cinit_idx})
+            res.methods['$cinit'] = self._all_methods[res.cinit_idx]
+            cinit = self.extract_function(res, '$cinit')
+            cinit([])
+
+        return res
+
     def extract_function(self, avm_class, func_name):
         p = self._patched_functions.get((avm_class, func_name))
         if p:
@@ -429,7 +457,7 @@ class SWFInterpreter(object):
             registers = [avm_class.variables] + list(args) + [None] * m.local_count
             stack = []
             scopes = collections.deque([
-                self._classes_by_name, avm_class.variables])
+                self._classes_by_name, avm_class.constants, avm_class.variables])
             while True:
                 opcode = _read_byte(coder)
                 if opcode == 9:  # label
@@ -577,6 +605,12 @@ class SWFInterpreter(object):
                 elif opcode == 72:  # returnvalue
                     res = stack.pop()
                     return res
+                elif opcode == 73:  # constructsuper
+                    # Not yet implemented, just hope it works without it
+                    arg_count = u30()
+                    args = list(reversed(
+                        [stack.pop() for _ in range(arg_count)]))
+                    obj = stack.pop()
                 elif opcode == 74:  # constructproperty
                     index = u30()
                     arg_count = u30()
@@ -657,8 +691,12 @@ class SWFInterpreter(object):
 
                     if mname in scope:
                         res = scope[mname]
+                    elif mname in _builtin_classes:
+                        res = _builtin_classes[mname]
                     else:
-                        res = avm_class.constants[mname]
+                        # Assume uninitialized
+                        # TODO warn here
+                        res = undefined
                     stack.append(res)
                 elif opcode == 97:  # setproperty
                     index = u30()
@@ -684,7 +722,12 @@ class SWFInterpreter(object):
                         stack.append(len(obj))
                     elif isinstance(pname, compat_str):  # Member access
                         obj = stack.pop()
-                        assert isinstance(obj, (dict, _ScopeDict)), \
+                        if isinstance(obj, _AVMClass):
+                            res = obj.static_properties[pname]
+                            stack.append(res)
+                            continue
+
+                        assert isinstance(obj, (dict, _ScopeDict)),\
                             'Accessing member %r on %r' % (pname, obj)
                         res = obj.get(pname, undefined)
                         stack.append(res)
@@ -694,6 +737,14 @@ class SWFInterpreter(object):
                         obj = stack.pop()
                         assert isinstance(obj, list)
                         stack.append(obj[idx])
+                elif opcode == 104:  # initproperty
+                    index = u30()
+                    value = stack.pop()
+                    idx = self.multinames[index]
+                    if isinstance(idx, _Multiname):
+                        idx = stack.pop()
+                    obj = stack.pop()
+                    obj[idx] = value
                 elif opcode == 115:  # convert_
                     value = stack.pop()
                     intvalue = int(value)
@@ -781,4 +832,3 @@ class SWFInterpreter(object):
 
         avm_class.method_pyfunctions[func_name] = resfunc
         return resfunc
-