[swfinterp] Add support for void methods
[youtube-dl] / youtube_dl / swfinterp.py
index 812ee7e8c54f97b2c6ef1e033f0975f22e5a6b6e..f4ee022f4e3fc3052f6ca9fbb9e19c5653b08a93 100644 (file)
@@ -2,12 +2,12 @@ from __future__ import unicode_literals
 
 import collections
 import io
-import struct
 import zlib
 
 from .utils import (
     compat_str,
     ExtractorError,
+    struct_unpack,
 )
 
 
@@ -23,17 +23,17 @@ 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 = 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 = 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 = 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)'
@@ -99,7 +99,7 @@ def _read_int(reader):
     for _ in range(5):
         buf = reader.read(1)
         assert len(buf) == 1
-        b = struct.unpack('<B', buf)[0]
+        b = struct_unpack('<B', buf)[0]
         res = res | ((b & 0x7f) << shift)
         if b & 0x80 == 0:
             break
@@ -111,7 +111,7 @@ def _u30(reader):
     res = _read_int(reader)
     assert res & 0xf0000000 == 0
     return res
-u32 = _read_int
+_u32 = _read_int
 
 
 def _s32(reader):
@@ -125,7 +125,7 @@ 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 struct_unpack('<i', bs + last_byte)[0]
 
 
 def _read_string(reader):
@@ -144,12 +144,16 @@ def _read_bytes(count, reader):
 
 def _read_byte(reader):
     resb = _read_bytes(1, reader=reader)
-    res = struct.unpack('<B', resb)[0]
+    res = struct_unpack('<B', resb)[0]
     return res
 
 
+StringClass = _AVMClass('(no name idx)', 'String')
+
+
 class SWFInterpreter(object):
     def __init__(self, file_contents):
+        self._patched_functions = {}
         code_tag = next(tag
                         for tag_code, tag in _extract_tags(file_contents)
                         if tag_code == 82)
@@ -212,6 +216,10 @@ class SWFInterpreter(object):
                 u30()  # namespace_idx
                 name_idx = u30()
                 self.multinames.append(self.constant_strings[name_idx])
+            elif kind == 0x09:
+                name_idx = u30()
+                u30()
+                self.multinames.append(self.constant_strings[name_idx])
             else:
                 self.multinames.append(_Multiname(kind))
                 for _c2 in range(MULTINAME_SIZES[kind]):
@@ -354,6 +362,9 @@ class SWFInterpreter(object):
 
         assert p + code_reader.tell() == len(code_tag)
 
+    def patch_function(self, avm_class, func_name, f):
+        self._patched_functions[(avm_class, func_name)] = f
+
     def extract_class(self, class_name):
         try:
             return self._classes_by_name[class_name]
@@ -361,7 +372,9 @@ class SWFInterpreter(object):
             raise ExtractorError('Class %r not found' % class_name)
 
     def extract_function(self, avm_class, func_name):
-        print('Extracting %s.%s' % (avm_class.name, func_name))
+        p = self._patched_functions.get((avm_class, func_name))
+        if p:
+            return p
         if func_name in avm_class.method_pyfunctions:
             return avm_class.method_pyfunctions[func_name]
         if func_name in self._classes_by_name:
@@ -377,15 +390,16 @@ class SWFInterpreter(object):
             s24 = lambda: _s24(coder)
             u30 = lambda: _u30(coder)
 
-            print('Invoking %s.%s(%r)' % (avm_class.name, func_name, tuple(args)))
             registers = [avm_class.variables] + list(args) + [None] * m.local_count
             stack = []
             scopes = collections.deque([
                 self._classes_by_name, avm_class.variables])
             while True:
                 opcode = _read_byte(coder)
-                print('opcode: %r, stack(%d): %r' % (opcode, len(stack), stack))
-                if opcode == 17:  # iftrue
+                if opcode == 16:  # jump
+                    offset = s24()
+                    coder.seek(coder.tell() + offset)
+                elif opcode == 17:  # iftrue
                     offset = s24()
                     value = stack.pop()
                     if value:
@@ -395,9 +409,27 @@ class SWFInterpreter(object):
                     value = stack.pop()
                     if not value:
                         coder.seek(coder.tell() + offset)
+                elif opcode == 19:  # ifeq
+                    offset = s24()
+                    value2 = stack.pop()
+                    value1 = stack.pop()
+                    if value2 == value1:
+                        coder.seek(coder.tell() + offset)
+                elif opcode == 20:  # ifne
+                    offset = s24()
+                    value2 = stack.pop()
+                    value1 = stack.pop()
+                    if value2 != value1:
+                        coder.seek(coder.tell() + offset)
+                elif opcode == 32:  # pushnull
+                    stack.append(None)
                 elif opcode == 36:  # pushbyte
                     v = _read_byte(coder)
                     stack.append(v)
+                elif opcode == 38:  # pushtrue
+                    stack.append(True)
+                elif opcode == 39:  # pushfalse
+                    stack.append(False)
                 elif opcode == 42:  # dup
                     value = stack[-1]
                     stack.append(value)
@@ -458,9 +490,23 @@ class SWFInterpreter(object):
                             res = args[0].join(obj)
                             stack.append(res)
                             continue
+                    elif obj == StringClass:
+                        if mname == 'String':
+                            assert len(args) == 1
+                            assert isinstance(args[0], (int, compat_str))
+                            res = compat_str(args[0])
+                            stack.append(res)
+                            continue
+                        else:
+                            raise NotImplementedError(
+                                'Function String.%s is not yet implemented'
+                                % mname)
                     raise NotImplementedError(
                         'Unsupported property %r on %r'
                         % (mname, obj))
+                elif opcode == 71:  # returnvoid
+                    res = None
+                    return res
                 elif opcode == 72:  # returnvalue
                     res = stack.pop()
                     return res
@@ -473,8 +519,7 @@ class SWFInterpreter(object):
 
                     mname = self.multinames[index]
                     assert isinstance(obj, _AVMClass)
-                    construct_method = self.extract_function(
-                        obj, mname)
+
                     # We do not actually call the constructor for now;
                     # we just pretend it does nothing
                     stack.append(obj.make_object())
@@ -485,6 +530,17 @@ class SWFInterpreter(object):
                     args = list(reversed(
                         [stack.pop() for _ in range(arg_count)]))
                     obj = stack.pop()
+                    if isinstance(obj, _AVMClass_Object):
+                        func = self.extract_function(obj.avm_class, mname)
+                        res = func(args)
+                        assert res is None
+                        continue
+                    if isinstance(obj, _ScopeDict):
+                        assert mname in obj.avm_class.method_names
+                        func = self.extract_function(obj.avm_class, mname)
+                        res = func(args)
+                        assert res is None
+                        continue
                     if mname == 'reverse':
                         assert isinstance(obj, list)
                         obj.reverse()
@@ -508,7 +564,10 @@ class SWFInterpreter(object):
                             break
                     else:
                         res = scopes[0]
-                    stack.append(res[mname])
+                    if mname not in res and mname == 'String':
+                        stack.append(StringClass)
+                    else:
+                        stack.append(res[mname])
                 elif opcode == 94:  # findproperty
                     index = u30()
                     mname = self.multinames[index]
@@ -539,7 +598,6 @@ class SWFInterpreter(object):
                     if isinstance(idx, _Multiname):
                         idx = stack.pop()
                     obj = stack.pop()
-                    print('Setting %r.%r = %r' % (obj, idx, value))
                     obj[idx] = value
                 elif opcode == 98:  # getlocal
                     index = u30()
@@ -553,8 +611,14 @@ class SWFInterpreter(object):
                     pname = self.multinames[index]
                     if pname == 'length':
                         obj = stack.pop()
-                        assert isinstance(obj, list)
+                        assert isinstance(obj, (compat_str, list))
                         stack.append(len(obj))
+                    elif isinstance(pname, compat_str):  # Member access
+                        obj = stack.pop()
+                        assert isinstance(obj, (dict, _ScopeDict)), \
+                            'Accessing member %r on %r' % (pname, obj)
+                        res = obj.get(pname, None)
+                        stack.append(res)
                     else:  # Assume attribute access
                         idx = stack.pop()
                         assert isinstance(idx, int)
@@ -584,6 +648,11 @@ class SWFInterpreter(object):
                     value1 = stack.pop()
                     res = value1 % value2
                     stack.append(res)
+                elif opcode == 171:  # equals
+                    value2 = stack.pop()
+                    value1 = stack.pop()
+                    result = value1 == value2
+                    stack.append(result)
                 elif opcode == 175:  # greaterequals
                     value2 = stack.pop()
                     value1 = stack.pop()