[swfinterp] Intepret more multinames
[youtube-dl] / youtube_dl / swfinterp.py
index d043c2f992ef0d9fa683d9667482f50f28a26fb4..7c0ee1e61a89d13d56bf0f2c54111bd738054890 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)'
@@ -85,13 +85,21 @@ class _AVMClass(object):
             for name, idx in methods.items()))
 
 
+class _Multiname(object):
+    def __init__(self, kind):
+        self.kind = kind
+
+    def __repr__(self):
+        return '[MULTINAME kind: 0x%x]' % self.kind
+
+
 def _read_int(reader):
     res = 0
     shift = 0
     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
@@ -103,7 +111,7 @@ def _u30(reader):
     res = _read_int(reader)
     assert res & 0xf0000000 == 0
     return res
-u32 = _read_int
+_u32 = _read_int
 
 
 def _s32(reader):
@@ -117,7 +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):
@@ -136,12 +144,13 @@ 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
 
 
 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)
@@ -204,8 +213,12 @@ 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: %d]' % kind)
+                self.multinames.append(_Multiname(kind))
                 for _c2 in range(MULTINAME_SIZES[kind]):
                     u30()
 
@@ -346,6 +359,9 @@ 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]
@@ -353,7 +369,9 @@ 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:
@@ -369,14 +387,12 @@ 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
                     offset = s24()
                     value = stack.pop()
@@ -399,6 +415,13 @@ class SWFInterpreter(object):
                 elif opcode == 48:  # pushscope
                     new_scope = stack.pop()
                     scopes.append(new_scope)
+                elif opcode == 66:  # construct
+                    arg_count = u30()
+                    args = list(reversed(
+                        [stack.pop() for _ in range(arg_count)]))
+                    obj = stack.pop()
+                    res = obj.avm_class.make_object()
+                    stack.append(res)
                 elif opcode == 70:  # callproperty
                     index = u30()
                     mname = self.multinames[index]
@@ -458,8 +481,7 @@ 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())
@@ -521,6 +543,8 @@ class SWFInterpreter(object):
                     index = u30()
                     value = stack.pop()
                     idx = self.multinames[index]
+                    if isinstance(idx, _Multiname):
+                        idx = stack.pop()
                     obj = stack.pop()
                     obj[idx] = value
                 elif opcode == 98:  # getlocal
@@ -537,6 +561,11 @@ class SWFInterpreter(object):
                         obj = stack.pop()
                         assert isinstance(obj, list)
                         stack.append(len(obj))
+                    elif isinstance(pname, compat_str):  # Member access
+                        obj = stack.pop()
+                        assert isinstance(obj, (dict, _ScopeDict)), \
+                            'Accessing member on %r' % obj
+                        stack.append(obj[pname])
                     else:  # Assume attribute access
                         idx = stack.pop()
                         assert isinstance(idx, int)