From 5425626790a46f9b5bdecf4e33bb254c4c2423ea Mon Sep 17 00:00:00 2001
From: Philipp Hagemeister <phihag@phihag.de>
Date: Fri, 18 Jul 2014 10:24:28 +0200
Subject: [youtube] Move swfinterp into its own file

---
 youtube_dl/extractor/youtube.py | 454 +---------------------------------------
 1 file changed, 4 insertions(+), 450 deletions(-)

(limited to 'youtube_dl/extractor')

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 16f4a047d..623056bd9 100644
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -14,6 +14,7 @@ import zlib
 from .common import InfoExtractor, SearchInfoExtractor
 from .subtitles import SubtitlesInfoExtractor
 from ..jsinterp import JSInterpreter
+from ..swfinterp import SWFInterpreter
 from ..utils import (
     compat_chr,
     compat_parse_qs,
@@ -450,457 +451,10 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
         return lambda s: initial_function([s])
 
     def _parse_sig_swf(self, file_contents):
-        if file_contents[1:3] != b'WS':
-            raise ExtractorError(
-                u'Not an SWF file; header is %r' % file_contents[:3])
-        if file_contents[:1] == b'C':
-            content = zlib.decompress(file_contents[8:])
-        else:
-            raise NotImplementedError(u'Unsupported compression format %r' %
-                                      file_contents[:1])
-
-        def extract_tags(content):
-            pos = 0
-            while pos < len(content):
-                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]
-                    pos += 4
-                assert pos+tag_len <= len(content)
-                yield (tag_code, content[pos:pos+tag_len])
-                pos += tag_len
-
-        code_tag = next(tag
-                        for tag_code, tag in extract_tags(content)
-                        if tag_code == 82)
-        p = code_tag.index(b'\0', 4) + 1
-        code_reader = io.BytesIO(code_tag[p:])
-
-        # Parse ABC (AVM2 ByteCode)
-        def read_int(reader=None):
-            if reader is None:
-                reader = code_reader
-            res = 0
-            shift = 0
-            for _ in range(5):
-                buf = reader.read(1)
-                assert len(buf) == 1
-                b = struct.unpack('<B', buf)[0]
-                res = res | ((b & 0x7f) << shift)
-                if b & 0x80 == 0:
-                    break
-                shift += 7
-            return res
-
-        def u30(reader=None):
-            res = read_int(reader)
-            assert res & 0xf0000000 == 0
-            return res
-        u32 = read_int
-
-        def s32(reader=None):
-            v = read_int(reader)
-            if v & 0x80000000 != 0:
-                v = - ((v ^ 0xffffffff) + 1)
-            return v
-
-        def s24(reader):
-            bs = reader.read(3)
-            assert len(bs) == 3
-            first_byte = b'\xff' if (ord(bs[0:1]) >= 0x80) else b'\x00'
-            return struct.unpack('!i', first_byte + bs)
-
-        def read_string(reader=None):
-            if reader is None:
-                reader = code_reader
-            slen = u30(reader)
-            resb = reader.read(slen)
-            assert len(resb) == slen
-            return resb.decode('utf-8')
-
-        def read_bytes(count, reader=None):
-            if reader is None:
-                reader = code_reader
-            resb = reader.read(count)
-            assert len(resb) == count
-            return resb
-
-        def read_byte(reader=None):
-            resb = read_bytes(1, reader=reader)
-            res = struct.unpack('<B', resb)[0]
-            return res
-
-        # minor_version + major_version
-        read_bytes(2 + 2)
-
-        # Constant pool
-        int_count = u30()
-        for _c in range(1, int_count):
-            s32()
-        uint_count = u30()
-        for _c in range(1, uint_count):
-            u32()
-        double_count = u30()
-        read_bytes((double_count-1) * 8)
-        string_count = u30()
-        constant_strings = [u'']
-        for _c in range(1, string_count):
-            s = read_string()
-            constant_strings.append(s)
-        namespace_count = u30()
-        for _c in range(1, namespace_count):
-            read_bytes(1)  # kind
-            u30()  # name
-        ns_set_count = u30()
-        for _c in range(1, ns_set_count):
-            count = u30()
-            for _c2 in range(count):
-                u30()
-        multiname_count = u30()
-        MULTINAME_SIZES = {
-            0x07: 2,  # QName
-            0x0d: 2,  # QNameA
-            0x0f: 1,  # RTQName
-            0x10: 1,  # RTQNameA
-            0x11: 0,  # RTQNameL
-            0x12: 0,  # RTQNameLA
-            0x09: 2,  # Multiname
-            0x0e: 2,  # MultinameA
-            0x1b: 1,  # MultinameL
-            0x1c: 1,  # MultinameLA
-        }
-        multinames = [u'']
-        for _c in range(1, multiname_count):
-            kind = u30()
-            assert kind in MULTINAME_SIZES, u'Invalid multiname kind %r' % kind
-            if kind == 0x07:
-                u30()  # namespace_idx
-                name_idx = u30()
-                multinames.append(constant_strings[name_idx])
-            else:
-                multinames.append('[MULTINAME kind: %d]' % kind)
-                for _c2 in range(MULTINAME_SIZES[kind]):
-                    u30()
-
-        # Methods
-        method_count = u30()
-        MethodInfo = collections.namedtuple(
-            'MethodInfo',
-            ['NEED_ARGUMENTS', 'NEED_REST'])
-        method_infos = []
-        for method_id in range(method_count):
-            param_count = u30()
-            u30()  # return type
-            for _ in range(param_count):
-                u30()  # param type
-            u30()  # name index (always 0 for youtube)
-            flags = read_byte()
-            if flags & 0x08 != 0:
-                # Options present
-                option_count = u30()
-                for c in range(option_count):
-                    u30()  # val
-                    read_bytes(1)  # kind
-            if flags & 0x80 != 0:
-                # Param names present
-                for _ in range(param_count):
-                    u30()  # param name
-            mi = MethodInfo(flags & 0x01 != 0, flags & 0x04 != 0)
-            method_infos.append(mi)
-
-        # Metadata
-        metadata_count = u30()
-        for _c in range(metadata_count):
-            u30()  # name
-            item_count = u30()
-            for _c2 in range(item_count):
-                u30()  # key
-                u30()  # value
-
-        def parse_traits_info():
-            trait_name_idx = u30()
-            kind_full = read_byte()
-            kind = kind_full & 0x0f
-            attrs = kind_full >> 4
-            methods = {}
-            if kind in [0x00, 0x06]:  # Slot or Const
-                u30()  # Slot id
-                u30()  # type_name_idx
-                vindex = u30()
-                if vindex != 0:
-                    read_byte()  # vkind
-            elif kind in [0x01, 0x02, 0x03]:  # Method / Getter / Setter
-                u30()  # disp_id
-                method_idx = u30()
-                methods[multinames[trait_name_idx]] = method_idx
-            elif kind == 0x04:  # Class
-                u30()  # slot_id
-                u30()  # classi
-            elif kind == 0x05:  # Function
-                u30()  # slot_id
-                function_idx = u30()
-                methods[function_idx] = multinames[trait_name_idx]
-            else:
-                raise ExtractorError(u'Unsupported trait kind %d' % kind)
-
-            if attrs & 0x4 != 0:  # Metadata present
-                metadata_count = u30()
-                for _c3 in range(metadata_count):
-                    u30()  # metadata index
-
-            return methods
-
-        class AVMClass(object):
-            def __init__(self, name_idx):
-                self.name_idx = name_idx
-                self.method_names = {}
-                self.method_idxs = {}
-                self.methods = {}
-                self.method_pyfunctions = {}
-                self.variables = {}
-
-            @property
-            def name(self):
-                return multinames[self.name_idx]
-
-        # Classes
-        class_count = u30()
-        classes = []
-        for class_id in range(class_count):
-            name_idx = u30()
-            classes.append(AVMClass(name_idx))
-            u30()  # super_name idx
-            flags = read_byte()
-            if flags & 0x08 != 0:  # Protected namespace is present
-                u30()  # protected_ns_idx
-            intrf_count = u30()
-            for _c2 in range(intrf_count):
-                u30()
-            u30()  # iinit
-            trait_count = u30()
-            for _c2 in range(trait_count):
-                parse_traits_info()
-        assert len(classes) == class_count
-
+        swfi = SWFInterpreter(file_contents)
         TARGET_CLASSNAME = u'SignatureDecipher'
-        searched_class = next(
-            c for c in classes if c.name == TARGET_CLASSNAME)
-        if searched_class is None:
-            raise ExtractorError(u'Target class %r not found' %
-                                 TARGET_CLASSNAME)
-
-        for avm_class in classes:
-            u30()  # cinit
-            trait_count = u30()
-            for _c2 in range(trait_count):
-                trait_methods = parse_traits_info()
-                avm_class.method_names.update(trait_methods.items())
-                avm_class.method_idxs.update(dict(
-                    (idx, name)
-                    for name, idx in trait_methods.items()))
-
-        # Scripts
-        script_count = u30()
-        for _c in range(script_count):
-            u30()  # init
-            trait_count = u30()
-            for _c2 in range(trait_count):
-                parse_traits_info()
-
-        # Method bodies
-        method_body_count = u30()
-        Method = collections.namedtuple('Method', ['code', 'local_count'])
-        for _c in range(method_body_count):
-            method_idx = u30()
-            u30()  # max_stack
-            local_count = u30()
-            u30()  # init_scope_depth
-            u30()  # max_scope_depth
-            code_length = u30()
-            code = read_bytes(code_length)
-            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):
-                u30()  # from
-                u30()  # to
-                u30()  # target
-                u30()  # exc_type
-                u30()  # var_name
-            trait_count = u30()
-            for _c2 in range(trait_count):
-                parse_traits_info()
-
-        assert p + code_reader.tell() == len(code_tag)
-
-        def extract_function(avm_class, func_name):
-            if func_name in avm_class.method_pyfunctions:
-                return avm_class.method_pyfunctions[func_name]
-            if func_name not in avm_class.methods:
-                raise ExtractorError(u'Cannot find function %r' % func_name)
-            m = avm_class.methods[func_name]
-
-            def resfunc(args):
-                registers = ['(this)'] + list(args) + [None] * m.local_count
-                stack = []
-                coder = io.BytesIO(m.code)
-                while True:
-                    opcode = struct.unpack('!B', coder.read(1))[0]
-                    if opcode == 17:  # iftrue
-                        offset = s24(coder)
-                        value = stack.pop()
-                        if value:
-                            coder.seek(coder.tell() + offset)
-                    elif opcode == 36:  # pushbyte
-                        v = struct.unpack('!B', coder.read(1))[0]
-                        stack.append(v)
-                    elif opcode == 44:  # pushstring
-                        idx = u30(coder)
-                        stack.append(constant_strings[idx])
-                    elif opcode == 48:  # pushscope
-                        # We don't implement the scope register, so we'll just
-                        # ignore the popped value
-                        stack.pop()
-                    elif opcode == 70:  # callproperty
-                        index = u30(coder)
-                        mname = multinames[index]
-                        arg_count = u30(coder)
-                        args = list(reversed(
-                            [stack.pop() for _ in range(arg_count)]))
-                        obj = stack.pop()
-                        if mname == u'split':
-                            assert len(args) == 1
-                            assert isinstance(args[0], compat_str)
-                            assert isinstance(obj, compat_str)
-                            if args[0] == u'':
-                                res = list(obj)
-                            else:
-                                res = obj.split(args[0])
-                            stack.append(res)
-                        elif mname == u'slice':
-                            assert len(args) == 1
-                            assert isinstance(args[0], int)
-                            assert isinstance(obj, list)
-                            res = obj[args[0]:]
-                            stack.append(res)
-                        elif mname == u'join':
-                            assert len(args) == 1
-                            assert isinstance(args[0], compat_str)
-                            assert isinstance(obj, list)
-                            res = args[0].join(obj)
-                            stack.append(res)
-                        elif mname in avm_class.method_pyfunctions:
-                            stack.append(avm_class.method_pyfunctions[mname](args))
-                        else:
-                            raise NotImplementedError(
-                                u'Unsupported property %r on %r'
-                                % (mname, obj))
-                    elif opcode == 72:  # returnvalue
-                        res = stack.pop()
-                        return res
-                    elif opcode == 79:  # callpropvoid
-                        index = u30(coder)
-                        mname = multinames[index]
-                        arg_count = u30(coder)
-                        args = list(reversed(
-                            [stack.pop() for _ in range(arg_count)]))
-                        obj = stack.pop()
-                        if mname == u'reverse':
-                            assert isinstance(obj, list)
-                            obj.reverse()
-                        else:
-                            raise NotImplementedError(
-                                u'Unsupported (void) property %r on %r'
-                                % (mname, obj))
-                    elif opcode == 86:  # newarray
-                        arg_count = u30(coder)
-                        arr = []
-                        for i in range(arg_count):
-                            arr.append(stack.pop())
-                        arr = arr[::-1]
-                        stack.append(arr)
-                    elif opcode == 93:  # findpropstrict
-                        index = u30(coder)
-                        mname = multinames[index]
-                        res = extract_function(avm_class, mname)
-                        stack.append(res)
-                    elif opcode == 94:  # findproperty
-                        index = u30(coder)
-                        mname = multinames[index]
-                        res = avm_class.variables.get(mname)
-                        stack.append(res)
-                    elif opcode == 96:  # getlex
-                        index = u30(coder)
-                        mname = multinames[index]
-                        res = avm_class.variables.get(mname)
-                        stack.append(res)
-                    elif opcode == 97:  # setproperty
-                        index = u30(coder)
-                        value = stack.pop()
-                        idx = stack.pop()
-                        obj = stack.pop()
-                        assert isinstance(obj, list)
-                        assert isinstance(idx, int)
-                        obj[idx] = value
-                    elif opcode == 98:  # getlocal
-                        index = u30(coder)
-                        stack.append(registers[index])
-                    elif opcode == 99:  # setlocal
-                        index = u30(coder)
-                        value = stack.pop()
-                        registers[index] = value
-                    elif opcode == 102:  # getproperty
-                        index = u30(coder)
-                        pname = multinames[index]
-                        if pname == u'length':
-                            obj = stack.pop()
-                            assert isinstance(obj, list)
-                            stack.append(len(obj))
-                        else:  # Assume attribute access
-                            idx = stack.pop()
-                            assert isinstance(idx, int)
-                            obj = stack.pop()
-                            assert isinstance(obj, list)
-                            stack.append(obj[idx])
-                    elif opcode == 128:  # coerce
-                        u30(coder)
-                    elif opcode == 133:  # coerce_s
-                        assert isinstance(stack[-1], (type(None), compat_str))
-                    elif opcode == 164:  # modulo
-                        value2 = stack.pop()
-                        value1 = stack.pop()
-                        res = value1 % value2
-                        stack.append(res)
-                    elif opcode == 175:  # greaterequals
-                        value2 = stack.pop()
-                        value1 = stack.pop()
-                        result = value1 >= value2
-                        stack.append(result)
-                    elif opcode == 208:  # getlocal_0
-                        stack.append(registers[0])
-                    elif opcode == 209:  # getlocal_1
-                        stack.append(registers[1])
-                    elif opcode == 210:  # getlocal_2
-                        stack.append(registers[2])
-                    elif opcode == 211:  # getlocal_3
-                        stack.append(registers[3])
-                    elif opcode == 214:  # setlocal_2
-                        registers[2] = stack.pop()
-                    elif opcode == 215:  # setlocal_3
-                        registers[3] = stack.pop()
-                    else:
-                        raise NotImplementedError(
-                            u'Unsupported opcode %d' % opcode)
-
-            avm_class.method_pyfunctions[func_name] = resfunc
-            return resfunc
-
-        initial_function = extract_function(searched_class, u'decipher')
+        searched_class = swfi.extract_class(TARGET_CLASSNAME)
+        initial_function = swfi.extract_function(searched_class, u'decipher')
         return lambda s: initial_function([s])
 
     def _decrypt_signature(self, s, video_id, player_url, age_gate=False):
-- 
cgit v1.2.3