Ticket #138: ole2.3.diff

msoffice.py

@@ -1 @@
-
-
-
-
-
+
+
-
-
-
-
-
+
+
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-    }
-
+
+
+
+
+
+
+
+
+
+
+
-
-
-
-
+
+
-
-
-
+
+
+
+
+
+
+
+
-
+        #-- OLE 2.01 ---
+
+        # Program ID
+        yield PascalString32(self, "prog_id", strip="\0")
+
+        if self["os"].value != self.OS_MAC:
+            # Magic number
+            yield textHandler(UInt32(self, "magic", "Magic number (0x71B239F4)"), hexadecimal)
+
+            # Unicode version
+            yield PascalStringWin32(self, "user_type_unicode", strip="\0")
+            yield PascalStringWin32(self, "clipboard_format_unicode", strip="\0")
+            yield PascalStringWin32(self, "prog_id_unicode", strip="\0")
+
+        size = self.datasize - (self._current_size // 8) # _current_size because current_size returns _current_max_size
+        if size:
+            yield NullBytes(self, "end_padding", size)
+
+        if self.datasize<self.size//8: yield RawBytes(self,"slack_space",(self.size//8)-self.datasize)
+
+class RootEntry(OLE2FragmentParser):
+    ENDIAN_CHECK=False
+    OS_CHECK=False
+
-    def createFields(self):
-        for index, property in enumerate(self.ole2.properties):
-            if index == 0:
-                continue
-            try:
-
+,parser
-            except LookupError:
-                name = property.name+"content"
-
+
+
-                yield field
+    def seekSBlock(self, block):
+        self.seekBit(block * self.ole2.ss_size)
-
-_index, property, name_prefix
+, name_prefix, parser=RawParser
-        ole2 = self.ole2
-        if not property["size"].value:
-            return
@@ -43 +93 @@
-        first = None
-        previous = None
-        size = 0
-        start = property["start"].value
-        chain = ole2.getChain(start, ole2.ss_fat)
-        blocksize = ole2.ss_size
-        seek = ole2.seekSBlock
-        desc_format = "Small blocks %s..%s (%s)"
-        fragment_group = None
+        chain = ole2.getChain(property["start"].value, ole2.ss_fat)
-        while True:
-            try:
-                block = chain.next()
-i
-not first
+u
+first is None
-                    first = block
-i
-
-i
-i
+u
+is not None 
+u
+u
-                    previous = block
-block
+ole2.ss_
-                    continue
-            except StopIteration:
-                block = None
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
-            if block is None:
-                break
-            first = block
-            previous = block
-            size = ole2.sector_size
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-    def createFields(self):
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+

msoffice_summary.py

@@ -1 @@
-from hachoir_parser import HachoirParser
-from hachoir_core.field import (FieldSet, ParserError,
-RootSeekableFieldSet, 
+
-    Bit, Bits, NullBits,
-    UInt16, UInt32, TimestampWin64, Enum,
-    Bytes, RawBytes, NullBytes, String,
-8, Int16, Int
+
-from hachoir_core.text_handler import textHandler, hexadecimal, filesizeHandler
-from hachoir_core.tools import createDict
-
-
+
-from hachoir_parser.image.bmp import BmpHeader, parseImageData
+from hachoir_parser.misc.ole2_util import OLE2FragmentParser, Bool
-
-MAX_SECTION_COUNT = 100
-
-class OSConfig:
-    def __init__(self, big_endian):
-        if big_endian:
-            self.charset = "MacRoman"
-            self.utf16 = "UTF-16-BE"
-        else:
-            self.charset = "ISO-8859-1"
-            self.utf16 = "UTF-16-LE"
-
-class PropertyIndex(FieldSet):
-    COMMON_PROPERTY = {
-        0: "Dictionary",
@@ -87 +77 @@
-        yield UInt32(self, "offset")
-
-    def createDescription(self):
-
+t
-
-class Bool(Int8):
-    def createValue(self):
-        value = Int8.createValue(self)
-        return (value == -1)
-
-class Thumbnail(FieldSet):
-    """
-    Thumbnail.
@@ -145 +130 @@
-            yield RawBytes(self, "data", size)
-
-class PropertyContent(FieldSet):
+    class NullHandler(FieldSet):
+        def createFields(self):
+            yield UInt32(self, "unknown[]")
+            yield PascalString32(self, "data")
+        def createValue(self):
+            return self["data"].value
+    class BlobHandler(FieldSet):
+        def createFields(self):
+            yield UInt32(self, "size")
+            yield UInt32(self, "count")
+            for i in range(self["count"].value):
+                yield PropertyContent(self, "item[]")
+                n=paddingSize(self.current_size,32)
+                if n: yield PaddingBits(self, "padding[]", n)
+    class WidePascalString32(FieldSet):
+        ''' uses number of characters instead of number of bytes '''
+        def __init__(self,parent,name,charset='ASCII'):
+            FieldSet.__init__(self,parent,name)
+            self.charset=charset
+        def createFields(self):
+            yield UInt32(self, "length", "Length of this string")
+            yield String(self, "data", self["length"].value*2, charset=self.charset)
+        def createValue(self):
+            return self["data"].value
+        def createDisplay(self):
+            return 'u'+self["data"].display
-    TYPE_LPSTR = 30
-    TYPE_INFO = {
-        0: ("EMPTY", None),
-one
-Int16", 
-Int32", 
+ullHandler
+UInt16", U
+UInt32", U
-        4: ("Float32", Float32),
-        5: ("Float64", Float64),
-        6: ("CY", None),
@@ -177 +188 @@
-        28: ("CARRAY", None),
-        29: ("USERDEFINED", None),
-        30: ("LPSTR", PascalString32),
-
+Wide
-        64: ("FILETIME", TimestampWin64),
-None
+BlobHandler
-        66: ("STREAM", None),
-        67: ("STORAGE", None),
-        68: ("STREAMED_OBJECT", None),
@@ -202 +213 @@
-        kw = {}
-        try:
-            handler = self.TYPE_INFO[tag][1]
-
-
+
+
+
+
+
+
+
-                if tag == self.TYPE_LPSTR:
-                    kw["charset"] = osconfig.charset
-                else:
@@ -211 +227 @@
-        except LookupError:
-            handler = None
-        if not handler:
-raise ParserError
+log.warning
-                % self["type"].display)
-
+
+
-            yield UInt32(self, "count")
-            for index in xrange(self["count"].value):
-                yield handler(self, "item[]", **kw)
@@ -243 +260 @@
-        yield String(self, "name", 16)
-        yield UInt32(self, "offset")
-
-
-
-
-
-
+
+
+
-
-    OS_MAC = 1
-    OS_NAME = {
-        0: "Windows 16-bit",
-        1: "Macintosh",
-        2: "Windows 32-bit",
-    }
-
-    def __init__(self, stream, **args):
-        RootSeekableFieldSet.__init__(self, None, "root", stream, None, stream.askSize(self))
-        HachoirParser.__init__(self, stream, **args)
-        if self["endian"].value == "\xFF\xFE":
-            self.endian = BIG_ENDIAN
-        elif self["endian"].value == "\xFE\xFF":
-            self.endian = LITTLE_ENDIAN
-        else:
-            raise ParserError("OLE2: Invalid endian value")
-        self.osconfig = OSConfig(self["os"].value == self.OS_MAC)
-
-    def validate(self):
-        return True
-
-
-    def createFields(self):
-