Misc Progress

* updated README to be more accurate.
* added support for askExtractMode
* preliminary support for FILETIMEs on PROPVARIANTS (like mtime)
* slightly more data/data/debugging/logging info

README.rst

@@ -1,23 +1,23 @@
 python-lib7zip
 ==============
-Python bindings for lib7zip_
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Python bindings for 7-Zip
+~~~~~~~~~~~~~~~~~~~~~~~~~
 :author: Mark Harviston <[email protected]>
 :version: 0.1
 
 pylib7zip is a direct binding to 7z.dll from the 7-zip project (7zip.org)
 
-7z.dll uses Windows COM+ calling conventions with an over-engineered slightly pathological 
-OOP API.
+7z.dll uses Windows COM+ calling conventions without registering itself with the COM server
+and has an over-engineered slightly pathological OOP API.
 
-Currently only works on Windows with Python 3.3
+Currently only works on Windows with Python 3.3.
 
 This provides roughly the same functionality as lib7zip does for C++ and SevenZipSharp does for C#
 but with a clean Pythonic API.
 
 Like lib7zip getting metadata and extracting files are the only operations supported, creating archives, or updating them in-place is not supported.
 
-This is beta software and may crash if used in an unusual way.
+This is beta software and may crash if used in an unusual (or even a usual) way.
 
 Dependencies
 ------------

conftest.py

@@ -0,0 +1,12 @@
+import logging
+#logger = logging.getLogger('lib7zip')
+logging.basicConfig(
+	level=logging.DEBUG,
+	#level=logging.INFO,
+	format='%(asctime)s - %(name)s::%(funcName)s @ %(filename)s:%(lineno)d - %(levelname)s - %(message)s',
+	datefmt='%H:%M',
+)
+logger = logging.getLogger('lib7zip.simplecom')
+logger.setLevel(logging.INFO)
+logger = logging.getLogger('lib7zip.stream')
+logger.setLevel(logging.INFO)

lib7zip/__init__.py

@@ -40,6 +40,7 @@
 HRESULT SetLargePageMode(); /* Unused */
 
 void* calloc(size_t, size_t);
+void* malloc(size_t);
 void* memset(void*, int, size_t);
 void free(void*);
 

lib7zip/extract_callback.py

@@ -2,10 +2,10 @@
 
 from .py7ziptypes import IID_ICryptoGetTextPassword, IID_IArchiveExtractCallback, \
 	IID_ISequentialOutStream, IID_ICompressProgressInfo, IID_ICryptoGetTextPassword2, \
-	OperationResult
+	OperationResult, AskMode
 
 from .wintypes import HRESULT
-from . import log, ffi, py7ziptypes
+from . import log, ffi, C, py7ziptypes
 from .simplecom import IUnknownImpl
 from .stream import FileOutStream
 
@@ -24,6 +24,14 @@ def __init__(self, password=''):
 		#self.out_file = FileOutStream(file)
 		#self.password = ffi.new('char[]', (password or '').encode('ascii'))
 		self.password = ffi.new('wchar_t[]', password or '')
+		#password = password or ''
+		'''
+		self._password = ffi.gc(C.malloc(ffi.sizeof('wchar_t') * len(password) + 1), C.free)
+		self.password = ffi.cast('wchar_t*', self._password)
+		self.password[0:len(password)] = password
+		self.password[len(password)] = '\0'
+		'''
+
 		super().__init__()
 
 	def cleanup(self):
@@ -68,20 +76,19 @@ def CryptoGetTextPassword(self, me, password):
 		assert password[0] == ffi.NULL
 		#log.debug('passowrd?=%s', ffi.string(password[0]))
 		#password = ffi.cast('wchar_t**', password)
-		#password[0] = self.password
+		password[0] = self.password
 		#password[0] = ffi.NULL
-		#log.debug('CryptoGetTextPassword returning, password=%s', ffi.string(password[0]))
+		log.debug('CryptoGetTextPassword returning, password=%s', ffi.string(password[0]))
 		return HRESULT.S_OK.value
 		#return len(self.password)
 
 	def CryptoGetTextPassword2(self, me, isdefined, password):
 		log.debug('CryptoGetTextPassword2 me=%r password=%r%r', me, ffi.string(self.password), self.password)
-		isdefined[0] = 1
+		isdefined[0] = bool(self.password)
 		password[0] = self.password
 		log.debug('CryptoGetTextPassword returning, password=%s', ffi.string(password[0]))
 		return HRESULT.S_OK.value
 
-
 	#STDMETHOD(SetRatioInfo)(const UInt64 *inSize, const UInt64 *outSize) PURE;
 	def SetRatioInfo(self, me, in_size, out_size):
 		log.debug('SetRatioInfo: in_size=%d, out_size=%d' % (int(in_size[0]), int(out_size[0])))
@@ -100,12 +107,17 @@ def __init__(self, archive, directory='', password=''):
 		super().__init__(password)
 
 	def GetStream(self, me, index, outStream, askExtractMode):
-		log.debug('GetStream [%d]', index)
+		askExtractMode = AskMode(askExtractMode)
+		log.debug('GetStream(%d, -, %d)', index, askExtractMode)
+
+		if askExtractMode != AskMode.kExtract:
+			return HRESULT.S_OK.value
+
+
 		path = os.path.join(self.directory, self.archive[index].path)
 		dirname = os.path.dirname(path)
 		log.debug('extracting to: %s', path)
 
-
 		if self.archive[index].isdir:
 			os.makedirs(path, exist_ok=True)
 			outStream[0] = ffi.NULL
@@ -140,7 +152,12 @@ def __init__(self, stream, index, password=''):
 		super().__init__(password)
 
 	def GetStream(self, me, index, outStream, askExtractMode):
-		log.debug('GetStream')
+		askExtractMode = AskMode(askExtractMode)
+		log.debug('GetStream(%d, -, %d)', index, askExtractMode)
+
+		if askExtractMode != AskMode.kExtract:
+			return HRESULT.S_OK.value
+
 		if self.index == index:
 			outStream[0] = self.stream.instances[py7ziptypes.IID_ISequentialOutStream]
 			return HRESULT.S_OK.value

lib7zip/py7ziptypes.py

@@ -252,7 +252,7 @@ def createIID(yy, xx):
 
 #PropID.h
 
-class ArchiveProps:
+class ArchiveProps(IntEnum):
 	"""Archive and Archive Item Propertys"""
 	noproperty = 0  # kpidNoProperty
 	mainsubfile = 1  # kpidMainSubfile
@@ -326,3 +326,8 @@ class OperationResult(Enum):
 	kUnSupportedMethod = 1
 	kDataError = 2
 	kCRCError = 3
+
+class AskMode(Enum):
+	kExtract = 0
+	kTest = 1
+	kSkip = 2

lib7zip/winhelpers.py

@@ -7,6 +7,7 @@
 from .wintypes import *
 #from bitstring import BitArray
 #import warnings
+from datetime import datetime, timedelta
 
 def guidp2uuid(guid):
 	"""GUID* -> uuid.UUID"""
@@ -93,7 +94,7 @@ def get_prop_val(fn, forcetype=None, checktype=None):
 	ptr = alloc_propvariant()
 	RNOK(fn(ptr))
 	pvar = ffi.cast('PROPVARIANT*', ptr)
-	vt = forcetype or pvar.vt
+	vt = forcetype or VARTYPE(pvar.vt)
 	if pvar.vt in (VARTYPE.VT_EMPTY, VARTYPE.VT_NULL):
 		# always check for null, pvar.vt is not a bug
 		return None
@@ -113,5 +114,22 @@ def get_prop_val(fn, forcetype=None, checktype=None):
 		return guidp2uu(pvar.puuid)
 	elif vt == VARTYPE.VT_BSTR:
 		return ffi.string(pvar.bstrVal)
+	elif False: #vt == VARTYPE.VT_FILETIME:
+		#FIXME This should work but it totally doesn't
+		timestamp = int(pvar.filetime.dwLowDateTime)
+		timestamp += int(pvar.filetime.dwLowDateTime) << 32
+		#timestamp is in 100-nanosecond intervals, convert to nanoseconds
+		timestamp *= 100
+		#convert to seconds
+		timestamp /= 1e9
+
+		#timestamp is now the number of seconds since Jan 1, 1601 CE
+		jan01_1601 = datetime(year=1601, month=1, day=1)
+		delta = timedelta(seconds=timestamp)
+		import pdb; pdb.set_trace()
+		return jan01_1601 + delta
+
+		# timestamp isn't guaranteed to be UTC (it isn't on FAT for example)
+		# it is however recommended, hope 7zip enforces that.
 	else:
 		raise TypeError("type code %d not supported" % vt)

lib7zip/wintypes.py

@@ -8,7 +8,12 @@
 
 typedef uint8_t GUID[16];
 
-typedef struct PROPVARIANT {
+typedef struct {
+	uint32_t dwLowDateTime;
+	uint32_t dwHighDateTime;
+} FILETIME;
+
+typedef struct {
 	VARTYPE           vt;
 	unsigned short    wReserved1;
 	unsigned short    wReserved2;
@@ -26,9 +31,10 @@
 		wchar_t*          bstrVal;
 		uint64_t          uhVal;
 		GUID*             puuid;
+		FILETIME          filetime;
 		/* snip */
 	};
-	} PROPVARIANT;
+} PROPVARIANT;
 
 
 typedef uint32_t HRESULT;

tests/test_archive.py

@@ -23,6 +23,7 @@ def IX(isdir=False, crc=None, contents=''):
 	#return _IX(isdir, crc, contents.encode('utf8') if contents is not None else None)
 	return _IX(isdir, crc, contents)
 J = os.path.join
+
 COMPLEX_MD = {
 	J('complex','articles'): IX(True),
 	J('complex','articles','the definate article.txt'): IX(False, 0x3C456DE6, 'the'),
@@ -40,7 +41,6 @@ def test_complex():
 	with Archive('tests/complex.7z') as archive:
 		for item in archive:
 			log.debug(item.path)
-			#print(item.path)
 			try:
 				md = COMPLEX_MD[item.path]
 			except KeyError as ex:
@@ -64,7 +64,10 @@ def test_extract_dir_complex(tmp_dir):
 	for path, md in COMPLEX_MD.items():
 		if md.contents:
 			with open(os.path.join(tmp_dir, path), encoding='utf-8') as f:
-				assert f.read() == md.contents
+				file_contents = f.read()
+				#if 'unicode' in path:
+				#	import pdb; pdb.set_trace()
+				assert file_contents == md.contents
 
 @pytest.mark.parametrize('path', simple_archives)
 def test_extract_stream(path):