def COMOpen(progID, events=None): try: if events == None: return win32com.client.Dispatch(progID) else: return win32com.client.DispatchWithEvents(progID, events) except: error.abortCheck() error.traceback() return None
def COMOpen(progID, events=None):
try:
if events == None:
return win32com.client.Dispatch(progID)
else:
return win32com.client.DispatchWithEvents(progID, events)
except:
error.abortCheck()
error.traceback()
return None
def getFromFile(filename):
props = docProps()
try:
props.fileOwner = fileOwner(filename)
except:
error.abortCheck()
props.fileOwner = None
stg = wincom.StgOpenStorage(
filename, None, wincom.storagecon.STGM_READ
| wincom.storagecon.STGM_SHARE_EXCLUSIVE)
pss = stg.QueryInterface(wincom.IID_IPropertySetStorage)
ps = pss.Open(FMTID_SummaryInformation)
data = ps.ReadMultiple(allSummaryInformation)
props.path = filename
props.title = unicodeFix(data[0])
props.author = unicodeFix(data[2])
props.subject = unicodeFix(data[1])
props.keywords = unicodeFix(data[3])
props.comments = unicodeFix(data[4])
props.template = unicodeFix(data[5])
props.lastAuthor = unicodeFix(data[6])
props.revNumber = unicodeFix(data[7])
props.lastEditTime = PyTime2datetime(data[8])
props.lastPrintedTime = PyTime2datetime(data[9])
props.createdTime = PyTime2datetime(data[10])
props.lastSavedTime = PyTime2datetime(data[11])
props.pageCount = data[12]
props.wordCount = data[13]
props.charCount = data[14]
props.appName = unicodeFix(data[15])
props.docSecurity = data[16]
try:
ps = pss.Open(FMTID_DocSummaryInformation)
data = ps.ReadMultiple([PIDDSI_COMPANY])
# except pythoncom.com_err: No workum, use plain except instead. Sigh.
except:
error.abortCheck()
error.traceback()
data = [""]
props.company = unicodeFix(data[0])
return props
def processFile(entry, options):
if entry.attributes & iolib.FILE_ATTRIBUTE_DIRECTORY:
return
filename = entry.path
filename2 = filename
tempFilename = None
if iolib.webfile.isWebPath(filename) or iolib.zipfile.isZipPath(filename):
tempFilename = iolib.makeTempCopy(filename)
if tempFilename == None:
return
filename2 = tempFilename
oldStdout = options.outRedirector.redirect(entry, options)
if oldStdout == None:
iolib.removeTempCopy(tempFilename)
return
if not wincom.StgIsStorageFile(filename2):
error.signal("File %s is not a Micorosft Office document file",
(filename))
iolib.removeTempCopy(tempFilename)
return
try:
props = docProps.getFromFile(filename2)
props.path = filename
except:
error.exception("Unable to access file %s", (filename))
iolib.removeTempCopy(tempFilename)
return
try:
revLog = wordRevLog(filename2)
except:
error.abortCheck()
revLog = None
entry.md5 = None
if options.md5f:
entry.md5 = md5file(filename2)
iolib.removeTempCopy(tempFilename)
options.outClass.showMetadata(entry, props, revLog, options)
options.outRedirector.restore(oldStdout)
return
def getFromFile(filename): props = docProps() try: props.fileOwner = fileOwner(filename) except: error.abortCheck() props.fileOwner = None stg = wincom.StgOpenStorage(filename, None, wincom.storagecon.STGM_READ | wincom.storagecon.STGM_SHARE_EXCLUSIVE) pss = stg.QueryInterface(wincom.IID_IPropertySetStorage) ps = pss.Open(FMTID_SummaryInformation) data = ps.ReadMultiple(allSummaryInformation) props.path = filename props.title = unicodeFix(data[0]) props.author = unicodeFix(data[2]) props.subject = unicodeFix(data[1]) props.keywords = unicodeFix(data[3]) props.comments = unicodeFix(data[4]) props.template = unicodeFix(data[5]) props.lastAuthor = unicodeFix(data[6]) props.revNumber = unicodeFix(data[7]) props.lastEditTime = PyTime2datetime(data[8]) props.lastPrintedTime = PyTime2datetime(data[9]) props.createdTime = PyTime2datetime(data[10]) props.lastSavedTime = PyTime2datetime(data[11]) props.pageCount = data[12] props.wordCount = data[13] props.charCount = data[14] props.appName = unicodeFix(data[15]) props.docSecurity = data[16] try: ps = pss.Open(FMTID_DocSummaryInformation) data = ps.ReadMultiple([PIDDSI_COMPANY]) # except pythoncom.com_err: No workum, use plain except instead. Sigh. except: error.abortCheck() error.traceback() data = [""] props.company = unicodeFix(data[0]) return props
def processFile(entry, options):
if entry.attributes & iolib.FILE_ATTRIBUTE_DIRECTORY:
return
filename = entry.path
filename2 = filename
tempFilename = None
if iolib.webfile.isWebPath(filename) or iolib.zipfile.isZipPath(filename):
tempFilename = iolib.makeTempCopy(filename)
if tempFilename == None:
return
filename2 = tempFilename
oldStdout = options.outRedirector.redirect(entry, options)
if oldStdout == None:
iolib.removeTempCopy(tempFilename)
return
if not wincom.StgIsStorageFile(filename2):
error.signal("File %s is not a Micorosft Office document file", (filename))
iolib.removeTempCopy(tempFilename)
return
try:
props = docProps.getFromFile(filename2)
props.path = filename
except:
error.exception("Unable to access file %s", (filename))
iolib.removeTempCopy(tempFilename)
return
try:
revLog = wordRevLog(filename2)
except:
error.abortCheck()
revLog = None
entry.md5 = None
if options.md5f:
entry.md5 = md5file(filename2)
iolib.removeTempCopy(tempFilename)
options.outClass.showMetadata(entry, props, revLog, options)
options.outRedirector.restore(oldStdout)
return
def wordRevLog(filename):
try:
stg = wincom.StgOpenStorage(
filename, None, wincom.storagecon.STGM_READ
| wincom.storagecon.STGM_SHARE_EXCLUSIVE)
except:
error.abortCheck("Cannot open file -- %s", filename)
return None
try:
stream = stg.OpenStream("WordDocument", None,
wincom.storagecon.STGM_SHARE_EXCLUSIVE, 0)
contents = stream.Read(0x02DA)
if contents == None:
return None
except:
error.abortCheck()
return None
filehdr = BinReader(contents)
sign = filehdr.getWord()
if sign != WIN95_IDENT and sign != WIN97_IDENT:
return None
filehdr.seek(0x02D2)
revLogOffset = filehdr.getDWord()
revLogSize = filehdr.getDWord()
if revLogOffset == None or revLogSize == None or revLogSize == 0:
return None
try:
stream = stg.OpenStream("1Table", None,
wincom.storagecon.STGM_SHARE_EXCLUSIVE, 0)
contents = stream.Read(revLogOffset + revLogSize)
if contents == None:
return None
except:
error.abortCheck()
return None
revLog = BinReader(contents)
revLogTable = []
revLog.seek(revLogOffset + 2)
count = revLog.getDWord()
if count == None or count == 0 or count > 20 or (count % 2) != 0:
return None
count = count / 2
for i in xrange(count):
entry = SimpleStruct()
n = revLog.getWord()
if n == 0 or n > 512:
break
entry.person = revLog.getUString(n)
n = revLog.getWord()
if n == 0 or n > 512:
break
entry.filePath = revLog.getUString(n)
revLogTable.append(entry)
return revLogTable
def wordRevLog(filename):
try:
stg = wincom.StgOpenStorage(filename, None, wincom.storagecon.STGM_READ | wincom.storagecon.STGM_SHARE_EXCLUSIVE)
except:
error.abortCheck("Cannot open file -- %s", filename)
return None
try:
stream = stg.OpenStream("WordDocument", None, wincom.storagecon.STGM_SHARE_EXCLUSIVE, 0)
contents = stream.Read(0x02DA);
if contents == None:
return None
except:
error.abortCheck()
return None
filehdr = BinReader(contents)
sign = filehdr.getWord();
if sign != WIN95_IDENT and sign != WIN97_IDENT:
return None
filehdr.seek(0x02D2)
revLogOffset = filehdr.getDWord()
revLogSize = filehdr.getDWord()
if revLogOffset == None or revLogSize == None or revLogSize == 0:
return None
try:
stream = stg.OpenStream("1Table", None, wincom.storagecon.STGM_SHARE_EXCLUSIVE, 0)
contents = stream.Read(revLogOffset + revLogSize);
if contents == None:
return None
except:
error.abortCheck()
return None
revLog = BinReader(contents)
revLogTable = []
revLog.seek(revLogOffset + 2)
count = revLog.getDWord()
if count == None or count == 0 or count > 20 or (count % 2) != 0:
return None
count = count / 2
for i in xrange(count):
entry = SimpleStruct()
n = revLog.getWord()
if n == 0 or n > 512:
break
entry.person = revLog.getUString(n)
n = revLog.getWord()
if n == 0 or n > 512:
break
entry.filePath = revLog.getUString(n)
revLogTable.append(entry)
return revLogTable
