def load_ifilter(filename, log=log):
try:
return ifilter.LoadIFilter(filename)
except pythoncom.com_error, e:
if e[0] == FILTER_E_UNKNOWNFORMAT:
log.warning("File %s is not a recognized format" % filename)
else:
log.warning("LoadIFilter for file %s, raised error %s, file cannot be processed" % (filename, str(e)))
raise
def _bind_to_filter(self, fileName):
"""
See if the file is a structured storage file or a normal file
and then return an ifilter interface by calling the appropriate bind/load function
"""
if pythoncom.StgIsStorageFile(fileName):
self.stg = pythoncom.StgOpenStorage(
fileName, None,
storagecon.STGM_READ | storagecon.STGM_SHARE_DENY_WRITE)
try:
self.f = ifilter.BindIFilterFromStorage(self.stg)
except pythoncom.com_error as e:
if e[0] == -2147467262: # 0x80004002: # no interface, try the load interface (this happens for some MSoft files)
self.f = ifilter.LoadIFilter(fileName)
else:
raise
else:
self.f = ifilter.LoadIFilter(fileName)
self.stg = None
def _bind_to_filter(self, fileName):
"""
See if the file is a structured storage file or a normal file
and then return an ifilter interface by calling the appropriate bind/load function
"""
if pythoncom.StgIsStorageFile(fileName):
self.stg = pythoncom.StgOpenStorage(fileName, None, storagecon.STGM_READ | storagecon.STGM_SHARE_DENY_WRITE)
try:
self.f = ifilter.BindIFilterFromStorage(self.stg)
except pythoncom.com_error, e:
if e[0] == -2147467262: # 0x80004002: # no interface, try the load interface (this happens for some MSoft files)
self.f = ifilter.LoadIFilter(fileName)
else:
raise
else:
self.f = ifilter.LoadIFilter(fileName)
self.stg = None
def _get_text(self, body_chunks):
"""
Gets all the text for a particular chunk. We need to keep calling get text till all the
segments for this chunk are retrieved
"""
while True:
try:
body_chunks.append(self.f.GetText())
except pythoncom.com_error, e:
if e[0] in [FILTER_E_NO_MORE_TEXT, FILTER_E_NO_MORE_TEXT, FILTER_E_NO_TEXT]:
break
else:
raise # not one of the values we were expecting