def updateBaseMetadata(self, **kw):
"""
Updates metadata information in the converted OOo document
based on the values provided by the user. This is implemented
through the invocation of the conversion server.
"""
if not self.hasBaseData():
# XXX please pass a meaningful description of error as argument
raise NotConvertedError()
server_proxy = DocumentConversionServerProxy(self)
response_code, response_dict, response_message = \
server_proxy.run_setmetadata(self.getId(),
enc(str(self.getBaseData())),
kw)
if response_code == 200:
# successful meta data extraction
self._setBaseData(dec(response_dict['data']))
self.updateFileMetadata(
) # record in workflow history # XXX must put appropriate comments.
else:
# Explicitly raise the exception!
raise ConversionError(
"OOoDocument: error getting document metadata (Code %s: %s)" %
(response_code, response_message))
def _getConversionFromProxyServer(self, format):
"""
Communicates with server to convert a file
"""
if not self.hasBaseData():
# XXX please pass a meaningful description of error as argument
raise NotConvertedError()
if format == 'text-content':
# Extract text from the ODF file
cs = cStringIO.StringIO()
cs.write(str(self.getBaseData()))
z = zipfile.ZipFile(cs)
s = z.read('content.xml')
s = self.rx_strip.sub(" ", s) # strip xml
s = self.rx_compr.sub(" ", s) # compress multiple spaces
cs.close()
z.close()
return 'text/plain', s
server_proxy = DocumentConversionServerProxy(self)
orig_format = self.getBaseContentType()
generate_result = server_proxy.run_generate(
self.getId(), enc(str(self.getBaseData())), None, format,
orig_format)
try:
_, response_dict, _ = generate_result
except ValueError:
# This is for backward compatibility with older oood version returning
# only response_dict
response_dict = generate_result
# XXX: handle possible OOOd server failure
return response_dict['mime'], Pdata(dec(response_dict['data']))
def _convertToBaseFormat(self):
"""
Converts the original document into ODF
by invoking the conversion server. Store the result
on the object. Update metadata information.
"""
server_proxy = DocumentConversionServerProxy(self)
response_code, response_dict, response_message = server_proxy.run_convert(
self.getFilename() or self.getId(), enc(str(self.getData())), None,
None, self.getContentType())
if response_code == 200:
# sucessfully converted document
self._setBaseData(dec(response_dict['data']))
metadata = response_dict['meta']
self._base_metadata = metadata
if metadata.get('MIMEType', None) is not None:
self._setBaseContentType(metadata['MIMEType'])
else:
# Explicitly raise the exception!
raise ConversionError(
"OOoDocument: Error converting document to base format. (Code %s: %s)"
% (response_code, response_message))
def convert(self, orig, data, context=None, **kwargs):
server_proxy = DocumentConversionServerProxy(context)
source_mimetype = self._getAllowedSourceMimetypeFromConversionServer(
server_proxy)
if source_mimetype is None:
raise ConversionError(
"Format(s) not allowed on Conversion Server %r" % self.inputs)
source_format = self._getFormatFromMimetype(source_mimetype)
destination_format = self._getFormatFromMimetype(self.output)
data.setData(
dec(
server_proxy.convertFile(
enc(orig),
source_format,
destination_format,
# Default values are ConversionServer default ones
kwargs.get('zip', False),
kwargs.get('refresh', False),
kwargs.get('conversion_kw', {}))))
return data