def process_document(self, session, doc):
data = doc.get_raw(session)
et = etree.parse(StringIO.StringIO(data), self.parser)
rec = LxmlRecord(et)
rec.byteCount = len(data)
self._copyData(doc, rec)
return rec
def process_document(self, session, doc):
data = doc.get_raw(session)
et = etree.parse(StringIO.StringIO(data), self.parser)
rec = LxmlRecord(et)
rec.byteCount = len(data)
self._copyData(doc, rec)
return rec
def process_document(self, session, doc):
# input must be string or stream
data = doc.get_raw(session)
try:
et = etree.parse(StringIO.StringIO(data), self.parser)
except AssertionError:
data = data.decode('utf8')
et = etree.parse(StringIO.StringIO(data), self.parser)
rec = LxmlRecord(et)
rec.byteCount = len(data)
self._copyData(doc, rec)
return rec
def process_document(self, session, doc):
# input must be string or stream
data = doc.get_raw(session)
try:
et = etree.parse(StringIO.StringIO(data), self.parser)
except AssertionError:
data = data.decode('utf8')
et = etree.parse(StringIO.StringIO(data), self.parser)
rec = LxmlRecord(et)
rec.byteCount = len(data)
self._copyData(doc, rec)
return rec
def _processResult(self, session, data):
"""Parse XML to create and return dict of metadata items.
Parse XML output from external program.
Process parsed XML using self.sources.
Populate and return a dictionary of metadata items.
"""
try:
et = etree.fromstring(data)
except AssertionError:
data = data.decode('utf8')
et = etree.fromstring(data)
except etree.XMLSyntaxError:
if session.logger is not None:
# log debug level
session.logger.log_lvl(session, 10, data)
raise
record = LxmlRecord(et)
record.byteCount = len(data)
mddict = {}
for key, src in self.sources.iteritems():
(xpath, process, preprocess) = src['source']
if preprocess is not None:
record = preprocess.process(session, record)
if xpath is not None:
rawlist = xpath.process_record(session, record)
processed = process.process(session, rawlist)
else:
processed = process.process(session, record)
if len(processed) > 1:
mddict[key] = []
for pl, k in sorted([(val['proxLoc'], k)
for k, val in processed.iteritems()]):
for x in pl:
mddict[key].append(k)
elif len(processed) == 1:
mddict[key] = processed.keys()[0]
elif src['default'] is not None:
mddict[key] = src['default']
return mddict
def _processResult(self, session, data):
"""Parse XML to create and return dict of metadata items.
Parse XML output from external program.
Process parsed XML using self.sources.
Populate and return a dictionary of metadata items.
"""
try:
et = etree.fromstring(data)
except AssertionError:
data = data.decode('utf8')
et = etree.fromstring(data)
except etree.XMLSyntaxError:
if session.logger is not None:
# log debug level
session.logger.log_lvl(session, 10, data)
raise
record = LxmlRecord(et)
record.byteCount = len(data)
mddict = {}
for key, src in self.sources.iteritems():
(xpath, process, preprocess) = src['source']
if preprocess is not None:
record = preprocess.process(session, record)
if xpath is not None:
rawlist = xpath.process_record(session, record)
processed = process.process(session, rawlist)
else:
processed = process.process(session, record)
if len(processed) > 1:
mddict[key] = []
for pl, k in sorted([(val['proxLoc'], k) for k, val in processed.iteritems()]):
for x in pl:
mddict[key].append(k)
elif len(processed) == 1:
mddict[key] = processed.keys()[0]
elif src['default'] is not None:
mddict[key] = src['default']
return mddict
