def parse_xml(file_name, size=0, method="json"):
"""Parse single XML file into the separated XML instances.
Args:
:param file_name(str): Name of file within same level as root.
:param size(int): Number of XML instances within XML file to be read.
:param method(str): Export method(Default:Json).
Returns:
int. The return code::
0 -- Success!
"""
# Set initial values
count = 0
results = []
# add benchmark
bm = BenchMark()
global MAX_NUMBER_OF_PATENTS
# create an XML Reader
parser = xml.sax.make_parser()
# turn off namespaces
parser.setFeature(xml.sax.handler.feature_namespaces, 0)
# turn off validation for DTD
parser.setFeature(xml.sax.handler.feature_external_ges, False)
# override the default Context Handler
xml_patent_handler = PatentHandler()
parser.setContentHandler(xml_patent_handler)
try:
with open(file_name) as citation:
bm.toggleOn('Start processing [ ]')
for xml_part in xml_documents(citation):
# Cast string back to file-like object to parse
parser.parse(cStringIO.StringIO(xml_part))
results.append(
copy.deepcopy(xml_patent_handler.serialization()))
count = count + 1
if not MAX_NUMBER_OF_PATENTS:
if count == int(size):
break
# Clean up stack after processing one xml paragraph
xml_patent_handler.reset()
bm.add(0)
bm.toggleOff(' \bOK] - ' + str(count) + ' patents ')
if method == "json":
export2json(results)
return 0
except IOError as e:
raise e
def parse_xml(file_name, size = 0, method = "json"):
"""Parse single XML file into the separated XML instances.
Args:
:param file_name(str): Name of file within same level as root.
:param size(int): Number of XML instances within XML file to be read.
:param method(str): Export method(Default:Json).
Returns:
int. The return code::
0 -- Success!
"""
# Set initial values
count = 0
results = []
# add benchmark
bm = BenchMark()
global MAX_NUMBER_OF_PATENTS
# create an XML Reader
parser = xml.sax.make_parser()
# turn off namespaces
parser.setFeature(xml.sax.handler.feature_namespaces, 0)
# turn off validation for DTD
parser.setFeature(xml.sax.handler.feature_external_ges, False)
# override the default Context Handler
xml_patent_handler = PatentHandler()
parser.setContentHandler(xml_patent_handler)
try:
with open(file_name) as citation:
bm.toggleOn('Start processing [ ]')
for xml_part in xml_documents(citation):
# Cast string back to file-like object to parse
parser.parse(cStringIO.StringIO(xml_part))
results.append(copy.deepcopy(xml_patent_handler.serialization()))
count = count+1
if not MAX_NUMBER_OF_PATENTS:
if count == int(size):
break
# Clean up stack after processing one xml paragraph
xml_patent_handler.reset()
bm.add(0)
bm.toggleOff(' \bOK] - '+ str(count) + ' patents ')
if method == "json":
export2json(results)
return 0
except IOError as e:
raise e
def export2json(data):
"""Export to json file
Args:
:param data(list): data list to be exported.
Returns:
int. The return code::
0 -- Success!
"""
bm = BenchMark()
bm.toggleOn('Dumping to json [ ]')
with open('data.json', 'w') as outfile:
json.dump(data, outfile, indent=4, ensure_ascii=False)
bm.toggleOff('\bOK] - saved to data.json')
return 0
def export2json(data):
"""Export to json file
Args:
:param data(list): data list to be exported.
Returns:
int. The return code::
0 -- Success!
"""
bm = BenchMark()
bm.toggleOn('Dumping to json [ ]')
with open('data.json', 'w') as outfile:
json.dump(data, outfile, indent=4, ensure_ascii=False)
bm.toggleOff('\bOK] - saved to data.json')
return 0
