def loads_xml(cls, bytestring, encoding=None, **kwargs):
"""
Currently processes "<dmrs>...</dmrs>"
To be updated for "<dmrslist>...</dmrslist>"...
Expects a bytestring; to load from a string instead, specify encoding
"""
from pydmrs.serial import loads_xml
return loads_xml(bytestring, encoding=encoding, cls=cls, **kwargs)
def loads_xml(cls, bytestring, encoding=None, **kwargs):
"""
Currently processes "<dmrs>...</dmrs>"
To be updated for "<dmrslist>...</dmrslist>"...
Expects a bytestring; to load from a string instead, specify encoding
"""
from pydmrs.serial import loads_xml
return loads_xml(bytestring, encoding=encoding, cls=cls, **kwargs)
test_connectedness = not allow_disconnected_dmrs and dmrs.is_connected(ignored_nodeids=filterable_nodeids)
# If DMRS should remain connected, check that removing filterable nodes will not result in a disconnected DMRS
if test_connectedness:
filtered_nodeids = set()
for nodeid in filterable_nodeids:
if dmrs.is_connected(removed_nodeids=filtered_nodeids|{nodeid}, ignored_nodeids=filterable_nodeids):
filtered_nodeids.add(nodeid)
else:
filtered_nodeids = filterable_nodeids
# Remove filtered nodes and their links from the DMRS
for nodeid in filtered_nodeids:
dmrs.remove_node(nodeid)
return dmrs
# If run from the command line, process the given file
if __name__ == '__main__':
with open(args.input_dmrs, 'r', encoding="utf-8") as fin, open(args.output_dmrs, 'w') as fout:
content = fin.read().strip()
for dmrs_string in split_dmrs_string(content):
dmrs = loads_xml(dmrs_string)
simplified_dmrs = gpred_filtering(dmrs)
simplified_dmrs_string = dumps_xml(simplified_dmrs)
fout.write('{}\n\n'.format(simplified_dmrs_string.decode('utf-8')))
# If DMRS should remain connected, check that removing filterable nodes will not result in a disconnected DMRS
if test_connectedness:
filtered_nodeids = set()
for nodeid in filterable_nodeids:
if dmrs.is_connected(removed_nodeids=filtered_nodeids | {nodeid},
ignored_nodeids=filterable_nodeids):
filtered_nodeids.add(nodeid)
else:
filtered_nodeids = filterable_nodeids
# Remove filtered nodes and their links from the DMRS
for nodeid in filtered_nodeids:
dmrs.remove_node(nodeid)
return dmrs
# If run from the command line, process the given file
if __name__ == '__main__':
with open(args.input_dmrs, 'r',
encoding="utf-8") as fin, open(args.output_dmrs, 'w') as fout:
content = fin.read().strip()
for dmrs_string in split_dmrs_string(content):
dmrs = loads_xml(dmrs_string)
simplified_dmrs = gpred_filtering(dmrs)
simplified_dmrs_string = dumps_xml(simplified_dmrs)
fout.write('{}\n\n'.format(simplified_dmrs_string.decode('utf-8')))