def get_aligned_frames_xml(tokenized, frame_instances, root):
# read DRG
tuples = get_drg(tokenized)
drgparser = drg.DRGParser()
d = drgparser.parse_tup_lines(tuples)
for instance_id, frame_instance in frame_instances.iteritems():
if len(frame_instance['roles']) > 0:
if frame_instance['frame'] != "Unmapped":
framebase_id = "{0}-{1}".format(frame_instance['frame'], mapping_net[frame_instance['synset']].split("#")[0].replace('-', '.'))
else:
log.info('No mapping found for synset {0}'.format(frame_instance['synset']))
continue
tag_frameinstance = objectify.SubElement(root, "frameinstance")
tag_frameinstance.attrib['id']=instance_id
tag_frameinstance.attrib['type']=framebase_id
tag_frameinstance.attrib['internalvariable']=frame_instance['variable']
for reificated_frame_var in d.reificated[frame_instance['variable']]:
tag_framelexicalization = objectify.SubElement(tag_frameinstance, "framelexicalization")
surface = []
unboxer.generate_from_referent(d, reificated_frame_var, surface, complete=False)
tag_framelexicalization[0] = ' '.join(surface)
tag_instancelexicalization = objectify.SubElement(tag_frameinstance, "instancelexicalization")
surface = []
unboxer.generate_from_referent(d, reificated_frame_var, surface, complete=True)
tag_instancelexicalization[0] = ' '.join(surface)
tag_frameelements = objectify.SubElement(tag_frameinstance, "frameelements")
for role, (variable, filler) in frame_instance['roles'].iteritems():
tag_frameelement = objectify.SubElement(tag_frameelements, "frameelement")
tag_frameelement.attrib['role'] = role
tag_frameelement.attrib['internalvariable'] = variable
tag_concept = objectify.SubElement(tag_frameelement, "concept")
tag_concept[0] = filler
try:
for reificated_role_var in d.reificated[variable]:
# composed lexicalization
surface = unboxer.generate_from_relation(d, reificated_frame_var, reificated_role_var)
if surface != None:
tag_rolelexicalization = objectify.SubElement(tag_frameelement, "rolelexicalization")
tag_rolelexicalization[0] = surface
# complete surface forms
surface = []
unboxer.generate_from_referent(d, reificated_role_var, surface, complete=True)
tag_conceptlexicalization = objectify.SubElement(tag_frameelement, "conceptlexicalization")
tag_conceptlexicalization[0] = ' '.join(surface)
except:
log.error("error with DRG reification: {0}".format(variable))
objectify.deannotate(root, xsi_nil=True)
etree.cleanup_namespaces(root)
return etree.tostring(root, pretty_print=True)
def get_aligned_frames_xml(tokenized, frame_instances, root):
# read DRG
tuples = get_drg(tokenized)
drgparser = drg.DRGParser()
d = drgparser.parse_tup_lines(tuples)
for instance_id, frame_instance in frame_instances.iteritems():
if len(frame_instance['roles']) > 0:
try:
framebase_id = "{0}-{1}".format(frame_instance['frame'], offset2wn[frame_instance['synset']].split("#")[0].replace('-', '.'))
except:
log.info('No mapping found for synset {0}'.format(frame_instance['synset']))
continue
tag_frameinstance = objectify.SubElement(root, "frameinstance")
tag_frameinstance.attrib['id']=instance_id
tag_frameinstance.attrib['type']=framebase_id
tag_frameinstance.attrib['internalvariable']=frame_instance['variable']
for reificated_frame_var in d.reificated[frame_instance['variable']]:
tag_framelexicalization = objectify.SubElement(tag_frameinstance, "framelexicalization")
surface = []
unboxer.generate_from_referent(d, reificated_frame_var, surface, complete=False)
tag_framelexicalization[0] = ' '.join(surface)
tag_instancelexicalization = objectify.SubElement(tag_frameinstance, "instancelexicalization")
surface = []
unboxer.generate_from_referent(d, reificated_frame_var, surface, complete=True)
tag_instancelexicalization[0] = ' '.join(surface)
tag_frameelements = objectify.SubElement(tag_frameinstance, "frameelements")
for role, (variable, filler) in frame_instance['roles'].iteritems():
tag_frameelement = objectify.SubElement(tag_frameelements, "frameelement")
tag_frameelement.attrib['role'] = role
tag_frameelement.attrib['internalvariable'] = variable
tag_concept = objectify.SubElement(tag_frameelement, "concept")
tag_concept[0] = filler
try:
for reificated_role_var in d.reificated[variable]:
# composed lexicalization
surface = unboxer.generate_from_relation(d, reificated_frame_var, reificated_role_var)
if surface != None:
tag_rolelexicalization = objectify.SubElement(tag_frameelement, "rolelexicalization")
tag_rolelexicalization[0] = surface
# complete surface forms
surface = []
unboxer.generate_from_referent(d, reificated_role_var, surface, complete=True)
tag_conceptlexicalization = objectify.SubElement(tag_frameelement, "conceptlexicalization")
tag_conceptlexicalization[0] = ' '.join(surface)
except:
log.error("error with DRG reification: {0}".format(variable))
objectify.deannotate(root, xsi_nil=True)
etree.cleanup_namespaces(root)
return etree.tostring(root, pretty_print=True)