def extract_make_instance_offsets(annotated_file):
"""Map MAKEINSTANCE entities to their respective offsets within text
"""
_, id_to_offset = extract_labeled_entities(annotated_file)
make_instances = get_make_instances(annotated_file)
make_instance_offsets = {}
for make_instance in make_instances:
offset = id_to_offset[make_instance.attrib["eventID"]]
make_instance_offsets[offset] = {"tense":make_instance.attrib["tense"],
"aspect":make_instance.attrib["aspect"],
"polarity":make_instance.attrib["polarity"],
"pos":make_instance.attrib["pos"]}
return make_instance_offsets
def extract_tlinks(annotated_timeml):
"""Return offset pairs and the relations between them.
"""
# offset pairs to rel types
tlinks = {}
_, id_to_offset = extract_labeled_entities(annotated_timeml)
make_instances = get_make_instances(annotated_timeml)
event_instance_id_to_event_id = {make_instance.attrib["eiid"]:make_instance.attrib["eventID"] for make_instance in make_instances}
# add in an offset for DOCTIME
id_to_offset['t0'] = (-1,-1)
for tlink in get_tlinks(annotated_timeml):
attribs = tlink.attrib
"""
print id_to_offset
print
print attribs["eventInstanceID"]
print event_instance_id_to_event_id[attribs["eventInstanceID"]]
print id_to_offset[event_instance_id_to_event_id[attribs["eventInstanceID"]]]
print
"""
if "eventInstanceID" not in attribs:
continue
event_offset = id_to_offset[event_instance_id_to_event_id[attribs["eventInstanceID"]]]
target_offset = id_to_offset[event_instance_id_to_event_id[attribs["relatedToEventInstance"]] if "relatedToEventInstance" in attribs else attribs["relatedToTime"]]
if (event_offset, target_offset) in tlinks:
sys.exit("ERROR: duplicate TLINKs")
tlinks[(event_offset, target_offset)] = attribs["relType"]
return tlinks
