def brat2bioc_entity(bratentity: BratEntity) -> BioCAnnotation:
ann = BioCAnnotation()
ann.id = bratentity.id
ann.text = bratentity.text
ann.infons['type'] = bratentity.type
for span in bratentity.locations:
ann.add_location(BioCLocation(span.begin, span.end - span.begin))
return ann
def to_bioc(self):
entity_bioc = BioCAnnotation()
entity_bioc.infons['type'] = self.type
entity_bioc.text = self.text
entity_bioc.id = str(self.id)
location = BioCLocation(self.start, len(self.text))
entity_bioc.add_location(location)
return entity_bioc
def __parse_annotation(self, tree):
annotation = BioCAnnotation()
annotation.id = tree.attrib['id']
annotation.infons = self.__parse_infons(tree)
annotation.text = tree.findtext('text')
for child in tree.findall('location'):
annotation.add_location(
BioCLocation(int(child.attrib['offset']), int(child.attrib['length'])))
return annotation
def add_annotation(triple, annotation_id):
# initialize annotation element
bioc_annotation = BioCAnnotation()
# MeSH term in a tag <text> ... </text> (origininal term, searched case insensitive)
bioc_annotation.text = triple[2]
# generate XML structure for the annotation and add infon
bioc_annotation.id = str(annotation_id)
bioc_annotation.put_infon('type', 'MeSH term')
# add location element
bioc_location = BioCLocation()
# add length of MeSH term
bioc_location.length = str(triple[1])
# add start position (offset)
bioc_location.offset = str(triple[0])
bioc_annotation.add_location(bioc_location)
return bioc_annotation
def add_annotation(triple, annotation_id):
# initialize annotation element
bioc_annotation = BioCAnnotation()
# MeSH term in a tag <text> ... </text> (origininal term, searched case insensitive)
bioc_annotation.text = triple[2]
# generate XML structure for the annotation and add infon
bioc_annotation.id = str(annotation_id)
bioc_annotation.put_infon('type', 'MeSH term')
# add location element
bioc_location = BioCLocation()
# add length of MeSH term
bioc_location.length = str(triple[1])
# add start position (offset)
bioc_location.offset = str(triple[0])
bioc_annotation.add_location(bioc_location)
return bioc_annotation
def __read_annotation(self, start_elem):
ann = BioCAnnotation()
ann.id = start_elem.get('id')
while self.__has_next():
event, elem = self.__next_event()
if event == 'start':
pass
elif event == 'end':
if elem.tag == 'text':
ann.text = elem.text
elif elem.tag == 'infon':
ann.infons[elem.get('key')] = elem.text
elif elem.tag == 'location':
ann.add_location(BioCLocation(int(elem.get('offset')), int(elem.get('length'))))
elif elem.tag == 'annotation':
return ann
raise RuntimeError("should not reach here") # pragma: no cover
