def get_or_create_participant(corpus, attr_map, target_child=None):
if not attr_map:
return None
age = parse_age(attr_map.get('age'))
code = attr_map.get('id')
name = attr_map.get('name')
role = attr_map.get('role')
language = attr_map.get('language')
group = attr_map.get('group')
sex = attr_map.get('sex')
ses = attr_map.get('SES')
education = attr_map.get('education')
custom = attr_map.get('custom')
query_set = Participant.objects.filter(code=code, name=name, role=role, corpus=corpus)
# Filter participant candidates by target child
if target_child:
query_set = query_set.filter(target_child=target_child)
participant = query_set.first()
if not participant:
participant = Participant(
code=code,
name=name,
role=role,
language=language,
group=group,
sex=sex,
ses=ses,
education=education,
custom=custom,
corpus=corpus,
corpus_name=corpus.name,
collection=corpus.collection,
collection_name=corpus.collection.name
)
if target_child:
participant.target_child = target_child
update_age(participant, age)
# TODO very confusing. in memory attribute gets passed to child process
# Mark the age for this participant for this transcript, to be saved in utterance / token as speaker_age
participant.age = age
participant.save()
return participant
