def mark_pair():
if not flask_session.get('session_id'):
return make_response(jsonify(status='error', message='need to start a session'), 400)
else:
action = request.args['action']
deduper_id = flask_session['session_id']
dedupers[deduper_id]['last_interaction'] = datetime.now()
if dedupers[deduper_id].get('counter'):
counter = dedupers[deduper_id]['counter']
else:
counter = {'yes': 0, 'no': 0, 'unsure': 0}
if dedupers[deduper_id].get('training_data'):
labels = dedupers[deduper_id]['training_data']
else:
labels = {'distinct' : [], 'match' : []}
deduper = dedupers[deduper_id]['deduper']
if action == 'yes':
current_pair = dedupers[deduper_id]['current_pair']
labels['match'].append(current_pair)
counter['yes'] += 1
resp = {'counter': counter}
elif action == 'no':
current_pair = dedupers[deduper_id]['current_pair']
labels['distinct'].append(current_pair)
counter['no'] += 1
resp = {'counter': counter}
elif action == 'finish':
file_io = dedupers[deduper_id]['csv']
training_file_path = os.path.join(UPLOAD_FOLDER, '%s-training.json' % file_io.file_path)
training_data = dedupers[deduper_id]['training_data']
with open(training_file_path, 'wb') as f:
f.write(json.dumps(training_data, default=_to_json))
field_defs = dedupers[deduper_id]['field_defs']
sample = deduper.data_sample
args = {
'field_defs': field_defs,
'training_data': training_file_path,
'file_io': file_io,
'data_sample': sample,
}
rv = dedupeit.delay(**args)
flask_session['deduper_key'] = rv.key
resp = {'finished': True}
flask_session['dedupe_start'] = time.time()
else:
counter['unsure'] += 1
dedupers[deduper_id]['counter'] = counter
resp = {'counter': counter}
deduper.markPairs(labels)
dedupers[deduper_id]['training_data'] = labels
dedupers[deduper_id]['counter'] = counter
if resp.get('finished'):
del dedupers[deduper_id]
resp = make_response(json.dumps(resp))
resp.headers['Content-Type'] = 'application/json'
return resp
