def generate_features(input_data, mode, filename=None):
new_data = []
from umls import UMLS
from generate_features import Generator
cui_retriever = UMLS()
feature_generator = Generator(cui_retriever)
def save_one_line(doc_id, sentence_id, features, filename):
line = ','.join([doc_id, sentence_id, features]) + "\n"
# save data to files
with open(filename, 'a') as f:
f.writelines(line)
for i, line in enumerate(input_data):
import datetime
print(i, datetime.datetime.now(), mode, line)
sentence = line[-1]
sentence_id = line[-2]
doc_id = line[-3]
features = feature_generator.produce_features_from_sentence(
sentence, sentence_id, doc_id, mode)
new_data.append(features)
if filename is not None:
# save the data
save_one_line(doc_id, sentence_id, features, filename)
return new_data
