in reference/src/main/python/similar.py [0:0]
def featurize_and_test_record(record_files, keywords):
set_tmp = None
record_final = None
for record_file in record_files:
record = read_and_featurize_record_file(record_file)
if record is not None:
record_final = record
if set_tmp is not None:
set_tmp = set_tmp & Counter(record["features"])
else:
set_tmp = Counter(record["features"])
# need to figure out how to merge asts as well
if set_tmp is None:
set_tmp = Counter()
for keyword in keywords:
set_tmp[vocab.get_index(keyword)] += 1
if record_final is None:
record_final = {"ast": None, "index": -1, "features": list(set_tmp.elements())}
else:
record_final["features"] = list(set_tmp.elements())
if len(record_final["features"]) > 0:
print_similar_and_completions(record_final, records, vectorizer, counter_matrix)