challenging-america-word-ga.../generate_out.py

12 lines
303 B
Python
Raw Permalink Normal View History

2023-03-29 11:09:48 +02:00
import sys
file = sys.argv[1]
with open(file, encoding='utf-8') as f1, open('out.tsv', 'w', encoding='utf-8') as f2:
for line in f1:
line = line.split('\t')
if line[-1][0].isupper():
f2.write('the:0.9 :0.1\n')
else:
f2.write('the:0.4 a:0.4 :0.2\n')