ireland-news-headlines/rockyRoadtoDublin.py
2022-05-28 16:44:54 +02:00

38 lines
1.5 KiB
Python

import inout as io
def generateChooChoo(data, target, categories):
data = [x[2].replace('\n', '').replace(':', '') for x in io.read(data)]
target = [x[0].replace('\n', '') for x in io.read(target)]
if categories == {}:
i = 0
for x in target:
if x not in categories:
categories[x] = i
i += 1
return {'data': data, 'target': target}, categories
def predictFuture(test):
data = [x[2].replace('\n', '').replace(':', '') for x in io.read(test + '/in.tsv')]
with open('vw_ireland_news_' + test, 'w', encoding='utf-8') as f:
for text in data:
f.write('1 |text ' + text + '\n')
if __name__ == '__main__':
ireland_news_train, categories = generateChooChoo('train/in.tsv.xz', 'train/expected.tsv.xz', categories={})
ireland_news_dev, _ = generateChooChoo('dev-0/in.tsv', 'dev-0/expected.tsv', categories)
with open('vw_ireland_news_train', 'w', encoding='utf-8') as f:
for target, text in zip(ireland_news_train['target'], ireland_news_train['data']):
f.write(str(categories[target] + 1) + ' |text ' + text + '\n')
with open('vw_ireland_news_dev0', 'w', encoding='utf-8') as f, open('ireland_news_dev0_targets', 'w', encoding='utf-8') as f_targets:
for target, text in zip(ireland_news_dev['target'], ireland_news_dev['data']):
f.write('1 |text ' + text + '\n')
f_targets.write(str(categories[target] + 1) + '\n')
predictFuture('test-A')
predictFuture('test-B')