import pandas as pd
from sklearn.naive_bayes import MultinomialNB
from sklearn.feature_extraction.text import TfidfVectorizer


# * Training data loading
with open('train/in.tsv', 'r', encoding='utf-8') as f:
    x_train = pd.DataFrame([line.strip().split('\t')
                           for line in f.readlines()], columns=['text', 'text_id'])
y_train = pd.read_csv('train/expected.tsv', sep='\t',
                      names=['paranormal'], encoding='utf-8')
# *Validation data loading
with open('dev-0/in.tsv', 'r', encoding='utf-8') as f:
    x_dev = pd.DataFrame([line.strip().split('\t')
                         for line in f.readlines()], columns=['text', 'text_id'])
# * Test data loading
with open('train/in.tsv', 'r', encoding='utf-8') as f:
    x_test = pd.DataFrame([line.strip().split('\t')
                          for line in f.readlines()], columns=['text', 'text_id'])

# * Training data preparation
tfidf_vectorizer = TfidfVectorizer(max_df=0.95, max_features=500)
x_train_vectorized = tfidf_vectorizer.fit_transform(
    x_train['text'].values)

# * Model training
mnb_model = MultinomialNB().fit(x_train_vectorized, y_train.values.ravel())

# * Validation data preparation
x_dev_prepared = tfidf_vectorizer.transform(x_dev['text'].values)
# * Validation data predictions
predictions = mnb_model.predict(x_dev_prepared)
# * Validation predicitons saving
with open('dev-0/out.tsv', 'w') as f:
    for pred in predictions:
        f.write(f'{pred}\n')

# * Test data preparation
x_test_vectorized = tfidf_vectorizer.transform(x_test['text'].values)
# * Test data predictions
predictions = mnb_model.predict(x_test_vectorized)
# * Test predictions saving
with open('test-A/out.tsv', 'w') as f:
    for pred in predictions:
        f.write(f'{pred}\n')