import numpy as np
import pandas as pd
import gzip
from sklearn.naive_bayes import MultinomialNB
from sklearn.feature_extraction.text import CountVectorizer

bayes = MultinomialNB()
vec_q = CountVectorizer()

with gzip.open('./train/train.tsv.gz', 'rb') as f:
    train = pd.read_csv(f, error_bad_lines=False, header=None, sep="\t")
dev = pd.read_csv("./dev-0/in.tsv", error_bad_lines=False, header=None, sep="\t")
test = pd.read_csv("./test-A/in.tsv", error_bad_lines=False, header=None, sep="\t")

# model
X_train, y_train = train[0].astype(str).tolist(), train[1].astype(str).tolist()
y_train=vec_q.fit_transform(y_train)
bayes.fit(y_train, X_train)

# dev 
X_dev = dev[0].astype(str).tolist()
y_dev = vec_q.transform(X_dev)
dev_pred = bayes.predict(y_dev)
pd.DataFrame(dev_pred).to_csv('./dev-0/out.tsv', sep='\t', index=False, header=False)

# test
X_test = train[0].astype(str).tolist()
y_test = vec_q.transform(X_test)
test_pred = bayes.predict(y_test)
pd.DataFrame(test_pred).to_csv('./test-A/out.tsv', sep='\t', index=False, header=False)