from re import L
import pandas as pd
import numpy as np
import csv
from sklearn.linear_model import LinearRegression
from stop_words import get_stop_words
from sklearn.feature_extraction.text import TfidfVectorizer
 
def linear_regression():

    # odczyt z plików
    colnames_train = ['start_date', 'end_date', 'title', 'sort_title', 'data']
    colnames_test = ['data']

    train = pd.read_csv("train/train.tsv", names = colnames_train, sep = "\t")
    dev_0 = pd.read_csv("dev-0/in.tsv", error_bad_lines = False, header = None, sep = "\t", quoting=csv.QUOTE_NONE)
    dev_1 = pd.read_csv("dev-1/in.tsv", error_bad_lines = False, header = None, sep = "\t", quoting=csv.QUOTE_NONE,)
    test = pd.read_csv("test-A/in.tsv", names = colnames_test, sep = "\t")

    # stworzenie instancji TFIDF i regresji liniowej
    tf = TfidfVectorizer(stop_words=get_stop_words('polish'))
    lin_reg = LinearRegression()

    # wydobycie daty
    date = (train['start_date'] + train['end_date']) / 2

    # regresja liniowa
    train_vec = tf.fit_transform(train['data'])
    lin_reg.fit(train_vec, date)

    # predykcja dla dev-0
    evaluate_dev = tf.transform(dev_0['data'])
    prediction_dev = lin_reg.predict(evaluate_dev)
    pd.DataFrame(prediction_dev).to_csv('dev-0/out2.tsv', sep = "\t", index = False, header = False)

    # predykcja dla test-A
    evaluate_test = tf.transform(test['data'])
    prediction_test = lin_reg.predict(evaluate_test)
    pd.DataFrame(prediction_test).to_csv('test-A/out2.tsv', sep = "\t", index = False, header = False)


    return  None

 
if __name__ == "__main__":
    linear_regression()

# geval: 21.80