import pandas as pd
from sklearn.model_selection import train_test_split


def main():
    data = pd.read_csv('resources/Amazon_Consumer_Reviews.csv', header=0, sep=',')

    columns = ['reviews.date', 'reviews.numHelpful', 'reviews.rating', 'reviews.doRecommend']
    string_columns = ['name', 'brand', 'categories', 'primaryCategories', 'keys', 'manufacturer', 'reviews.title',
                      'reviews.username', 'reviews.text']

    data = data[string_columns + columns]

    for c in string_columns:
        data[c] = data[c].str.lower()

    print("Empty rows summary:")
    print(data.isnull().sum())
    data.dropna()

    data.to_csv('resources/data.csv')

    train, test = train_test_split(data, train_size=0.6, random_state=1)
    test, dev = train_test_split(test, test_size=0.5, random_state=1)
    test.to_csv('resources/test.csv')
    train.to_csv('resources/train.csv')
    dev.to_csv('resources/dev.csv')

    print("\n\nMean reviews rating for each primary category: ")
    print(data[["primaryCategories", "reviews.rating"]].groupby("primaryCategories").mean())

    print("\n\nCounted primary categories: ")
    print(data["primaryCategories"].value_counts())

    print("\n\nGeneral data statistics: ")
    print(data.describe(include='all'))


if __name__ == '__main__':
    main()