import matplotlib.pyplot as plt 
import pandas as pd
import kaggle
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler

def download_file():
    kaggle.api.authenticate()
    kaggle.api.dataset_download_files('nasa/meteorite-landings', path='.', unzip=True)
    

def split(data):
    meteorite_train, meteorite_test = train_test_split(data, test_size=0.2, random_state=1)
    meteorite_train, meteorite_val = train_test_split(meteorite_train, test_size=0.25, random_state=1)
    return meteorite_train, meteorite_test, meteorite_val

def normalization(data):
    scaler = StandardScaler()
    data['mass'] = scaler.fit_transform(data[['mass']])
    return data

def preprocessing(data):
    data = data.dropna(subset=['reclat'])

    incorrect_years_index  = data.loc[(data['year'] > 2016) | (data['year'] < 860)].index
    incorrect_location_index  = data.loc[(data['reclat'] == 0) & (data['reclong'] == 0)].index

    data.drop(incorrect_years_index.union(incorrect_location_index), inplace=True)
    data.loc[(data['mass'].isnull()) & (data['name'].str.startswith('Österplana')), 'mass'] = 0
    return data

download_file()
data = pd.read_csv("meteorite-landings.csv")
meteorite_train, meteorite_test, meteorite_val = split(data)

meteorite_train = normalization(meteorite_train)
meteorite_test = normalization(meteorite_test)
meteorite_val = normalization(meteorite_val)

meteorite_train = normalization(meteorite_train)
meteorite_test = normalization(meteorite_test)
meteorite_val = normalization(meteorite_val)

meteorite_train.to_csv('meteorite_train.csv', encoding='utf-8')
meteorite_test.to_csv('meteorite_test.csv', encoding='utf-8')
meteorite_val.to_csv('meteorite_val.csv', encoding='utf-8')