import pandas
import os
import sys
from sklearn.linear_model import LinearRegression

IN_FILE_NAME = "in.tsv"
OUT_FILE_NAME = "out.tsv"
MAIN_DIR = "."
NAMES_FILE_NAME = "names"
TRAIN_DIR = "train"
TRAIN_FILE_NAME = "train.tsv"
VALUE_SEP = "\t"
LINE_SEP = "\n"
CATEGORY_TYPE = "category"


def main(dirname: str):
    names = get_names()

    X, Y = get_train_data(names)

    clf = LinearRegression().fit(X, Y)

    clf.predict(get_input_data(dirname, names)).tofile(os.path.join(
        dirname, OUT_FILE_NAME), sep=LINE_SEP)


def get_train_data(names: list):
    train_path = os.path.join(MAIN_DIR, TRAIN_DIR, TRAIN_FILE_NAME)
    check_file(train_path)
    train_data = process_input(pandas.read_csv(
        train_path, header=None, sep=VALUE_SEP, names=names))

    train_data = train_data.loc[(train_data[names[0]] > 1000)]

    X = train_data.loc[:, train_data.columns != names[0]]
    Y = train_data[names[0]]

    return X, Y


def get_input_data(dirname, names):
    in_path = os.path.join(dirname, IN_FILE_NAME)
    check_file(in_path)
    return process_input(pandas.read_csv(in_path, header=None, sep=VALUE_SEP, names=names[1:]))


def process_input(df):
    for c in df.select_dtypes(include=object).columns.values:
        df[c] = df[c].astype(CATEGORY_TYPE).cat.codes
    return df


def get_names() -> list:
    names_path = os.path.join(MAIN_DIR, NAMES_FILE_NAME)
    check_file(names_path)

    with open(names_path) as f_names:
        return f_names.read().strip().split(VALUE_SEP)


def check_file(filename: str):
    if not os.path.exists(filename):
        raise FileNotFoundError(filename)


if __name__ == "__main__":
    if len(sys.argv) < 2:
        raise Exception("Name of working dir not specified!")
    main(sys.argv[1])