import numpy as np
import pandas as pd


# get immonet data
def get_immonet_data(path_pre, include_chemnitz):
    immonet_data = pd.read_csv(path_pre + "Data/RealEstate/immonet_data_ffm.csv")
    immonet_data = immonet_data.append(pd.read_csv(path_pre + "Data/RealEstate/immonet_data_koeln.csv"))
    immonet_data = immonet_data.append(pd.read_csv(path_pre + "Data/RealEstate/immonet_data_muenchen.csv"))
    immonet_data = immonet_data.append(pd.read_csv(path_pre + "Data/RealEstate/immonet_data_hamburg.csv"))
    immonet_data = immonet_data.append(pd.read_csv(path_pre + "Data/RealEstate/immonet_data_stuttgart.csv"))
    immonet_data = immonet_data.append(pd.read_csv(path_pre + "Data/RealEstate/immonet_data_duesseldorf.csv"))
    immonet_data = immonet_data.append(pd.read_csv(path_pre + "Data/RealEstate/immonet_data_berlin.csv"))
    if include_chemnitz:
        immonet_data = immonet_data.append(pd.read_csv(path_pre + "Data/RealEstate/immonet_data_chemnitz.csv"))
    immonet_data = immonet_data[
        (~immonet_data.n_rooms.isna()) & (~immonet_data.sq_meters.isna()) & (~immonet_data.price.isna())
        & (~immonet_data.year_built.isna()) & (~immonet_data.year_built.isna())]\
        .reset_index(drop=True)
    return immonet_data


# print out price statistics
def print_descriptive_statistics(model_data):
    print("City\tAverage\tSt.deviation\t0.25 quantile\tMedian\t0.75 quantile")
    for city in ['Frankfurt am Main', "Muenchen"]:  # "Koeln",
        focal_data = model_data[model_data.ort == city]
        print(
            f"{city}\t{len(focal_data)}\t{round(np.average(focal_data.price), 2)}\t{round(np.std(focal_data.price), 2)}"
            f"\t{round(np.quantile(focal_data.price, 0.25), 2)}\t{round(np.quantile(focal_data.price, 0.5), 2)}\t{round(np.quantile(focal_data.price, 0.75), 2)}")


# find subset for experiment
def find_subset(model_data, year_range, floor_range, room_range, unemployment=1):
    mask = (~model_data.garden) & model_data.elevator & model_data.basement &\
           (model_data["floor (storey)"].isin(floor_range)) & (model_data['nmbr of rooms'].isin(room_range)) & \
           (model_data['construction year'].isin(year_range)) & (model_data['unemployment'] == unemployment)
    print(f"Instances in subset: {np.sum(mask)}")
    mask_data = model_data[mask].copy(deep=True)

    # print availabilities
    for stadt in ["Frankfurt", "Hamburg", "Koeln", "Berlin", "Duesseldorf", "Stuttgart", "Muenchen", "Chemnitz"]:
        try:
            mask_data_stadt = mask_data[mask_data[stadt] == 1]
        except KeyError:
            print(f"{stadt} not included.")
            continue
        mask_data_stadt_by, mask_data_stadt_bn = \
            mask_data_stadt[mask_data_stadt["balcony"] == 1], mask_data_stadt[mask_data_stadt["balcony"] == 0]
        print(f"{stadt}: Balcony yes: {(mask_data_stadt_by['Anteil Gruenenwaehler'].unique())}")
        print(f"{stadt}: Balcony no: {(mask_data_stadt_bn['Anteil Gruenenwaehler'].unique())}")
    return mask_data.copy(deep=True)


# zip preprocess zip mapping
def get_zip_mapping_for_city(city, zip_mapping):
    city_list, district, unemployment, zips, green_share = [], [], [], [], []
    for row in zip_mapping.itertuples():
        for plz in row[2].split(";"):
            city_list.append(city)
            district.append(row[1])
            unemployment.append(row[3])
            green_share.append(row[4])
            zips.append(int(plz))
    thres_1 = np.nanquantile(unemployment, 0.3333)
    thres_2 = np.nanquantile(unemployment, 0.6666)
    unemployment = [1 if u <= thres_1 else 3 if u >= thres_2 else 2 for u in unemployment]
    thres_1 = np.nanquantile(green_share, 0.3333)
    thres_2 = np.nanquantile(green_share, 0.6666)
    green_share = [1 if u <= thres_1 else 3 if u >= thres_2 else 2 for u in green_share]
    zip_mapping_complete = pd.DataFrame(list(zip(city_list, district, zips, unemployment, green_share)),
                                        columns=['city', 'district', 'zips', 'unemployment', 'green_share'])
    return zip_mapping_complete


# get zip mapping for complete data using "get_zip_mapping_for_city"
def get_zip_mapping(path_pre):
    muc_zip_map = pd.read_csv(path_pre + "Data/WebScraper/plz_mapping_muenchen.csv")
    muc_zip_map["ArbQuote"] = muc_zip_map["Arbeitslose"] / muc_zip_map["Einwohner"]
    muc_zip_map = muc_zip_map[['Stadtteil', 'PLZ', 'ArbQuote', 'GreenShare21']]
    zip_mapping = get_zip_mapping_for_city("Muenchen", muc_zip_map)
    zip_mapping = zip_mapping.append(
        get_zip_mapping_for_city("Koeln", pd.read_csv(path_pre + "Data/WebScraper/plz_mapping_koeln.csv")))
    zip_mapping = zip_mapping.append(
        get_zip_mapping_for_city("Frankfurt", pd.read_csv(path_pre + "Data/WebScraper/plz_mapping_ffm.csv")))
    zip_mapping = zip_mapping.append(
        get_zip_mapping_for_city("Hamburg", pd.read_csv(path_pre + "Data/WebScraper/plz_mapping_hamburg.csv")))
    zip_mapping = zip_mapping.append(
        get_zip_mapping_for_city("Stuttgart", pd.read_csv(path_pre + "Data/WebScraper/plz_mapping_stuttgart.csv")))
    zip_mapping = zip_mapping.append(
        get_zip_mapping_for_city("Duesseldorf", pd.read_csv(path_pre + "Data/WebScraper/plz_mapping_duesseldorf.csv")))
    zip_mapping = zip_mapping.append(
        get_zip_mapping_for_city("Berlin", pd.read_csv(path_pre + "Data/WebScraper/plz_mapping_berlin.csv", encoding='latin1')))
    zip_mapping = zip_mapping.append(
        get_zip_mapping_for_city("Chemnitz", pd.read_csv(path_pre + "Data/WebScraper/plz_mapping_chemnitz.csv")))
    return zip_mapping


# get zip based features (green voters / unemployment) by getting zip helper frame to join with immonet data
def get_zip_features(zip_mapping):
    zip_helper = pd.DataFrame(index=zip_mapping.zips.unique())
    zip_helper["unemployment"] = [zip_mapping[zip_mapping.zips == z]["unemployment"].mode()[0] for z in zip_helper.index]
    zip_helper["green_share"] = [zip_mapping[zip_mapping.zips == z]["green_share"].mode()[0] for z in zip_helper.index]
    return zip_helper


# uses names without tex unfriendly symbols (e.g., '_')
def rename_house_features(feature_names):
    rename_c = {"n_rooms": "nmbr of rooms", "floor_n": "floor (storey)", "year_built": "construction year",
                "ort_Frankfurt am Main": "Frankfurt", "ort_Muenchen": "Muenchen",  "ort_Koeln": "Koeln",
                "ort_Hamburg": "Hamburg", "ort_Stuttgart": "Stuttgart",
                "ort_Duesseldorf": "Duesseldorf", "ort_Berlin": "Berlin",
                "ort_Chemnitz": "Chemnitz", "green_share": "Anteil Gruenenwaehler"}
    return [rename_c[c] if c in rename_c else c for c in feature_names]


# get features for prediction in stages 1, 2, 3
def get_12_var_comb_for_cities(city_1, city_2, include_chemnitz=False):
    other_cities = ['Berlin', 'Duesseldorf', 'Frankfurt', 'Hamburg', 'Koeln', 'Muenchen', 'Stuttgart']
    if include_chemnitz:
        other_cities.append("Chemnitz")
    other_cities.remove(city_1)
    other_cities.remove(city_2)

    var_prop_combinations = pd.DataFrame()
    garden, basement, elevator, balcony, floor, nmbr_of_rooms, construction_year, unemployment, \
    Anteil_Gruenenwaehler, city_val_1, city_val_2 = \
        [], [], [], [], [], [], [], [], [], [], []
    for j in range(1, 13):
        # fixed properties
        garden.append(False), basement.append(True), elevator.append(True), floor.append(1), \
        nmbr_of_rooms.append(3), construction_year.append(9), unemployment.append(1)

        # variable properties
        Anteil_Gruenenwaehler.append(int((j + 3) / 4))
        city_val_1.append(j % 2)
        city_val_2.append((j + 1) % 2)
        balcony.append(int(((j + 1) / 2)) % 2)

    var_prop_combinations["garden"], var_prop_combinations["basement"], var_prop_combinations["elevator"], \
    var_prop_combinations["balcony"], var_prop_combinations["floor (storey)"], var_prop_combinations[
        "nmbr of rooms"], \
    var_prop_combinations["construction year"], var_prop_combinations["unemployment"], \
    var_prop_combinations["Anteil Gruenenwaehler"], var_prop_combinations[city_1], var_prop_combinations[city_2] \
        = garden, basement, elevator, balcony, floor, nmbr_of_rooms, construction_year, unemployment, \
          Anteil_Gruenenwaehler, city_val_1, city_val_2
    for city in other_cities:
        var_prop_combinations[city] = 0

    return var_prop_combinations


# get features for average price of category A cities in Germany (for stages 1, 3)
def get_comb_for_cat_A_cities():
    cities = ['Berlin', 'Duesseldorf', 'Frankfurt', 'Hamburg', 'Koeln', 'Muenchen', 'Stuttgart']

    var_prop_combinations = pd.DataFrame()
    garden, basement, elevator, balcony, floor, nmbr_of_rooms, construction_year, unemployment, \
    Anteil_Gruenenwaehler, city = \
        [], [], [], [], [], [], [], [], [], []
    for j in range(1, 43):
        # fixed properties
        garden.append(False), basement.append(True), elevator.append(True), floor.append(1), \
        nmbr_of_rooms.append(3), construction_year.append(9), unemployment.append(1)

        # variable properties
        Anteil_Gruenenwaehler.append((int((j-1)/2) % 3)+1)
        city.append(cities[int((j-1)/6)])
        balcony.append(j % 2)

    var_prop_combinations["garden"], var_prop_combinations["basement"], var_prop_combinations["elevator"], \
    var_prop_combinations["balcony"], var_prop_combinations["floor (storey)"], var_prop_combinations[
        "nmbr of rooms"], \
    var_prop_combinations["construction year"], var_prop_combinations["unemployment"], \
    var_prop_combinations["Anteil Gruenenwaehler"], var_prop_combinations["city"] \
        = garden, basement, elevator, balcony, floor, nmbr_of_rooms, construction_year, unemployment, \
          Anteil_Gruenenwaehler, city
    var_prop_combinations = pd.get_dummies(var_prop_combinations, columns=['city'])
    var_prop_combinations.columns = [col.replace("city_", "") for col in var_prop_combinations.columns]

    return var_prop_combinations