import pandas as pd
from tqdm import tqdm_notebook as tqdm
from sklearn.metrics import accuracy_score

df_train = pd.read_csv("train_data.csv")
df_test = pd.read_csv("test_data.csv")

df_train.drop("y", axis=1, inplace=True)
df_train.rename(columns={"id": "ID"}, inplace=True)
df = pd.concat([df_train, df_test])
df = df.reset_index()

query_str1 = "X1 == @X1 and X2 == @X2 and X3== @X3 and X4 == @X4 and X5 == @X5"
query_str2 = "X6 == @X7 and X7 == @X8 and X8 == @X9 and X9 == @X10 and X10 == @X11"
query_str3 = "X12 == @X13 and X13 == @X14 and X14 == @X15 and X15 == @X16 and  X16 == @X17"
query_str4 = "X18 == @X19 and X19 == @X20 and X20 == @X21 and X21 == @X22 and  X22 == @X23"

leak_id = []
leak_val = []
for ind in tqdm(df.index):
    X1 = df.loc[ind, "X1"]
    X2 = df.loc[ind, "X2"]
    X3 = df.loc[ind, "X3"]
    X4 = df.loc[ind, "X4"]
    X5 = df.loc[ind, "X5"]
    X6 = df.loc[ind, "X6"]
    X7 = df.loc[ind, "X7"]
    X8 = df.loc[ind, "X8"]
    X9 = df.loc[ind, "X9"]
    X10 = df.loc[ind, "X10"]
    X11 = df.loc[ind, "X11"]
    X12 = df.loc[ind, "X12"]
    X13 = df.loc[ind, "X13"]
    X14 = df.loc[ind, "X14"]
    X15 = df.loc[ind, "X15"]
    X16 = df.loc[ind, "X16"]
    X17 = df.loc[ind, "X17"]
    X18 = df.loc[ind, "X18"]
    X19 = df.loc[ind, "X19"]
    X20 = df.loc[ind, "X20"]
    X21 = df.loc[ind, "X21"]
    X22 = df.loc[ind, "X22"]
    X23 = df.loc[ind, "X23"]
    
    # 新規顧客, ほぼ新規顧客は飛ばす
    if X6+X7+X8+X9+X10+X11 == -12 and X12+X13+X14+X15+X16+X17+X18+X19+X20+X21+X22+X23 == 0:
        continue
    if X6+X7+X8+X9+X10+X11 == -9 and X12+X13+X14+X15+X16+X17+X18+X19+X20+X21+X22+X23 == 0:
        continue

    queried1 = df.query(query_str1)
    if len(queried1) == 0:
        continue

    queried2 = queried1.query(query_str2)
    if len(queried2) == 0:
        continue

    queried3 = queried2.query(query_str3)
    if len(queried3) == 0:
        continue

    queried4 = queried3.query(query_str4)
    if len(queried4) != 1:
        continue

    if ind != queried4.index[0]:
        leak_id.append(queried4.index[0])
        leak_id.append(ind)

df.loc[leak_id, :]