import pandas as pd

df = pd.read_csv(r'C:\Users\Administrator\Documents\WeChat Files\anmox-\FileStorage\File\2024-12\26故障.csv',
                 encoding='gbk')
df['开始时间'] = pd.to_datetime(df['开始时间'], errors='coerce')
df['结束时间'] = pd.to_datetime(df['结束时间'], errors='coerce')
time_df = pd.DataFrame(df.groupby(['开始时间'])['结束时间'].max())
time_df.reset_index(inplace=True)
time_df.sort_values(by='开始时间', inplace=True)

datas = set()
max_row = None
for index, row in time_df.iterrows():
    if index == 0:
        datas.add((row['开始时间'], row['结束时间']))
        max_row = row
        continue

    if row['结束时间'] > max_row['结束时间']:
        datas.add((row['开始时间'], row['结束时间']))
        max_row = row

result_df = pd.DataFrame()
for begin, end in datas:
    print(begin, end)
    now_df = df[(df['开始时间'] == begin) & (df['结束时间'] == end)]
    now_df = now_df.tail(1)
    result_df = pd.concat([result_df, now_df])

result_df.sort_values(by='开始时间', inplace=True)
result_df.to_csv(r'd:\data\26故障_new.csv', encoding='utf8', index=False)