from datetime import datetime

import pandas as pd


def convert_date(date_str):
    cut_index = str(date_str).rfind("_")
    date = date_str[0:cut_index].replace("_", "-")
    time = date_str[cut_index + 1:].replace(":", ".")

    return datetime.strptime(f"{date} {time}", '%Y-%m-%d %H.%M.%S.%f')


df = pd.read_csv(r"d:/data/b2_240828_2324_Err 1.csv", header=1)
df.dropna(subset='TimeStamp', inplace=True)
df.drop_duplicates(subset='TimeStamp', keep="first", inplace=True)

origin_columns = list(df.columns)

df['TimeStamp1'] = df['TimeStamp'].apply(convert_date)
df.sort_values(by='TimeStamp1', inplace=True)

# df['DateTime'] = pd.to_datetime(df['TimeStamp'], format="%Y-%m-%d %H:%M:%S")
df['DateTime'] = df['TimeStamp1'].apply(lambda x: x.strftime("%Y-%m-%d %H:%M:%S"))

print(df.shape)

dateTime_count = df['DateTime'].value_counts()

dateTime_count_1 = dateTime_count[dateTime_count == 1]
dateTime_count_gt1 = dateTime_count[dateTime_count > 1]

df1 = df[df['DateTime'].isin(dateTime_count_1.index.values)]
df2 = df[df['DateTime'].isin(dateTime_count_gt1.index.values)]

print(df1.shape)
print(df2.shape)
origin_columns.insert(0, 'DateTime')
df1.to_csv("1秒数据.csv", encoding='utf-8', index=False, columns=origin_columns, date_format="%Y-%m-%d %H:%M:%S.%f")
df2.to_csv("毫秒数据.csv", encoding='utf-8', index=False, columns=origin_columns, date_format="%Y-%m-%d %H:%M:%S.%f")