timeseries_shaper.filter.datetime_filter

  1import pandas as pd
  2from ..base import Base
  3
  4class DateTimeFilter(Base):
  5    """
  6    Provides class methods for filtering time columns in a pandas DataFrame.
  7    Allows specification of which column to operate on.
  8    
  9    Inherits from:
 10        Base (class): Base class with common initializations for DataFrame handling.
 11    """
 12
 13    @classmethod
 14    def filter_after_date(cls, dataframe: pd.DataFrame, column_name: str = 'systime', date: str = None) -> pd.DataFrame:
 15        """
 16        Filters the DataFrame to include only rows after the specified date.
 17
 18        Args:
 19        - date (str): The cutoff date in 'YYYY-MM-DD' format.
 20
 21        Returns:
 22        - pd.DataFrame: A DataFrame containing rows where the 'systime' is after the specified date.
 23
 24        Example:
 25        --------
 26        >>> filtered_data = DateTimeFilter.filter_after_date(df, "systime", "2023-01-01")
 27        >>> print(filtered_data)
 28        """
 29        return dataframe[dataframe[column_name] > pd.to_datetime(date)]
 30
 31    @classmethod
 32    def filter_before_date(cls, dataframe: pd.DataFrame, column_name: str = 'systime', date: str = None) -> pd.DataFrame:
 33        """
 34        Filters the DataFrame to include only rows before the specified date.
 35
 36        Args:
 37        - date (str): The cutoff date in 'YYYY-MM-DD' format.
 38
 39        Returns:
 40        - pd.DataFrame: A DataFrame containing rows where the 'systime' is before the specified date.
 41
 42        Example:
 43        --------
 44        >>> filtered_data = DateTimeFilter.filter_before_date(df, "systime", "2023-01-01")
 45        >>> print(filtered_data)
 46        """
 47        return dataframe[dataframe[column_name] < pd.to_datetime(date)]
 48
 49    @classmethod
 50    def filter_between_dates(cls, dataframe: pd.DataFrame, column_name: str = 'systime', start_date: str = None, end_date: str = None) -> pd.DataFrame:
 51        """
 52        Filters the DataFrame to include only rows between the specified start and end dates.
 53
 54        Args:
 55        - start_date (str): The start date of the interval in 'YYYY-MM-DD' format.
 56        - end_date (str): The end date of the interval in 'YYYY-MM-DD' format.
 57
 58        Returns:
 59        - pd.DataFrame: A DataFrame containing rows where the 'systime' is between the specified dates.
 60
 61        Example:
 62        --------
 63        >>> filtered_data = DateTimeFilter.filter_between_dates(df, "systime", "2023-01-01", "2023-02-01")
 64        >>> print(filtered_data)
 65        """
 66        mask = (dataframe[column_name] > pd.to_datetime(start_date)) & (dataframe[column_name] < pd.to_datetime(end_date))
 67        return dataframe[mask]
 68
 69    @classmethod
 70    def filter_after_datetime(cls, dataframe: pd.DataFrame, column_name: str = 'systime', datetime: str = None) -> pd.DataFrame:
 71        """
 72        Filters the DataFrame to include only rows after the specified datetime.
 73
 74        Args:
 75        - datetime (str): The cutoff datetime in 'YYYY-MM-DD HH:MM:SS' format.
 76
 77        Returns:
 78        - pd.DataFrame: A DataFrame containing rows where the 'systime' is after the specified datetime.
 79
 80        Example:
 81        --------
 82        >>> filtered_data = DateTimeFilter.filter_after_datetime(df, "systime", "2023-01-01 12:00:00")
 83        >>> print(filtered_data)
 84        """
 85        return dataframe[dataframe[column_name] > pd.to_datetime(datetime)]
 86
 87    @classmethod
 88    def filter_before_datetime(cls, dataframe: pd.DataFrame, column_name: str = 'systime', datetime: str = None) -> pd.DataFrame:
 89        """
 90        Filters the DataFrame to include only rows before the specified datetime.
 91
 92        Args:
 93        - datetime (str): The cutoff datetime in 'YYYY-MM-DD HH:MM:SS' format.
 94
 95        Returns:
 96        - pd.DataFrame: A DataFrame containing rows where the 'systime' is before the specified datetime.
 97
 98        Example:
 99        --------
100        >>> filtered_data = DateTimeFilter.filter_before_datetime(df, "systime", "2023-01-01 12:00:00")
101        >>> print(filtered_data)
102        """
103        return dataframe[dataframe[column_name] < pd.to_datetime(datetime)]
104
105    @classmethod
106    def filter_between_datetimes(cls, dataframe: pd.DataFrame, column_name: str = 'systime', start_datetime: str = None, end_datetime: str = None) -> pd.DataFrame:
107        """
108        Filters the DataFrame to include only rows between the specified start and end datetimes.
109
110        Args:
111        - start_datetime (str): The start datetime of the interval in 'YYYY-MM-DD HH:MM:SS' format.
112        - end_datetime (str): The end datetime of the interval in 'YYYY-MM-DD HH:MM:SS' format.
113
114        Returns:
115        - pd.DataFrame: A DataFrame containing rows where the 'systime' is between the specified datetimes.
116
117        Example:
118        --------
119        >>> filtered_data = DateTimeFilter.filter_between_datetimes(df, "systime", "2023-01-01 12:00:00", "2023-02-01 12:00:00")
120        >>> print(filtered_data)
121        """
122        mask = (dataframe[column_name] > pd.to_datetime(start_datetime)) & (dataframe[column_name] < pd.to_datetime(end_datetime))
123        return dataframe[mask]