ebtools 0.1.0__py3-none-any.whl

ebtools/data/ukclockchangedays.py

@@ -0,0 +1,110 @@
+# -*- coding: utf-8 -*-
+"""
+Created on Tue Apr  4 17:36:46 2023
+
+@author: Eamonn.Bell
+"""
+
+
+# =============================================================================
+# ---- Clock Change days - DAYLIGHT SAVINGS
+# =============================================================================
+
+# Maintain a list of clock change days here.
+# The official place to look for clock change days is here:
+# https://www.gov.uk/when-do-the-clocks-change
+#
+# The Daylight Savings Clock Change dates and times can be found as part of 
+# the Market Domain Data (MDD) on the Elexon portal page here (may have to be
+# logged in):
+# www.elexonportal.co.uk/mddviewer
+#  
+# An alternative source can be found here:
+# https://www.timeanddate.com/time/change/uk/london
+#
+# Clock change dates should be checked against the gov.uk site on a regular
+# basis and then set moved to the 'Confirmed' section.
+
+
+# LAST SUNDAY IN MARCH - These days will have 46 SPs
+uk_clock_change_forward  = [
+    
+    # Confirmed
+    '1998-03-29',
+    '1999-03-28',
+    '2000-03-26',
+    '2001-03-25',
+    '2002-03-31',
+    '2003-03-30',
+    '2004-03-28',
+    '2005-03-27',
+    '2006-03-26',
+    '2007-03-25',
+    '2008-03-30',
+    '2009-03-29',
+    '2010-03-28',
+    '2011-03-27',
+    '2012-03-25',
+    '2013-03-31',
+    '2014-03-30',
+    '2015-03-29',
+    '2016-03-27',
+    '2017-03-26',
+    '2018-03-25',
+    '2019-03-31',
+    '2020-03-29',
+    '2021-03-28',
+    '2022-03-27',
+    '2023-03-26',
+    '2024-03-31',
+    '2025-03-30',
+    '2026-03-29',
+    
+    # Unconfirmed
+    '2027-03-28',
+    '2028-03-26',
+    '2029-03-25',    
+    '2030-03-31'
+    ]
+
+# LAST SUNDAY IN OCTOBER - These days will have 50 SPs
+uk_clock_change_backward = [
+
+    # Confirmed
+    '1998-10-25',
+    '1999-10-31',
+    '2000-10-29',
+    '2001-10-28',
+    '2002-10-27',
+    '2003-10-26',
+    '2004-10-31',
+    '2005-10-30',
+    '2006-10-29',
+    '2007-10-28',
+    '2008-10-26',
+    '2009-10-25',
+    '2010-10-31',
+    '2011-10-30',
+    '2012-10-28',
+    '2013-10-27',
+    '2014-10-26',
+    '2015-10-25',
+    '2016-10-30',
+    '2017-10-29',
+    '2018-10-28',
+    '2019-10-27',
+    '2020-10-25',
+    '2021-10-31',
+    '2022-10-30',
+    '2023-10-29',
+    '2024-10-27',
+    '2025-10-26',
+    '2026-10-25',
+    
+    # Unconfirmed
+    '2027-10-31',
+    '2028-10-29',
+    '2029-10-28',
+    '2030-10-27' 
+    ]
+

ebtools/data/units.py

@@ -0,0 +1,22 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Created on Sun Dec 31 13:52:29 2023
+
+@author: eamonnbell
+"""
+
+
+# Dictionary to reference strings to the number of potential periods per
+# day in a dataset. Add here if there are more.
+
+unit_dict = {
+    '5min' : 288,
+    '15min': 96,
+    '30min': 48,
+    '60min': 24,
+    'hh'   : 48,
+    'hour' : 24,
+    'efa'  : 6
+    }
+

ebtools/general/__init__.py

@@ -0,0 +1,260 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""
+Created on Tue Oct 24 21:12:28 2023
+
+@author: eamonnbell
+"""
+
+
+from ebtools.general.helper import (
+    
+    # Dict Tools
+    sort_dict_by_keys,
+    make_dict_keys_from_list,
+    
+    # Force to Type
+    force_to_list,
+    force_to_np_array
+
+    )
+
+
+from ebtools.general.read_data import (
+    
+    # Reading different file_types into dataframes
+    read_from_csv,
+    read_from_xlsx,
+    read_from_parquet,
+    
+    # Reading HH Data into a dataframe
+    load_hh_data
+    
+    )
+
+
+from ebtools.general.save_data import (
+    
+    save_to_csv,
+    save_to_xlsx,
+    save_to_parquet
+    
+    )
+
+
+from ebtools.general.inspect_funcs import (
+    
+    get_call_func_name,
+    get_function_arguments
+    
+    )
+
+
+from ebtools.general.file_reading_tools import (
+    
+    does_file_exist,
+    get_file_type,
+    list_all_files_in_folder,
+    zip_checker
+    
+    )
+
+
+from ebtools.general.datetime_calendar import (
+    
+    # Helper funcs
+    is_year_leap_year,
+    get_daylight_savings_dates,
+    bst_date_range,
+    )
+
+from ebtools.general.datetime_conversion import (
+
+    # Helper funcs
+    convert_single_datetime_to_sp_start,
+    convert_single_datetime_to_sp_end,
+    
+    # Convert Datetime
+    convert_date_to_standard_date_str,
+    convert_date_to_datetime,
+    convert_datetime_to_date,
+    convert_date_to_datetime_str,
+    convert_date_to_first_of_month,
+    convert_date_to_last_of_month,
+    
+    # Month-start and Month-end
+    move_date_to_start_of_month,
+    move_date_to_end_of_month
+    
+    )
+
+from ebtools.general.datetime_timezone import (
+
+    # Convert timezone-aware datetime columns
+    convert_datetime_remove_tz_aware,
+    convert_datetime_by_tz,
+
+    )
+
+from ebtools.general.datetime_dataframe import (
+
+    # Add dataframe datetime columns
+    convert_datetime_sp_start,
+    convert_datetime_sp_end,
+    add_first_of_month_col_from_int_cols,
+    add_first_of_month_col_from_date_col,
+
+    )
+
+from ebtools.general.datetime_parsing import (
+    
+    # Check Datetime
+    check_datetime_formats,
+    check_datetime_formats_tz,
+    check_string_datetime_formats,
+
+    )
+
+from ebtools.general.datetime_adjustments import (
+    
+    # Adjust Dates
+    start_of_delivery_year,
+    roll_back_by_month_str,
+    roll_back_by_month_df,
+    
+    )
+
+
+from ebtools.general.sp_efa_conversion import (
+    
+    # General
+    sp_to_efa,
+    hour_to_efa,
+    current_sp,
+    current_efa,
+    make_sp_efa_table,
+
+    )
+
+from ebtools.general.sp_efa_datetime import (
+    
+    # EFA date and datetime
+    convert_date_to_efa_datetime,
+    convert_datetime_to_date_and_efa,
+
+    )
+
+from ebtools.general.sp_efa_strings import (
+    
+    # SP and EFA from strings
+    make_sp_from_string,
+    make_efa_from_string,
+
+    )
+
+from ebtools.general.sp_efa_dataframe import (
+    
+    # Add cols to df
+    add_sp_from_datetime_col,
+    add_sp_start_end_datetimes,
+    add_sp_to_efa_column,
+    add_efa_datetime_col,
+    add_date_efa_cols,
+    add_sp_and_efa_from_datetime
+    
+    )
+
+
+from ebtools.general.dataframe_dates import (
+    
+    # Dataframe - Augmentation
+    add_buffer_days,
+    add_columns,
+    )
+
+from ebtools.general.dataframe_selection import (
+
+    # Dataframe - Date Range Selection
+    select_year,
+    select_weeks,
+    select_days,
+    )
+
+from ebtools.general.dataframe_plotting import (
+
+    # Dataframe - Analysis and Plotting
+    make_full_year_pivot,
+    make_full_plotting_df,
+    make_pivot_calendar,
+    )
+
+from ebtools.general.dataframe_base import (
+
+    # Dataframe - BASE
+    adjust_for_clock_change,
+    base_df,
+    )
+
+from ebtools.general.dataframe_filtering import (
+
+    # Dataframe - Filtering
+    df_filter,
+    df_cutoff,
+    drop_empty_df_cols,
+    melt_df,
+    )
+
+from ebtools.general.dataframe_processing import (
+
+    # Dataframe - Processing
+    unique_element_count,
+    non_numeric_pivot,
+    df_round_to_decimal,
+    )
+
+from ebtools.general.dataframe_formatting import (
+
+    # Dataframe - Formatting
+    remove_df_index_name,
+    remove_df_column_name,
+    )
+
+from ebtools.general.dataframe_conversion import (
+
+    # Dataframe - Correct Strings
+    correct_str_elements_float,
+    correct_str_elements_int,
+
+    # Dataframe - Create df from JSON
+    create_df_from_json,
+    create_df_from_json_from_file
+    )
+
+
+from ebtools.general.df_tools_string_dates import (
+    
+    # STRING date functions
+    week_starting,
+    week_number,
+    weekday_number,
+    day_of_the_year_number
+    
+    )
+
+from ebtools.general.array_selection import (
+    
+    # General
+    add_newaxis,
+    
+    # Select data from ROWS
+    n_smallest_values_per_row,
+    n_largest_values_per_row
+    
+    )
+
+
+from ebtools.general.online import (
+
+    is_online
+    
+    )

ebtools/general/_sp_efa_helpers.py

@@ -0,0 +1,89 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+"""Private helpers for Settlement Period and EFA conversions."""
+
+import datetime
+
+import numpy as np
+import pandas as pd
+
+import ebtools.general.helper as helper
+
+
+EFA_START_HOURS = {
+    1: -1,
+    2: 3,
+    3: 7,
+    4: 11,
+    5: 15,
+    6: 19,
+}
+
+TIME_TO_EFA = {
+    datetime.time(23, 0): 1,
+    datetime.time(3, 0): 2,
+    datetime.time(7, 0): 3,
+    datetime.time(11, 0): 4,
+    datetime.time(15, 0): 5,
+    datetime.time(19, 0): 6,
+}
+
+
+def as_integer_array(values: object, name: str) -> np.ndarray:
+    """Return values as an integer NumPy array, preserving integer semantics."""
+    value_array = helper.force_to_np_array(values)
+
+    try:
+        value_int = value_array.astype(int)
+    except (TypeError, ValueError):
+        raise ValueError(f"{name} values must be integer-like.") from None
+
+    try:
+        integer_like = np.all(np.equal(value_array, value_int))
+    except TypeError:
+        integer_like = False
+
+    if not integer_like:
+        raise ValueError(f"{name} values must be integer-like.")
+
+    return value_int
+
+
+def return_scalar_or_array(
+        values: np.ndarray,
+        scalar_input: bool
+        ) -> int | np.ndarray:
+    """Return a scalar integer for scalar input, otherwise return an array."""
+    if scalar_input:
+        return int(values[0])
+
+    return values
+
+
+def require_columns(
+        df: pd.DataFrame,
+        columns: list[str]
+        ) -> None:
+    """Raise a KeyError if any expected DataFrame columns are missing."""
+    missing_cols = [col for col in columns if col not in df.columns]
+
+    if missing_cols:
+        raise KeyError(f"Column(s) not found in dataframe: {missing_cols}")
+
+
+def validate_range(
+        values: np.ndarray | pd.Series,
+        name: str,
+        min_value: int,
+        max_value: int
+        ) -> None:
+    """Raise a ValueError if values fall outside an inclusive range."""
+    if np.any((values < min_value) | (values > max_value)):
+        raise ValueError(
+            f"{name} values must be in the range {min_value} to {max_value}."
+        )
+
+
+def validate_sp_values(values: np.ndarray | pd.Series) -> None:
+    """Validate that values are valid Settlement Period numbers."""
+    validate_range(values, "SP", 1, 48)