sensospot_parser/sensospot_data/dynamic_range.py

from pandas.api.types import is_numeric_dtype

from .utils import split_data_frame
from .columns import (
    RAW_DATA_POS_ID,
    CALC_SPOT_OVERFLOW,
    META_DATA_WELL_ROW,
    RAW_DATA_SPOT_MEAN,
    META_DATA_WELL_COLUMN,
    SETTINGS_EXPOSURE_TIME,
    SETTINGS_EXPOSURE_CHANNEL,
    RAW_DATA_NORMALIZATION_MAP,
    SETTINGS_NORMALIZED_EXPOSURE_TIME,
)

PROBE_MULTI_INDEX = [
    META_DATA_WELL_ROW,
    META_DATA_WELL_COLUMN,
    RAW_DATA_POS_ID,
]


def _check_if_xdr_ready(data_frame):
    """ check if a data frame meets the constraints for xdr """
    required_columns = {SETTINGS_EXPOSURE_CHANNEL, SETTINGS_EXPOSURE_TIME}
    if not required_columns.issubset(data_frame.columns):
        raise ValueError("XDR: Apply an exposure map first")
    if len(data_frame[SETTINGS_EXPOSURE_CHANNEL].unique()) != 1:
        raise ValueError("XDR: Mixed Exposure Channels")
    if not is_numeric_dtype(data_frame[SETTINGS_EXPOSURE_TIME]):
        raise ValueError("XDR: Exposure time is not numerical")
    if data_frame[SETTINGS_EXPOSURE_TIME].hasnans:
        raise ValueError("XDR: Exposure time contains NaNs")


def _calc_overflow_info(data_frame, column=RAW_DATA_SPOT_MEAN, limit=0.5):
    """ add overflow info, based on column and limit """
    data_frame[CALC_SPOT_OVERFLOW] = data_frame[column] > limit
    return data_frame


def _reduce_overflow(data_frame):
    """ the heavy lifting for creating an extended dynamic range """

    split_frames = split_data_frame(data_frame, SETTINGS_EXPOSURE_TIME)

    # get the exposure times, longest first
    exposure_times = sorted(split_frames.keys(), reverse=True)
    max_time, *rest_times = exposure_times

    result_frame = split_frames[max_time].set_index(PROBE_MULTI_INDEX)

    for next_time in rest_times:
        mask = result_frame[CALC_SPOT_OVERFLOW] == True  # noqa: E712
        next_frame = split_frames[next_time].set_index(PROBE_MULTI_INDEX)
        result_frame.loc[mask] = next_frame.loc[mask]

    return result_frame.reset_index()


def blend(data_frame, column=RAW_DATA_SPOT_MEAN, limit=0.5):
    """ creates an extended dynamic range, eliminating overflowing spots """
    _check_if_xdr_ready(data_frame)
    data_frame = _calc_overflow_info(data_frame, column, limit)
    return _reduce_overflow(data_frame)


def normalize_values(data_frame, normalized_time=None):
    """add exposure time normalized values to a data frame

    will use the maximum exposure time, if none is provided
    """
    if normalized_time is None:
        normalized_time = data_frame[SETTINGS_EXPOSURE_TIME].max()
    print(normalized_time)

    data_frame[SETTINGS_NORMALIZED_EXPOSURE_TIME] = normalized_time

    for original_col, normalized_col in RAW_DATA_NORMALIZATION_MAP.items():
        data_frame[normalized_col] = (
            data_frame[original_col] / data_frame[SETTINGS_EXPOSURE_TIME]
        ) * data_frame[SETTINGS_NORMALIZED_EXPOSURE_TIME]

    return data_frame


def create_xdr(
    data_frame,
    normalized_time=None,
    column=RAW_DATA_SPOT_MEAN,
    limit=0.5,
):
    """normalize measurement exposures

    normalized_time:
        if it is None, the max exposure time is used for normalization.
    """
    data_frame = blend(data_frame, column, limit)
    return normalize_values(data_frame, normalized_time)