interval_index_from_str.py

import pandas as pd


_interval_regex = r"^(?P<open>[[(])"
_interval_regex += r"(?P<low>-inf|[0-9][.0-9]*)"
_interval_regex += r"\s*,\s*"
_interval_regex += r"(?P<high>\+?inf|[0-9][.0-9]*)"
_interval_regex += r"(?P<close>[)\]])$"


def interval_from_string(series):
    if not pd.api.types.is_string_dtype(series):
        return series
    extracted = series.str.extract(_interval_regex)
    extracted = extracted.dropna()
    if len(extracted) != len(series):
        return series
    left_closed = extracted.open.unique()
    right_closed = extracted.close.unique()
    if len(left_closed) != 1 or len(right_closed) != 1:
        return series
    left_closed = left_closed[0] == "["
    right_closed = right_closed[0] == "]"
    if left_closed:
        if right_closed:
            closed = "both"
        else:
            closed = "left"
    else:
        if right_closed:
            closed = "right"
        else:
            closed = "neither"
    interval = pd.IntervalIndex.from_arrays(left=pd.to_numeric(extracted.low), 
                                            right=pd.to_numeric(extracted.high), 
                                            closed=closed)
    return interval