Parse a GEP-2257 like duration string into a Python timedelta

parse_duration_string.py

import re
from datetime import timedelta

# Ordered case-insensitive duration strings, loosely based on the [GEP-2257](https://gateway-api.sigs.k8s.io/geps/gep-2257/) spec
# Discrepancies between this pattern and GEP-2257 duration strings:
# - this pattern accepts units `w|d|h|m|s|ms|[uµ]s` (all units supported by the datetime.timedelta constructor), GEP-2257 accepts only `h|m|s|ms`
# - this pattern allows for optional whitespace around the units, GEP-2257 does not
# - this pattern is compiled to be case-insensitive, GEP-2257 expects lowercase units
# - this pattern expects ordered (descending) units, GEP-2257 allows arbitrary order
# - this pattern does not allow duplicate unit occurrences, GEP-2257 does
# - this pattern allows for negative integers, GEP-2257 does not
_TIMEDELTA_PATTERN = re.compile(
    r"^(?:\s*)"  # optional whitespace at the beginning of the string
    r"(?:(-?\d+)\s*w\s*)?"  # weeks with optional whitespace around unit
    r"(?:(-?\d+)\s*d\s*)?"  # days with optional whitespace around unit
    r"(?:(-?\d+)\s*h\s*)?"  # hours with optional whitespace around unit
    r"(?:(-?\d+)\s*m\s*)?"  # minutes with optional whitespace around unit
    r"(?:(-?\d+)\s*s\s*)?"  # seconds with optional whitespace around unit
    r"(?:(-?\d+)\s*ms\s*)?"  # milliseconds with optional whitespace around unit
    r"(?:(-?\d+)\s*[µu]s\s*)?$",  # microseconds with optional whitespace around unit
    flags=re.IGNORECASE,
)

def parse_duration_string(value: str) -> timedelta | None:
    """Parse a GEP-2257 like duration string into a timedelta.

    >>> parse_duration_string('42w 42d 42h 42m 42s 42ms 42us')
    datetime.timedelta(days=337, seconds=67362, microseconds=42042)
    """
    match = _TIMEDELTA_PATTERN.match(value)
    if match is not None and any(groups := match.groups(default=0)):
        return timedelta(
            weeks=int(groups[0]),
            days=int(groups[1]),
            hours=int(groups[2]),
            minutes=int(groups[3]),
            seconds=int(groups[4]),
            milliseconds=int(groups[5]),
            microseconds=int(groups[6]),
        )

parse_duration_string('5w -7us')

ref sloria/environs#366