quickopts

Tiny opinionated Python library for parsing script options.

Overview

quickopts parses POSIX getopts-style command-line arguments from the same docstring you use for help text. It supports:

• single-letter options, with clustering, so -Cz behaves like -C -z and -Czbmain behaves like -C -z -b main;
• positional arguments, where the first positional argument stops option parsing, plus -- to stop option parsing explicitly.

It is intentionally small and does not support long options like --flag, optional flag values, or validation of Synopsis: patterns.

Sample usage

quickopts works well for single-file scripts using PEP 723 inline dependency metadata. Tools such as uv can read the script header, create an environment with quickopts installed, and then run the file directly from its shebang.

#!/usr/bin/env -S uv run --script
# /// script
# requires-python = ">=3.12"
# dependencies = [
#   "quickopts",
# ]
# ///

"""A cool tool

Synopsis:
  $prog [-L]
  $prog -C [-zb BRANCH]
  $prog -D VAL
  $prog -W [--] [ARGS...]

Commands:
  -C  Create a value.
  -D  Delete a value.
  -L  List all the values.
  -W  Wrap a command.
  -h  Show this message and exit.

Options:
  -b BRANCH  Name of the value's branch.
  -z         Sleep for a little while first.
"""

import quickopts


if __name__ == "__main__":
    q = quickopts.parse_or_exit(
        __doc__,
        template_mapping={"prog": quickopts.Placeholder.PROGRAM},
    )
    match q.command or "L":
        case "C": _create(branch=q.flags.get("b", "main"), sleep="z" in q.switches)
        case "D": _delete(q.args[0])
        case "L": ...
        case "W": ...

Parsing rules

quickopts reads option definitions from Commands: and Options: sections in the docstring. Blank lines end the current section. Synopsis: is not parsed; it is only help text.

Command entries define mutually exclusive commands. We recommend using uppercase letters for these, other than the standard -h help command:

Commands:
  -C  Create a value.
  -L  List all values.
  -h  Show help.

Option entries define either value flags or switches. In Options:, an entry is a value flag when non-whitespace text appears between the option name and the two-or-more-space gap before the description. Otherwise it is a switch. We recommend lowercase letters in this section:

Options:
  -b BRANCH  Value flag.
  -z         Switch.

parse(doc, args) parses args without a program name and returns a Parsed object with the following attributes:

• command: the selected command, or None;
• flags: value flags, for example {"b": "main"}; repeated values are available through q.flags.getall("b");
• switches: switch repeat counts, for example {"z": 2}; use "z" in q.switches to check presence;
• args: remaining positional arguments.

Single-dash options use POSIX getopts-style clustering:

When a value flag appears inside a cluster, it consumes the rest of that token as its value. If no characters remain, it consumes the following arg. Repeated value flags collect all values in order and scalar access returns the last value. Repeated switches are counted.

parse_or_exit(doc, template_mapping={"prog": quickopts.Placeholder.PROGRAM}) reads sys.argv, prints parse errors to stderr with exit code 2, and handles -h by printing the docstring and exiting with code 0. Template mappings use string.Template substitution, and values may be literal strings or built-in placeholders such as quickopts.Placeholder.PROGRAM, which resolves to the display program name, including python -m module when invoked with -m. The older program_var="prog" shortcut is still supported for compatibility.