config.py

from typing import Dict, Union
from shlex import quote as shquote, split as shsplitwords


def shquotewords(s: str) -> str:
    """
    Split string *s* into (POSIX) shell words, quote each word, and join them
    back into a string.

    This is suitable for properly quoting multi-word, user-defined values which
    should follow shell quoting and escaping semantics (e.g. to allow spaces in
    single words) but not allow shell features like variable interpolation,
    command substition, redirection, piping, etc.

    >>> shquotewords('''a "b c" d $E''')
    "a 'b c' d '$E'"

    >>> shquotewords('''foo >bar''')
    "foo '>bar'"
    """
    return " ".join(shquote(word) for word in shsplitwords(s))


def augur_options(opts: Dict[str, Union[str, list, None]]) -> str:
    """
    Converts a dictionary of command-line config options to a string suitable
    for use in an ``augur`` invocation.

    Values are appropriately quoted/escaped:

    >>> augur_options({"--min-length": 9000, "--include-where": "state = WA"})
    "--min-length 9000 --include-where 'state = WA'"

    Flags are keys where the value is ``None``:

    >>> augur_options({"--exclude-all": None})
    '--exclude-all'

    Options that take multiple values can use shell-like quoting and escaping
    in a single Python string:

    >>> augur_options({"--metadata-id-columns": 'strain "seq name" seq\ id $id'})
    "--metadata-id-columns strain 'seq name' 'seq id' '$id'"

    or use a Python list:

    >>> augur_options({"--metadata-id-columns": ["strain", "seq name", "seq id", "$id"]})
    "--metadata-id-columns strain 'seq name' 'seq id' '$id'"
    """
    return " ".join(
        shquotewords(opt if val is None else f"{opt} {val}")
            for opt, val in opts.items())