Function arg_split_with_quotes

IPython/utils/_process_common.py:226–270

Split a command line and report which tokens were originally quoted. Returns a list of ``(token, was_quoted)`` pairs. ``token`` is the unquoted form, as ``shlex.split(posix=True)`` returns, and ``was_quoted`` is True if that token had any single- or double-quote characters in ``commandl

(
    commandline: str, strict: bool = True
)

def arg_split_with_quotes(
    commandline: str, strict: bool = True
) -> list[tuple[str, bool]]:
    """Split a command line and report which tokens were originally quoted.

    Returns a list of ``(token, was_quoted)`` pairs. ``token`` is the unquoted
    form, as ``shlex.split(posix=True)`` returns, and ``was_quoted`` is True
    if that token had any single- or double-quote characters in ``commandline``.

    Useful for callers like ``%run`` that want to honor shell quoting when
    deciding wether to apply further expansion (glob, tilde) to a token.

    Detection is shlex-based on both passes so the quote semantics are the
    same on Posix and Windows. If ``strict`` is False, malformed input (e.g.
    an unbalanced quote) returns whatever was parsed so far instead of raising.
    """
    def _tokenize(s: str, posix: bool) -> list[str]:
        lex = shlex.shlex(s, posix=posix)
        lex.whitespace_split = True
        lex.commenters = ''
        out = []
        while True:
            try:
                out.append(next(lex))
            except StopIteration:
                break
            except ValueError:
                if strict:
                    raise
                out.append(lex.token)
                break
        return out

    raw_tokens = _tokenize(commandline, posix=False)
    clean_tokens = _tokenize(commandline, posix=True)

    if len(raw_tokens) != len(clean_tokens):
        # If the two passes disagree (exotic input) report nothing as quoted
        # so callers get the legacy, non-quote-aware behavior.
        return [(t, False) for t in clean_tokens]

    return [
        (clean, ("'" in raw) or ('"' in raw))
        for clean, raw in zip(clean_tokens, raw_tokens)
    ]