Subcommand requirements not being negated

[marcaddeo]

Hi all,

I'm trying to have it so that my program can be run with or without a subcommand. If a subcommand is passed, the global required arguments should no longer be required. If there is no subcommand, the global required arguments should remain required.

This is what I've got:

use clap::{Parser, Subcommand};

#[derive(Subcommand, Clone)]
pub enum Commands {
    /// Print a config template.
    ConfigTemplate,
}

#[derive(Parser)]
#[command(arg_required_else_help(true))]
#[command(subcommand_negates_reqs(true))]
#[command(args_conflicts_with_subcommands(true))]
pub struct Cli {
    /// Output what would happen, but don't actually submit to Clockify.
    #[arg(short, long)]
    pub dry_run: bool,

    #[command(subcommand)]
    pub command: Option<Commands>,

    /// The Jira timesheet CSV export file. Use '-' to read from stdin.
    pub file: String,
}

cargo run -- help outputs:

Usage: clockify-transfer [OPTIONS] <FILE>
       clockify-transfer <COMMAND>

Commands:
  config-template  Print a config template
  help             Print this message or the help of the given subcommand(s)

Arguments:
  <FILE>  The Jira timesheet CSV export file. Use '-' to read from stdin

Options:
  -d, --dry-run  Output what would happen, but don't actually submit to Clockify
  -h, --help     Print help information

The multiple Usage statements indicate that this confirmation should make it do what I want, but it doesn't seem to work:

cargo run -- config-template outputs:

error: The following required argument was not provided: file

Usage: clockify-transfer [OPTIONS] <FILE>
       clockify-transfer <COMMAND>

For more information try '--help'

Which is still requiring the file argument.

Is there something I'm doing wrong, or is this a bug?

[epage]

That error isn't coming from clap's validation logic but the derive when it is trying to fill in the file field within the struct.

Options

• Change the type to Option<String> and explicitly mark it as required.
• Move all of the arguments into a GlobalArgs struct that you flatten with args: Option<GlobalArgs>
  • In the future, Support for using ArgGroup as Enum with derive #2621 will make it so disjoint sets of arguments can be in an enum. We might be able to do the same with subcommands and implicitly set args_conflicts_with_subcommands

[marcaddeo]

Thanks for clarifying that for me! I ended up going with the GlobalArgs option.

[explosion-mental]

This works great, tho clap_completion doesn't seem to generate proper options, and even then, if a file has the same name as a subcommand it ignores that and treats it as a subcommand.

For that reason I think it's preferable to use the good old program run file when using subcommands.