// Working with subcommands is simple. There are a few key points to remember when working with // subcommands in clap. First, s are really just Apps. This means they can have their own // settings, version, authors, args, and even their own subcommands. The next thing to remember is // that subcommands are set up in a tree like heirachy. // // An ASCII art depiction may help explain this better. Using a fictional version of git as the demo // subject. Imagine the following are all subcommands of git (note, the author is aware these aren't // actually all subcommands in the real git interface, but it makes explanation easier) // // Top Level App (git) TOP // | // ----------------------------------------- // / | \ \ // clone push add commit LEVEL 1 // | / \ / \ | // url origin remote ref name message LEVEL 2 // / /\ // path remote local LEVEL 3 // // Given the above fictional subcommand hierarchy, valid runtime uses would be (not an all inclusive // list): // // $ git clone url // $ git push origin path // $ git add ref local // $ git commit message // // Notice only one command per "level" may be used. You could not, for example, do: // // $ git clone url push origin path // // It's also important to know that subcommands each have their own set of matches and may have args // with the same name as other subcommands in a different part of the tree heirachy (i.e. the arg // names aren't in a flat namespace). // // In order to use subcommands in clap, you only need to know which subcommand you're at in your // tree, and which args are defined on that subcommand. // // Let's make a quick program to illustrate. We'll be using the same example as above but for // brevity sake we won't implement all of the subcommands, only a few. extern crate clap; use clap::{App, AppSettings, Arg, }; fn main() { let matches = App::new("git") .about("A fictional versioning CLI") .version("1.0") .author("Me") .subcommand( App::new("clone").about("clones repos").arg( Arg::with_name("repo") .help("The repo to clone") .required(true), ), ) .subcommand( App::new("push") .about("pushes things") .setting(AppSettings::SubcommandRequiredElseHelp) .subcommand( App::new("remote") // Subcommands can have thier own subcommands, // which in turn have their own subcommands .about("pushes remote things") .arg(Arg::with_name("repo") .required(true) .help("The remote repo to push things to")), ) .subcommand(App::new("local").about("pushes local things")), ) .subcommand( App::new("add") .about("adds things") .author("Someone Else") // Subcommands can list different authors .version("v2.0 (I'm versioned differently") // or different version from their parents .setting(AppSettings::ArgRequiredElseHelp) // They can even have different settings .arg(Arg::with_name("stuff") .long("stuff") .help("Stuff to add") .takes_value(true) .multiple(true)), ) .get_matches(); // At this point, the matches we have point to git. Keep this in mind... // You can check if one of git's subcommands was used if matches.is_present("clone") { println!("'git clone' was run."); } // You can see which subcommand was used if let Some(subcommand) = matches.subcommand_name() { println!("'git {}' was used", subcommand); // It's important to note, this *only* check's git's DIRECT children, **NOT** it's // grandchildren, great grandchildren, etc. // // i.e. if the command `git push remove --stuff foo` was run, the above will only print out, // `git push` was used. We'd need to get push's matches to see futher into the tree } // An alternative to checking the name is matching on known names. Again notice that only the // direct children are matched here. match matches.subcommand_name() { Some("clone") => println!("'git clone' was used"), Some("push") => println!("'git push' was used"), Some("add") => println!("'git add' was used"), None => println!("No subcommand was used"), _ => unreachable!(), // Assuming you've listed all direct children above, this is unreachable } // You could get the independent subcommand matches, although this is less common if let Some(clone_matches) = matches.subcommand_matches("clone") { // Now we have a reference to clone's matches println!("Cloning repo: {}", clone_matches.value_of("repo").unwrap()); } // The most common way to handle subcommands is via a combined approach using // `ArgMatches::subcommand` which returns a tuple of both the name and matches match matches.subcommand() { ("clone", Some(clone_matches)) => { // Now we have a reference to clone's matches println!("Cloning {}", clone_matches.value_of("repo").unwrap()); } ("push", Some(push_matches)) => { // Now we have a reference to push's matches match push_matches.subcommand() { ("remote", Some(remote_matches)) => { // Now we have a reference to remote's matches println!("Pushing to {}", remote_matches.value_of("repo").unwrap()); } ("local", Some(_)) => { println!("'git push local' was used"); } _ => unreachable!(), } } ("add", Some(add_matches)) => { // Now we have a reference to add's matches println!( "Adding {}", add_matches .values_of("stuff") .unwrap() .collect::>() .join(", ") ); } ("", None) => println!("No subcommand was used"), // If no subcommand was usd it'll match the tuple ("", None) _ => unreachable!(), // If all subcommands are defined above, anything else is unreachabe!() } // Continued program logic goes here... }