cmd_lib

Rust command-line library

Common rust command-line macros and utilities, to write shell-script like tasks easily in rust programming language. Available at crates.io.

Why you need this

If you need to run some external commands in rust, the std::process::Command is a good abstraction layer on top of different OS syscalls. It provides fine-grained control over how a new process should be spawned, and it allows you to wait for process to finish and check the exit status or collect all of its output. However, when Redirection or Piping is needed, you need to set up the parent and child IO handles manually, like this in the rust cookbook, which is often tedious and error prone.

A lot of developers just choose shell(sh, bash, ...) scripts for such tasks, by using < to redirect input, > to redirect output and | to pipe outputs. In my experience, this is the only good parts of shell script. You can find all kinds of pitfalls and mysterious tricks to make other parts of shell script work. As the shell scripts grow, they will ultimately be unmaintainable and no one wants to touch them any more.

This cmd_lib library is trying to provide the redirection and piping capabilities, and other facilities to make writing shell-script like tasks easily without launching any shell. For the rust cookbook examples, they can usually be implemented as one line of rust macro with the help of this library, as in the examples/rust_cookbook.rs. Since they are rust code, you can always rewrite them in rust natively in the future, if necessary without spawning external commands.

What this library provides

Macros to run external commands

• run_cmd! -> CmdResult

let msg = "I love rust";
run_cmd!(echo $msg)?;
run_cmd!(echo "This is the message: $msg")?;

// pipe commands are also supported
let dir = "/var/log";
run_cmd!(du -ah $dir | sort -hr | head -n 10)?;

// or a group of commands
// if any command fails, just return Err(...), which is similar to bash's `set -euo pipefail`
let file = "/tmp/f";
let keyword = "rust";
run_cmd! {
    cat ${file} | grep ${keyword};
    echo "bad cmd" >&2;
    ignore ls /nofile;
    date;
    ls oops;
    cat oops;
}?;

• run_fun! -> FunResult

let version = run_fun!(rustc --version | awk r"{print $2}")?;
info!("Your rust version is {version}");

Abstraction without overhead

Since all the macros' lexical analysis and syntactic analysis happen at compile time, it can basically generate code the same as calling std::process APIs manually. It also includes command type checking, so most of the errors can be found at compile time instead of at runtime. With tools like rust-analyzer, it can give you real-time feedback for broken commands being used.

You can use cargo expand to check the generated code.

Intuitive parameters passing

When passing parameters to run_cmd! and run_fun! macros, if they are not part to rust String literals, they will be converted to string as an atomic component, so you don't need to quote them. The parameters will be like $a or ${a} in run_cmd! or run_fun! macros.

let dir = "my folder";
run_cmd!(echo "Creating $dir at /tmp")?;
run_cmd!(mkdir -p /tmp/$dir)?;

// or with group commands:
let dir = "my folder";
run_cmd!(echo "Creating $dir at /tmp"; mkdir -p /tmp/$dir)?;

You can consider "" as glue, so everything inside the quotes will be treated as a single atomic component.

If they are part of Raw string literals, there will be no string interpolation, the same as in idiomatic rust. However, you can always use format! macro to form the new string. For example:

// string interpolation
let key_word = "time";
let awk_opts = format!(r#"/{}/ {{print $(NF-3) " " $(NF-1) " " $NF}}"#, key_word);
run_cmd!(ping -c 10 www.google.com | awk $awk_opts)?;

Notice here $awk_opts will be treated as single option passing to awk command.

If you want to use dynamic parameters, you can use $[] to access vector variable:

let gopts = vec![vec!["-l", "-a", "/"], vec!["-a", "/var"]];
for opts in gopts {
    run_cmd!(ls $[opts])?;
}

Redirection and Piping

Right now piping and stdin, stdout, stderr redirection are supported. Most parts are the same as in bash scripts.

Logging

This library provides convenient macros and builtin commands for logging. All messages which are printed to stderr will be logged. It will also include the full running commands in the error result.

let dir: &str = "folder with spaces";
run_cmd!(mkdir /tmp/$dir; ls /tmp/$dir)?;
run_cmd!(mkdir /tmp/$dir; ls /tmp/$dir; rmdir /tmp/$dir)?;
// output:
// [INFO ] mkdir: cannot create directory ‘/tmp/folder with spaces’: File exists
// Error: Running ["mkdir" "/tmp/folder with spaces"] exited with error; status code: 1

It is using rust log crate, and you can use your actual favorite logger implementation. Notice that if you don't provide any logger, it will use env_logger to print messages from process's stderr.

You can also mark your main() function with #[cmd_lib::main], which will log error from main() by default. Like this:

[ERROR] FATAL: Running ["mkdir" "/tmp/folder with spaces"] exited with error; status code: 1

Builtin commands

cd

cd: set process current directory.

run_cmd! (
    cd /tmp;
    ls | wc -l;
)?;

Notice that builtin cd will only change with current scope and it will restore the previous current directory when it exits the scope.

Use std::env::set_current_dir if you want to change the current working directory for the whole program.

ignore

Ignore errors for command execution.

echo

Print messages to stdout.

-n     do not output the trailing newline

error, warn, info, debug, trace

Print messages to logging with different levels. You can also use the normal logging macros, if you don't need to do logging inside the command group.

run_cmd!(error "This is an error message")?;
run_cmd!(warn "This is a warning message")?;
run_cmd!(info "This is an information message")?;
// output:
// [ERROR] This is an error message
// [WARN ] This is a warning message
// [INFO ] This is an information message

Low-level process spawning macros

spawn! macro executes the whole command as a child process, returning a handle to it. By default, stdin, stdout and stderr are inherited from the parent. The process will run in the background, so you can run other stuff concurrently. You can call wait() to wait for the process to finish.

With spawn_with_output! you can get output by calling wait_with_output(), wait_with_all() or even do stream processing with wait_with_pipe().

There are also other useful APIs, and you can check the docs for more details.

let mut proc = spawn!(ping -c 10 192.168.0.1)?;
// do other stuff
// ...
proc.wait()?;

let mut proc = spawn_with_output!(/bin/cat file.txt | sed s/a/b/)?;
// do other stuff
// ...
let output = proc.wait_with_output()?;

spawn_with_output!(journalctl)?.wait_with_pipe(&mut |pipe| {
    BufReader::new(pipe)
        .lines()
        .filter_map(|line| line.ok())
        .filter(|line| line.find("usb").is_some())
        .take(10)
        .for_each(|line| println!("{}", line));
})?;

Macro to register your own commands

Declare your function with the right signature, and register it with use_custom_cmd! macro:

fn my_cmd(env: &mut CmdEnv) -> CmdResult {
    let args = env.get_args();
    let (res, stdout, stderr) = spawn_with_output! {
        orig_cmd $[args]
            --long-option xxx
            --another-option yyy
    }?
    .wait_with_all();
    writeln!(env.stdout(), "{}", stdout)?;
    writeln!(env.stderr(), "{}", stderr)?;
    res
}

use_custom_cmd!(my_cmd);

Macros to define, get and set thread-local global variables

• tls_init! to define thread local global variable
• tls_get! to get the value
• tls_set! to set the value

  tls_init!(DELAY, f64, 1.0);
  const DELAY_FACTOR: f64 = 0.8;
  tls_set!(DELAY, |d| *d *= DELAY_FACTOR);
  let d = tls_get!(DELAY);
  // check more examples in examples/tetris.rs

Other Notes

Minimum supported Rust version

cmd_lib(2.0)'s MSRV is 1.88.

Environment Variables

You can use std::env::var to fetch the environment variable key from the current process. It will report error if the environment variable is not present, and it also includes other checks to avoid silent failures.

To set environment variables in single-threaded programs, you can use std::env::set_var and std::env::remove_var. While those functions must not be called if any other threads might be running, you can always set environment variables for one command at a time, by putting the assignments before the command:

run_cmd!(FOO=100 /tmp/test_run_cmd_lib.sh)?;

Security Notes

Using macros can actually avoid command injection, since we do parsing before variable substitution. For example, below code is fine even without any quotes:

fn cleanup_uploaded_file(file: &Path) -> CmdResult {
    run_cmd!(/bin/rm -f /var/upload/$file)
}

It is not the case in bash, which will always do variable substitution at first.

Glob/Wildcard

This library does not provide glob functions, to avoid silent errors and other surprises. You can use the glob package instead.

Thread Safety

This library tries very hard to not set global state, so parallel cargo test can be executed just fine. That said, there are some limitations to be aware of:

• std::env::set_var and std::env::remove_var must not be called in a multi-threaded program
• tls_init!, tls_get!, and tls_set! create thread-local variables, which means each thread will have its own independent version of the variable
• set_debug and set_pipefail are global and affect all threads; to change those settings without affecting other threads, use ScopedDebug and ScopedPipefail

License: MIT OR Apache-2.0