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 a tedious work.
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 looks like
To get a first impression, here is an example from examples/dd_test.rs:
run_cmd! (
info "Dropping caches at first";
sudo bash -c "echo 3 > /proc/sys/vm/drop_caches";
info "Running with thread_num: $thread_num, block_size: $block_size";
)?;
let cnt = DATA_SIZE / thread_num / block_size;
let now = Instant::now();
(0..thread_num).into_par_iter().for_each(|i| {
let off = cnt * i;
let bandwidth = run_fun!(
sudo bash -c "dd if=$file of=/dev/null bs=$block_size skip=$off count=$cnt 2>&1"
| awk r#"/copied/{print $(NF-1) " " $NF}"#
)
.unwrap_or_else(|_| cmd_die!("thread $i failed"));
cmd_info!("thread $i bandwidth: $bandwidth");
});
let total_bandwidth = Byte::from_bytes((DATA_SIZE / now.elapsed().as_secs()) as u128)
.get_appropriate_unit(true)
.to_string();
cmd_info!("Total bandwidth: ${total_bandwidth}/s");
Output will be like this:
➜ rust_cmd_lib git:(master) ✗ cargo run --example dd_test -- -b 4096 -f /dev/nvme0n1 -t 4
Finished dev [unoptimized + debuginfo] target(s) in 1.56s
Running `target/debug/examples/dd_test -b 4096 -f /dev/nvme0n1 -t 4`
INFO - Dropping caches at first
INFO - Running with thread_num: 4, block_size: 4096
INFO - thread 1 bandwidth: 286 MB/s
INFO - thread 3 bandwidth: 269 MB/s
INFO - thread 2 bandwidth: 267 MB/s
INFO - thread 0 bandwidth: 265 MB/s
INFO - Total bandwidth: 1.01 GiB/s
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(...)
let file = "/tmp/f";
let keyword = "rust";
if run_cmd! {
cat ${file} | grep ${keyword};
echo "bad cmd" >&2;
ignore ls /nofile;
date;
ls oops;
cat oops;
}.is_err() {
// your error handling code
}
- run_fun! --> FunResult
let version = run_fun!(rustc --version)?;
eprintln!("Your rust version is {}", version);
// with pipes
let n = run_fun!(echo "the quick brown fox jumped over the lazy dog" | wc -w)?;
eprintln!("There are {} words in above sentence", n);
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 $[opts])?;
}
Redirection and Piping
Right now piping and stdin, stdout, stderr redirection are supported. Most parts are the same as in bash scripts.
Logging
This library provides convenient macros and builtin commands for logging. All messages which are printed to stderr will be logged. Since it is returning result type, you can also log the errors if command execution fails.
// this code snppit is using a builtin simple logger, you can replace it with a real logger
init_builtin_logger();
let dir: &str = "folder with spaces";
assert!(run_cmd!(mkdir /tmp/$dir; ls /tmp/$dir).is_ok());
assert!(run_cmd!(mkdir /tmp/"$dir"; ls /tmp/"$dir"; rmdir /tmp/"$dir").is_err());
// output:
// INFO - mkdir: cannot create directory ‘/tmp/folder with spaces’: File exists
It is using rust log crate, and you can use your actual favorite logging implementation. Notice that if you don't provide any logger, the stderr output will be discarded.
Builtin commands
cd
cd: set process current directory, which can be used without importing.
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, which can be used without importing.
echo
Print messages to stdout, which needs to be imported with use_builtin_cmd!
macro.
use_builtin_cmd!(echo, warn); // find more builtin commands in src/builtins.rs
run_cmd!(echo "This is from builtin command!")?;
run_cmd!(warn "This is from builtin command!")?;
Macros to register your own commands
Declare your function with #[export_cmd(..)]
attribute, and import it with use_custom_cmd!
macro:
#[export_cmd(my_cmd)]
fn foo(env: &mut CmdEnv) -> CmdResult {
let msg = format!("msg from foo(), args: {:?}", env.args());
writeln!(env.stderr(), "{}", msg)?;
writeln!(env.stdout(), "bar")
}
use_custom_cmd!(my_cmd);
run_cmd!(my_cmd)?;
println!("get result: {}", run_fun!(my_cmd)?);
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()
, or even do stream
processing with wait_with_pipe()
.
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));
})?;
Macros to define, get and set thread-local global variables
tls_init!
to define thread local global variabletls_get!
to get the valuetls_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
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, you can use std::env::set_var. There are also other related APIs in the std::env module.
To set environment variables for the command only, you can put the assignments before the command. Like this:
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 states, so parallel cargo test
can be executed just fine.
The only known APIs not supported in multi-thread environment are the
tls_init/tls_get/tls_set
macros, and you should only use them for thread local variables.
License: MIT OR Apache-2.0