src/content/docs/develop/calling-rust.mdx
import { Content as FrontendListen } from './_sections/frontend-listen.mdx';
This document includes guides on how to communicate with your Rust code from your application frontend. To see how to communicate with your frontend from your Rust code, see Calling the Frontend from Rust.
Tauri provides a command primitive for reaching Rust functions with type safety, along with an event system that is more dynamic.
Tauri provides a simple yet powerful command system for calling Rust functions from your web app.
Commands can accept arguments and return values. They can also return errors and be async.
Commands can be defined in your src-tauri/src/lib.rs file.
To create a command, just add a function and annotate it with #[tauri::command]:
#[tauri::command]
fn my_custom_command() {
println!("I was invoked from JavaScript!");
}
:::note Command names must be unique. :::
:::note
Commands defined in the lib.rs file cannot be marked as pub due to a limitation in the glue code generation.
You will see an error like this if you mark it as a public function:
error[E0255]: the name `__cmd__command_name` is defined multiple times
--> src/lib.rs:28:8
|
27 | #[tauri::command]
| ----------------- previous definition of the macro `__cmd__command_name` here
28 | pub fn x() {}
| ^ `__cmd__command_name` reimported here
|
= note: `__cmd__command_name` must be defined only once in the macro namespace of this module
:::
You will have to provide a list of your commands to the builder function like so:
#[cfg_attr(mobile, tauri::mobile_entry_point)]
pub fn run() {
tauri::Builder::default()
.invoke_handler(tauri::generate_handler![my_custom_command])
.run(tauri::generate_context!())
.expect("error while running tauri application");
}
Now, you can invoke the command from your JavaScript code:
// When using the Tauri API npm package:
import { invoke } from '@tauri-apps/api/core';
// When using the Tauri global script (if not using the npm package)
// Be sure to set `app.withGlobalTauri` in `tauri.conf.json` to true
const invoke = window.__TAURI__.core.invoke;
// Invoke the command
invoke('my_custom_command');
If your application defines a lot of components or if they can be grouped,
you can define commands in a separate module instead of bloating the lib.rs file.
As an example let's define a command in the src-tauri/src/commands.rs file:
#[tauri::command]
pub fn my_custom_command() {
println!("I was invoked from JavaScript!");
}
:::note
When defining commands in a separate module they should be marked as pub.
:::
:::note The command name is not scoped to the module so they must be unique even between modules. :::
In the lib.rs file, define the module and provide the list of your commands accordingly;
mod commands;
#[cfg_attr(mobile, tauri::mobile_entry_point)]
pub fn run() {
tauri::Builder::default()
.invoke_handler(tauri::generate_handler![commands::my_custom_command])
.run(tauri::generate_context!())
.expect("error while running tauri application");
}
Note the commands:: prefix in the command list, which denotes the full path to the command function.
The command name in this example is my_custom_command so you can still call it by executing invoke("my_custom_command")
in your frontend, the commands:: prefix is ignored.
When using a Rust frontend to call invoke() without arguments, you will need to adapt your frontend code as below.
The reason is that Rust doesn't support optional arguments.
#[wasm_bindgen]
extern "C" {
// invoke without arguments
#[wasm_bindgen(js_namespace = ["window", "__TAURI__", "core"], js_name = invoke)]
async fn invoke_without_args(cmd: &str) -> JsValue;
// invoke with arguments (default)
#[wasm_bindgen(js_namespace = ["window", "__TAURI__", "core"])]
async fn invoke(cmd: &str, args: JsValue) -> JsValue;
// They need to have different names!
}
Your command handlers can take arguments:
#[tauri::command]
fn my_custom_command(invoke_message: String) {
println!("I was invoked from JavaScript, with this message: {}", invoke_message);
}
Arguments should be passed as a JSON object with camelCase keys:
invoke('my_custom_command', { invokeMessage: 'Hello!' });
:::note
You can use snake_case for the arguments with the rename_all attribute:
#[tauri::command(rename_all = "snake_case")]
fn my_custom_command(invoke_message: String) {}
The corresponding JavaScript:
invoke('my_custom_command', { invoke_message: 'Hello!' });
:::
Arguments can be of any type, as long as they implement serde::Deserialize.
Command handlers can return data as well:
#[tauri::command]
fn my_custom_command() -> String {
"Hello from Rust!".into()
}
The invoke function returns a promise that resolves with the returned value:
invoke('my_custom_command').then((message) => console.log(message));
Returned data can be of any type, as long as it implements serde::Serialize.
Return values that implements serde::Serialize are serialized to JSON when the response is sent to the frontend.
This can slow down your application if you try to return a large data such as a file or a download HTTP response.
To return array buffers in an optimized way, use tauri::ipc::Response:
use tauri::ipc::Response;
#[tauri::command]
fn read_file() -> Response {
let data = std::fs::read("/path/to/file").unwrap();
tauri::ipc::Response::new(data)
}
If your handler could fail and needs to be able to return an error, have the function return a Result:
#[tauri::command]
fn login(user: String, password: String) -> Result<String, String> {
if user == "tauri" && password == "tauri" {
// resolve
Ok("logged_in".to_string())
} else {
// reject
Err("invalid credentials".to_string())
}
}
If the command returns an error, the promise will reject, otherwise, it resolves:
invoke('login', { user: 'tauri', password: '0j4rijw8=' })
.then((message) => console.log(message))
.catch((error) => console.error(error));
As mentioned above, everything returned from commands must implement serde::Serialize, including errors.
This can be problematic if you're working with error types from Rust's std library or external crates as most error types do not implement it.
In simple scenarios you can use map_err to convert these errors to String:
#[tauri::command]
fn my_custom_command() -> Result<(), String> {
std::fs::File::open("path/to/file").map_err(|err| err.to_string())?;
// Return `null` on success
Ok(())
}
Since this is not very idiomatic you may want to create your own error type which implements serde::Serialize.
In the following example, we use the thiserror crate to help create the error type.
It allows you to turn enums into error types by deriving the thiserror::Error trait.
You can consult its documentation for more details.
// create the error type that represents all errors possible in our program
#[derive(Debug, thiserror::Error)]
enum Error {
#[error(transparent)]
Io(#[from] std::io::Error)
}
// we must manually implement serde::Serialize
impl serde::Serialize for Error {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::ser::Serializer,
{
serializer.serialize_str(self.to_string().as_ref())
}
}
#[tauri::command]
fn my_custom_command() -> Result<(), Error> {
// This will return an error
std::fs::File::open("path/that/does/not/exist")?;
// Return `null` on success
Ok(())
}
A custom error type has the advantage of making all possible errors explicit so readers can quickly identify what errors can happen. This saves other people (and yourself) enormous amounts of time when reviewing and refactoring code later.
It also gives you full control over the way your error type gets serialized. In the above example, we simply returned the error message as a string, but you could assign each error a code so you could more easily map it to a similar looking TypeScript error enum for example:
#[derive(Debug, thiserror::Error)]
enum Error {
#[error(transparent)]
Io(#[from] std::io::Error),
#[error("failed to parse as string: {0}")]
Utf8(#[from] std::str::Utf8Error),
}
#[derive(serde::Serialize)]
#[serde(tag = "kind", content = "message")]
#[serde(rename_all = "camelCase")]
enum ErrorKind {
Io(String),
Utf8(String),
}
impl serde::Serialize for Error {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::ser::Serializer,
{
let error_message = self.to_string();
let error_kind = match self {
Self::Io(_) => ErrorKind::Io(error_message),
Self::Utf8(_) => ErrorKind::Utf8(error_message),
};
error_kind.serialize(serializer)
}
}
#[tauri::command]
fn read() -> Result<Vec<u8>, Error> {
let data = std::fs::read("/path/to/file")?;
Ok(data)
}
In your frontend you now get a { kind: 'io' | 'utf8', message: string } error object:
type ErrorKind = {
kind: 'io' | 'utf8';
message: string;
};
invoke('read').catch((e: ErrorKind) => {});
Asynchronous commands are preferred in Tauri to perform heavy work in a manner that doesn't result in UI freezes or slowdowns.
:::note
Async commands are executed on a separate async task using async_runtime::spawn.
Commands without the async keyword are executed on the main thread unless defined with #[tauri::command(async)].
:::
If your command needs to run asynchronously, simply declare it as async.
:::caution
You need to be careful when creating asynchronous functions using Tauri.
Currently, you cannot simply include borrowed arguments in the signature of an asynchronous function.
Some common examples of types like this are &str and State<'_, Data>.
This limitation is tracked here: https://github.com/tauri-apps/tauri/issues/2533 and workarounds are shown below.
:::
When working with borrowed types, you have to make additional changes. These are your two main options:
Option 1: Convert the type, such as &str to a similar type that is not borrowed, such as String.
This may not work for all types, for example State<'_, Data>.
Example:
// Declare the async function using String instead of &str, as &str is borrowed and thus unsupported
#[tauri::command]
async fn my_custom_command(value: String) -> String {
// Call another async function and wait for it to finish
some_async_function().await;
value
}
Option 2: Wrap the return type in a Result. This one is a bit harder to implement, but works for all types.
Use the return type Result<a, b>, replacing a with the type you wish to return, or () if you wish to return null, and replacing b with an error type to return if something goes wrong, or () if you wish to have no optional error returned. For example:
Result<String, ()> to return a String, and no error.Result<(), ()> to return null.Result<bool, Error> to return a boolean or an error as shown in the Error Handling section above.Example:
// Return a Result<String, ()> to bypass the borrowing issue
#[tauri::command]
async fn my_custom_command(value: &str) -> Result<String, ()> {
// Call another async function and wait for it to finish
some_async_function().await;
// Note that the return value must be wrapped in `Ok()` now.
Ok(format!(value))
}
Since invoking the command from JavaScript already returns a promise, it works just like any other command:
invoke('my_custom_command', { value: 'Hello, Async!' }).then(() =>
console.log('Completed!')
);
The Tauri channel is the recommended mechanism for streaming data such as streamed HTTP responses to the frontend. The following example reads a file and notifies the frontend of the progress in chunks of 4096 bytes:
use tokio::io::AsyncReadExt;
#[tauri::command]
async fn load_image(path: std::path::PathBuf, reader: tauri::ipc::Channel<&[u8]>) {
// for simplicity this example does not include error handling
let mut file = tokio::fs::File::open(path).await.unwrap();
let mut chunk = vec![0; 4096];
loop {
let len = file.read(&mut chunk).await.unwrap();
if len == 0 {
// Length of zero means end of file.
break;
}
reader.send(&chunk).unwrap();
}
}
See the channels documentation for more information.
Commands can access the WebviewWindow instance that invoked the message:
#[tauri::command]
async fn my_custom_command(webview_window: tauri::WebviewWindow) {
println!("WebviewWindow: {}", webview_window.label());
}
Commands can access an AppHandle instance:
#[tauri::command]
async fn my_custom_command(app_handle: tauri::AppHandle) {
let app_dir = app_handle.path().app_dir();
use tauri::GlobalShortcutManager;
app_handle.global_shortcut_manager().register("CTRL + U", move || {});
}
:::tip
AppHandle and WebviewWindow both take a generic parameter R: Runtime,
when the wry feature is enabled in tauri (which is enabled by default),
we default the generic to the Wry runtime so you can use it directly,
but if you want to use a different runtime, for example the mock runtime,
you need to write your functions like this
use tauri::{AppHandle, GlobalShortcutManager, Runtime, WebviewWindow};
#[tauri::command]
async fn my_custom_command<R: Runtime>(app_handle: AppHandle<R>, webview_window: WebviewWindow<R>) {
let app_dir = app_handle.path().app_dir();
app_handle
.global_shortcut_manager()
.register("CTRL + U", move || {});
println!("WebviewWindow: {}", webview_window.label());
}
:::
Tauri can manage state using the manage function on tauri::Builder.
The state can be accessed on a command using tauri::State:
struct MyState(String);
#[tauri::command]
fn my_custom_command(state: tauri::State<MyState>) {
assert_eq!(state.0 == "some state value", true);
}
#[cfg_attr(mobile, tauri::mobile_entry_point)]
pub fn run() {
tauri::Builder::default()
.manage(MyState("some state value".into()))
.invoke_handler(tauri::generate_handler![my_custom_command])
.run(tauri::generate_context!())
.expect("error while running tauri application");
}
Tauri commands can also access the full tauri::ipc::Request object which includes the raw body payload and the request headers.
#[derive(Debug, thiserror::Error)]
enum Error {
#[error("unexpected request body")]
RequestBodyMustBeRaw,
#[error("missing `{0}` header")]
MissingHeader(&'static str),
}
impl serde::Serialize for Error {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where
S: serde::ser::Serializer,
{
serializer.serialize_str(self.to_string().as_ref())
}
}
#[tauri::command]
fn upload(request: tauri::ipc::Request) -> Result<(), Error> {
let tauri::ipc::InvokeBody::Raw(upload_data) = request.body() else {
return Err(Error::RequestBodyMustBeRaw);
};
let Some(authorization_header) = request.headers().get("Authorization") else {
return Err(Error::MissingHeader("Authorization"));
};
// upload...
Ok(())
}
In the frontend you can call invoke() sending a raw request body by providing an ArrayBuffer or Uint8Array on the payload argument, and include request headers in the third argument:
const data = new Uint8Array([1, 2, 3]);
await __TAURI__.core.invoke('upload', data, {
headers: {
Authorization: 'apikey',
},
});
The tauri::generate_handler! macro takes an array of commands. To register
multiple commands, you cannot call invoke_handler multiple times. Only the last
call will be used. You must pass each command to a single call of
tauri::generate_handler!.
#[tauri::command]
fn cmd_a() -> String {
"Command a"
}
#[tauri::command]
fn cmd_b() -> String {
"Command b"
}
#[cfg_attr(mobile, tauri::mobile_entry_point)]
pub fn run() {
tauri::Builder::default()
.invoke_handler(tauri::generate_handler![cmd_a, cmd_b])
.run(tauri::generate_context!())
.expect("error while running tauri application");
}
Any or all of the above features can be combined:
struct Database;
#[derive(serde::Serialize)]
struct CustomResponse {
message: String,
other_val: usize,
}
async fn some_other_function() -> Option<String> {
Some("response".into())
}
#[tauri::command]
async fn my_custom_command(
window: tauri::Window,
number: usize,
database: tauri::State<'_, Database>,
) -> Result<CustomResponse, String> {
println!("Called from {}", window.label());
let result: Option<String> = some_other_function().await;
if let Some(message) = result {
Ok(CustomResponse {
message,
other_val: 42 + number,
})
} else {
Err("No result".into())
}
}
#[cfg_attr(mobile, tauri::mobile_entry_point)]
pub fn run() {
tauri::Builder::default()
.manage(Database {})
.invoke_handler(tauri::generate_handler![my_custom_command])
.run(tauri::generate_context!())
.expect("error while running tauri application");
}
import { invoke } from '@tauri-apps/api/core';
// Invocation from JavaScript
invoke('my_custom_command', {
number: 42,
})
.then((res) =>
console.log(`Message: ${res.message}, Other Val: ${res.other_val}`)
)
.catch((e) => console.error(e));
The event system is a simpler communication mechanism between your frontend and the Rust. Unlike commands, events are not type safe, are always async, cannot return values and only supports JSON payloads.
To trigger a global event you can use the event.emit or the WebviewWindow#emit functions:
import { emit } from '@tauri-apps/api/event';
import { getCurrentWebviewWindow } from '@tauri-apps/api/webviewWindow';
// emit(eventName, payload)
emit('file-selected', '/path/to/file');
const appWebview = getCurrentWebviewWindow();
appWebview.emit('route-changed', { url: window.location.href });
:::note Global events are delivered to all listeners :::
To trigger an event to a listener registered by a specific webview you can use the event.emitTo or the WebviewWindow#emitTo functions:
import { emitTo } from '@tauri-apps/api/event';
import { getCurrentWebviewWindow } from '@tauri-apps/api/webviewWindow';
// emitTo(webviewLabel, eventName, payload)
emitTo('settings', 'settings-update-requested', {
key: 'notification',
value: 'all',
});
const appWebview = getCurrentWebviewWindow();
appWebview.emitTo('editor', 'file-changed', {
path: '/path/to/file',
contents: 'file contents',
});
:::note
Webview-specific events are not triggered to regular global event listeners.
To listen to any event you must provide the { target: { kind: 'Any' } } option to the event.listen function,
which defines the listener to act as a catch-all for emitted events:
import { listen } from '@tauri-apps/api/event';
listen(
'state-changed',
(event) => {
console.log('got state changed event', event);
},
{
target: { kind: 'Any' },
}
);
:::
To learn how to listen to events and emit events from your Rust code, see the Rust Event System documentation.