22 releases

0.9.1 Dec 15, 2023
0.9.0-alpha4 Jan 8, 2023
0.9.0-alpha3 Sep 15, 2022
0.9.0-alpha1 Jul 24, 2022
0.2.0 Oct 28, 2020

#52 in FFI

Download history 46/week @ 2024-06-10 47/week @ 2024-06-17 16/week @ 2024-06-24 15/week @ 2024-07-08 27/week @ 2024-07-15 13/week @ 2024-07-22 7/week @ 2024-07-29 38/week @ 2024-08-05 13/week @ 2024-08-12 16/week @ 2024-08-19 30/week @ 2024-08-26 235/week @ 2024-09-02 1/week @ 2024-09-09 8/week @ 2024-09-16 46/week @ 2024-09-23

298 downloads per month
Used in 4 crates (2 directly)

MIT/Apache

510KB
4K SLoC

tealr

A crate to enhance the APIs provided by the rlua and mlua crates

It aims to do this by improving the following:

  • Allow the api to have easily accessible documentation embedded into it
  • Allow the documentation to be built to web pages (using tealr_doc_gen )
  • To go along with the documentation, tealr also allow you to be more precise in the types your api works with. Think generic methods and typed lambdas. No more Lua::Value
  • Add macros to make it easier to work with teal, a statically typed dialect of lua.

It does this by adding new traits and replacing/extending the existing ones from rlua and mlua. As a result, the api that tealr exposes is as similar as the api from those 2 crates as possible.

It also contains some macro's to easily generate new types to better express the API type wise.

Example of instance.help()

The library shown is https://github.com/lenscas/tealsql

https://github.com/lenscas/tealr/tree/master/tealr/images/help_example.gif

html rendered documentation

Rendered html is also available at https://lenscas.github.io/tealsql/

Note:

Both rlua and mlua are behind the feature flags rlua and mlua.

Tealr reexports these crates and allows you to set flags through it (the forwarded flags are the same with either the prefix rlua_ or mlua_. For example if you want to enable mlua/async then you need to enable tealr/mlua_async).

Please, do not set feature flags directly in mlua/rlua and instead set them through tealr. The API of these crates can change depending on what feature flags are set and tealr needs to be made aware of those changes.

Expose a value to lua/teal

Exposing types to lua as userdata is almost the same using tealr as it is using rlua and mlua

Rlua:

use tealr::ToTypename;
#[derive(Clone, tealr::rlu::UserData, ToTypename)]
struct ExampleRlua {}

//now, implement rlu::TealData.
//This tells rlua what methods are available and tealr what the types are
impl tealr::rlu::TealData for ExampleRlua {
    //implement your methods/functions
    fn add_methods<'lua, T: tealr::rlu::TealDataMethods<'lua, Self>>(methods: &mut T) {
        methods.document_type("This is documentation added to the type itself.");

        methods.document("This documentation gets added to the exposed function bellow.");
        methods.add_method("example_method", |_, _, x: i8| Ok(x));
        methods.add_method_mut("example_method_mut", |_, _, x: (i8, String)| Ok(x.1));
        methods.add_function("example_function", |_, x: Vec<String>| Ok((x, 8)));
        methods.document("***You*** can also embed markdown to the documentation, which gets picked up by [tealr_doc_gen](https://github.com/lenscas/type_generator)`");
        methods.document("It is also possible to use this function multiple times. These are added as paragraphs.");
        methods.add_function_mut("example_function_mut", |_, x: (bool, Option<ExampleRlua>)| {
            Ok(x)
        });
        ///This creates the instance.help() function, which returns the documentation as a string.
        methods.generate_help();
    }
}

Mlua:

use tealr::ToTypename;
#[derive(Clone, tealr::mlu::UserData, ToTypename)]
struct ExampleMlua {}
impl<'lua> FromLua<'lua> for ExampleMlua {
    fn from_lua(value: mlua::prelude::LuaValue<'lua>, _: &'lua Lua) -> Result<Self> {
        value
            .as_userdata()
            .map(|x| x.take())
            .unwrap_or(Err(mlua::Error::FromLuaConversionError {
                from: value.type_name(),
                to: "Example",
                message: None,
            }))
    }
}

impl tealr::mlu::TealData for ExampleMlua {
    //implement your methods/functions
    fn add_methods<'lua, T: tealr::mlu::TealDataMethods<'lua, Self>>(methods: &mut T) {
        methods.document_type("This is documentation added to the type itself.");
        methods.document("This documentation gets added to the exposed function bellow.");
        methods.add_method("example_method", |_, _, x: i8| Ok(x));
        methods.add_method_mut("example_method_mut", |_, _, x: (i8, String)| Ok(x.1));
        methods.add_function("example_function", |_, x: Vec<String>| Ok((x, 8)));
        methods.document("***You*** can also embed markdown to the documentation, which gets picked up by [tealr_doc_gen](https://github.com/lenscas/type_generator)`");
        methods.document("It is also possible to use this function multiple times. These are added as paragraphs.");
        methods.add_function_mut("example_function_mut", |_, x: (bool, Option<ExampleMlua>)| {
            Ok(x)
        });
        ///This creates the instance.help() function, which returns the documentation as a string.
        methods.generate_help();
    }
}

Replacing lua::Value with better type information

Though it is perfectly possible to use the lua::Value from rlua and mlua they aren't the most descriptive type wise. Using them will hurt your documentation as a result.

To help avoid lua::Value tealr comes with new types and macros that help you define your API better type wise.

Simple unions:

These allow you to easily create a type that is only one of the types you give.

use tealr::{
    create_union_mlua,
};
create_union_mlua!(enum YourTypeName = i32 | String);

Typed functions:

Though the normal function type from both mlua and rlua is perfectly useable it doesn't contain contain any type information. To help add more type information to your api tealr comes with its own version of this function type that contains type information.

use tealr::{
    mlu::{
        mlua::Lua,
        TypedFunction
    },
}

let lua = mlua::Lua::new();
let add_1 = TypedFunction::<u8, u8>::from_rust(|_lua, x| Ok(x + 1), &lua)?;

assert_eq!(add_1.call(2)?, 3);

Generics

To go along with typed functions, tealr also comes with a way to mimic generics. Though they at first glance will just look like another way to use lua::Value due to not being able to put bounds on the generic, they are still very useful to properly model how input and output rely on each other.

In the following example we take a generic function and call it, returning whatever it returned back to lua. Thanks to the use of generics, it i clear that the return type of the method is equal to the return type of the lambda. If lua::Value was used instead this was not clear.

use mlua::IntoLua;
use tealr::{
    create_generic_mlua,
    mlu::{mlua::FromLua, TealData, TealDataMethods, TypedFunction,UserData},
    ToTypename, TypeWalker,
};

create_generic_mlua!(X);

#[derive(Clone, UserData, ToTypename)]
struct Example {}

impl TealData for Example {
    fn add_methods<'lua, T: TealDataMethods<'lua, Self>>(methods: &mut T) {
        methods.add_method(
            "generic_function_callback",
            |lua, _, fun: TypedFunction<String, X>| {
                fun.call("A nice string!".to_string())
            },
        );
    }
}

impl<'lua> FromLua<'lua> for Example {
    fn from_lua(value: mlua::prelude::LuaValue<'lua>, _: &'lua Lua) -> Result<Self> {
        value
            .as_userdata()
            .map(|x| x.take())
            .unwrap_or(Err(mlua::Error::FromLuaConversionError {
                from: value.type_name(),
                to: "Example",
                message: None,
            }))
    }
}


For rlua, all you have to do is replace mlua for rlua

Teal integration

The teal language is basically just a statically typed variant of lua and can even be made to run in the lua vm without compiling to lua first.

As a result of this and tealr's focus on enabling a richer typed api causes the 2 projects to work well together. However, to further help bind the 2 projects, tealr contains some extra helpers for those that want to use teal.

Compile inline teal code into lua

Both rlua and mlua allow you to run lua code embedded in your application.

Similarly, tealr allows you to compile embedded teal code to lua while compiling your application. This can then be executed by rlua and mlua.

This means that you can make use of teal's static type system even for small scripts inside your rust codebase.

use tealr::compile_inline_teal;
let code = compile_inline_teal!("local x : number = 5 return x");

Embed the teal compiler

Teal makes it possible for the lua vm to load teal files as if they are normal lua files.

Tealr makes doing this from withing rust a bit easier, by exposing a macro that can embed the teal compiler in your application and create a function that creates the needed lua code to set the VM up. This function takes a string, which is the file that needs to get required.

use tealr::embed_compiler;
let compiler = embed_compiler!("v0.13.1");
#[cfg(feature = "rlua")]
{
    let res : u8 = tealr::rlu::rlua::Lua::new().context(|ctx| {
        let code = compiler("example/basic_teal_file");
        ctx.load(&code).set_name("embedded_compiler")?.eval()
    })?;
};

#[cfg(feature = "mlua")]
{
    let code = compiler("example/basic_teal_file");
    let lua = tealr::mlu::mlua::Lua::new();
    let res: u8 = lua.load(&code).set_name("embedded_compiler").eval()?;
};
Ok::<(), Box<dyn std::error::Error>>(())

There are a few sources tealr can use to get the compiler. If no source is specified it defaults to github releases. Other sources can be specified as follows:

//get the teal compiler using the given path
embed_compiler!(Local(path = "some/path/to/tl.tl"));
//this uses luarocks to try and discover the location of the compiler
embed_compiler!(Local());
//download the compiler at compile time from github (default)
embed_compiler!(GitHub(version = "v0.13.1"));
//download the compiler at compile time from luarocks
embed_compiler!(Luarocks(version = "v0.13.1"));

You can find longer ones with comments on what each call does here

Dependencies

~2–13MB
~173K SLoC