Crates.io | mavspec |
lib.rs | mavspec |
version | |
source | src |
created_at | 2023-12-06 19:22:58.406237+00 |
updated_at | 2025-03-15 19:53:43.543283+00 |
description | A set of code generation utilities for MAVLink protocol. |
homepage | https://mavka.gitlab.io/home/projects/mavspec |
repository | https://gitlab.com/mavka/libs/mavspec |
max_upload_size | |
id | 1060337 |
Cargo.toml error: | TOML parse error at line 22, column 1 | 22 | autolib = false | ^^^^^^^ unknown field `autolib`, expected one of `name`, `version`, `edition`, `authors`, `description`, `readme`, `license`, `repository`, `homepage`, `documentation`, `build`, `resolver`, `links`, `default-run`, `default_dash_run`, `rust-version`, `rust_dash_version`, `rust_version`, `license-file`, `license_dash_file`, `license_file`, `licenseFile`, `license_capital_file`, `forced-target`, `forced_dash_target`, `autobins`, `autotests`, `autoexamples`, `autobenches`, `publish`, `metadata`, `keywords`, `categories`, `exclude`, `include` |
size | 0 |
A code-generator for MAVLink.
MAVLink is a lightweight open protocol for communicating between drones, onboard components and ground control stations.
It is used by such autopilots like PX4 or ArduPilot. MAVLink has simple and
compact serialization model. The basic abstraction is message
which can be sent through a link (UDP, TCP, UNIX
socket, UART, whatever) and deserialized into a struct with fields of primitive types or arrays of primitive types.
Such fields can be additionally restricted by enum
variants, annotated with metadata like units of measurements,
default or invalid values. There are several MAVLink dialects. Official dialect definitions are
XML files that can be found in the MAVlink
repository. Based on message
abstractions,
MAVLink defines so-called microservices
that specify how clients should respond on
a particular message under certain conditions or how they should initiate a particular action.
This library is a building block for other MAVLink-related tools (telemetry collectors, IO, etc.). It is only responsible for code generation. Other Mavka projects are focused on their own areas:
MAVSpec
is using this library to discover and parse MAVLink dialects.no-std
(and no-alloc
) targets and focuses on stateless parts of MAVLink protocol.Mavio
that
provides a high-level interface for MAVLink messaging and takes care about stateful features of the protocol:
sequencing, message time-stamping, automatic heartbeats, simplifies message signing, and so on.This project respects semantic versioning
.
Install as a Cargo dependency.
cargo add mavspec --features all
The all
feature will generate all standard MAVLink dialects.
In case you want to generate code as a part of you build sequence, we suggest to also add MAVSpec as a build dependency.
cargo add --build mavspec --featurs generators
The generators
feature enables all code-generators.
MAVSpec is designed to be both feature-rich and flexible. The following is just a short guide for the common use-cases. If you want a deep understanding, check API docs.
The following explains how to use library API, for command-line tool usage check CLI section.
Standard MAVLink dialects are generated from the corresponding
XML specifications and are available in
mavspec::rust::microservices
module. To use bundled dialects just install MAVSpec with the specified dialects
features:
cargo add mavspec --featurs all
Available dialects are:
minimal
β minimal dialect required to
expose your presence to other MAVLink devices.standard
β a superset of minimal
dialect,
that expected to be used by almost all flight stack.common
β minimum viable dialect with most of
the features, a building block for other future-rich dialects.ardupilotmega
β feature-full dialect used by
ArduPilot. In most cases this dialect is the go-to choice if you want
to recognize almost all MAVLink messages used by existing flight stacks.all
β meta-dialect which includes all other
standard dialects including those which were created for testing purposes. It is guaranteed
that namespaces of the dialects in all
family do not collide.asluav
, avssuas
, csairlink
, cubepilot
, development
, icarous
, matrixpilot
,
paparazzi
, ualberta
, uavionix
. These do not include python_array_test
and test
dialects which should be either generated manually or as a part of all
meta-dialect.When standard MAVLink dialects are used (at least minimal
Cargo feature is enabled), this
library exposes default_dialect
and DefaultDialect
entities that allow to access the most feature-rich enabled
MAVLink dialect.
You might think about default dialect as a sequence of smaller and smaller dialects where all
it the most rich dialect
and minimal
provides a bare minimum of functionality:
all
>
ardupilotmega
>
common
>
standard
>
minimal
MAVSpec allows to generate additional structures tailored for MAVLink microservices.
Each microservice is a subdialect with only those messages and enums which are necessary. To generate microservice
subdialects use msrv-*
feature flags.
β οΈ We do not recommend to enable microservices for libraries that perform generic MAVLink operations as this may include compilation time.
MAVSpec also provides additional utils to work with MAVLink microservices. These tools can be
enabled by msrv-utils-*
feature flags and available in mavspec::rust::microservices
module.
β οΈ
msrv-utils-*
are considered unstable for now! Useunstable
feature flag to enable them.
It is possible to bundle metadata generated by MAVInspect into meta
module.
This can be useful for ground control stations that require to present the user with the descriptions of MAVLink
entities.
To enable metadata bundling use metadata
feature flag.
β οΈ Metadata available only with
std
feature enabled. Otherwise, this will cause build to fail.
API documentation for Rust code-generation can be found here.
Add MAVSpec with rust
feature to your dependencies.
cargo add mavspec --features rust
This feature enables interfaces upon which your generated code will depend. You can access these interfaces through
use mavspec::rust::spec
.
Optionally enable std
(for Rust standard library) or alloc
(for memory allocation support) features if your target
supports them (if you are not developing for an embedded devices, then we suggest to always enable std
).
Add MAVSpec with rust_gen
as a build dependency:
cargo add --build mavspec --features rust_gen
If necessary, add optional section to your Cargo.toml
to generate only specific MAVLink entities:
[package.metadata.mavspec]
microservices = ["HEARTBEAT", "MISSION"]
messages = ["PROTOCOL_VERSION", "MAV_INSPECT_V1", "PING"]
enums = ["STORAGE_STATUS", "GIMBAL_*"]
commands = ["MAV_CMD_DO_CHANGE_SPEED", "MAV_CMD_DO_SET_ROI*"]
generate_tests = false
This will greatly reduce compile time and may slightly reduce memory footprint (if you are not going to expose autogenerated code as a part of your library API, then Rust compiler will probably optimize away all unused pieces).
If you want to generate tests for generated code, set generate_tests
to true
. This mode is disabled by default.
Update your build.rs
:
use std::env::var;
use std::path::Path;
use mavspec::rust::BuildHelper;
fn main() {
// Assume that your library and `message_definitions` are both in the root of your project.
let sources = vec![
"./message_definitions/standard",
"./message_definitions/extra",
];
// Output path
let destination = Path::new(&var("OUT_DIR").unwrap()).join("mavlink");
// Path to your `Cargo.toml` manifest
let manifest_path = Path::new(env!("CARGO_MANIFEST_DIR")).join("Cargo.toml");
// Parse XML definitions and generate Rust code
BuildHelper::builder(&destination)
.set_sources(&sources)
.set_manifest_path(&manifest_path)
.generate()
.unwrap();
}
The OUT_DIR
environment variable is provided by Rust build toolchain and points to output library for your crate. It
is considered a bad practice to write outside this path in the build scripts.
Finally, import generated code in your lib.rs
(or anywhere it seems appropriate):
mod mavlink {
include!(concat!(env!("OUT_DIR"), "/mavlink/mod.rs"));
}
pub use mavlink::dialects;
Check examples/rust
for a slightly more elaborated example which uses Cargo features as
flags for MAVLink
dialect selection.
In MAVSpec
we are trying to keep balance between names as they appear in MAVLink XML definitions and Rust naming
conventions. In most situation we favor the Rust way unless it introduces confusions. In case we failed, and you are
confused, all entities are supplemented with descriptions where canonical MAVlink names are mentioned. Here is the list
of the naming rules:
MAVSpec
uses UpperCamelCase
.snake_case
.UpperCamelCase
with MAVLink enum name prefix stripped
(whenever applicable). For example, if bitmask enum has name IMPORTANCE_LEVEL
and flag name is
IMPORTANCE_LEVEL_THE_MATTER_OF_LIFE_AND_DEATH
, then flag name will be TheMatterOfLifeAndDeath
.SCREAMING_SNAKE_CASE
with MAVLink enum name
prefix stripped (whenever applicable). For example, if bitmask enum has name VERY_IMPORTANT_FLAGS
and flag name is
VERY_IMPORTANT_FLAGS_THE_MATTER_OF_LIFE_AND_DEATH_FLAG
, then flag name will be THE_MATTER_OF_LIFE_AND_DEATH_FLAG
.type
field of HEARTBEAT
message will be encoded as type_
._
.Check mavspec_examples_rust.rs
which shows how the last two cases of
inconvenient names are handled (this is not something of high aesthetic value but in our defence we must say that all
approaches we've considered looked equally ugly).
MAVInspect may skip code re-generation if dialects haven't changed. It uses 64-bit CRC fingerprint to monitor
changes. Set fingerprints
feature flag to enable this behavior.
This feature is useful for reducing build time during development and CI runs. Make sure that your releases are clean and do not depend on fingerprints.
Unstable features are enabled by unstable
feature flag. Such features are experimental and can be changed or
excluded in future releases.
Install mavspec
command-line tool.
cargo install mavspec --features cli
Check installation:
mavspec -V
If you are working from the MAVSpec
repository, then you always can run CLI-tool using cargo:
cargo run --bin mavspec --features cli --
Parse XML definitions from ./message_definitions/standard
and generate dialects in
tmp/mavlink
directory:
mavspec --src message_definitions/standard --out tmp/mavlink rust
Print mavspec
help for Rust code generator:
mavspec rust -h
examples/build_custom_dialects
β an example library with autogenerated code.
cargo run --package mavspec_examples_custom_dialects --bin custom_dialects_example
API is considered relatively stable but certain advanced features are yet to be developed. However, most of these features are nice to have, rather than something necessary to consider this library complete.
Milestone v1
contains features considered necessary to
reach stable version 1.0.0
. Most of these features are related to Rust code generator.
Other code generators (will form a basis for other Mavka projects):
telemetry
milestone is focused on code generators required
for storing MAVLink data in time-series databases like InfluxDB or
Timescale.gRPC
milestone is reserved for
Protobuf and gRPC bindings.Propositions and pull-requests are welcomed.
First of all, there is an official MAVLink client for Rust worth mentioning:
rust-mavlink
. One of the reasons behind writing this library was my desire
to decouple parser and code generator into the separate projects.
I was personally inspired by gomavlib
library for MAVLink (Go). I like the
way it is written, and its source code helped me in several cases when official MAVLink documentation wasn't clear
enough.
If you want to autogenerate language bindings and prefer Python, you might be interested in the official
mavgen
code-generation tool. If you are
looking for a router for MAVLink messages, then we suggest mavp2p
. If
you want a solution that supports MAVLink microservices, then it worth
checking MAVSDK
that uses gRPC API.
MAVLink is almost 15 years old, but the ecosystem around this protocol is still dynamic and developing. Some projects are stable and robust, while others are nice and feature-rich but incomplete.
Here we simply comply with the suggested dual licensing according to Rust API Guidelines (C-PERMISSIVE).
Licensed under either of
at your option.
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.