A high-performance inference engine for AI models
use anyhow::Result;
use crate::{
configs::{Paths, PlatformsConfig},
languages::{LanguageBackend, LanguageBackendTarget},
types::{Command, Configuration, Language},
};
pub struct RustLanguageBackend {
config: PlatformsConfig,
}
impl RustLanguageBackend {
pub fn new(config: PlatformsConfig) -> Self {
Self {
config,
}
}
}
impl LanguageBackend for RustLanguageBackend {
fn config(&self) -> PlatformsConfig {
self.config.clone()
}
fn language(&self) -> Language {
Language::Rust
}
fn expects_prebuild_for_run(&self) -> bool {
false
}
fn build_targets(
&self,
configuration: Configuration,
targets: Vec<LanguageBackendTarget>,
) -> Result<()> {
let paths = Paths::new()?;
for target in targets {
Command::cargo_build(paths.main_crate.clone(), target.name.clone(), target.features.clone(), configuration)
.with_envs(self.config.required_envs_for_target(target.name.clone())?)
.run()?;
}
Ok(())
}
fn test_target(
&self,
configuration: Configuration,
target: LanguageBackendTarget,
) -> Result<()> {
Command::cargo_test(target.name.clone(), target.features.clone(), configuration).run()
}
fn example_target(
&self,
name: &str,
configuration: Configuration,
target: LanguageBackendTarget,
) -> Result<()> {
let paths = Paths::new()?;
let name = self.language().convert_command_name(name);
Command::cargo_run_example(
paths.main_crate.clone(),
name,
target.name.clone(),
target.features.clone(),
configuration,
)
.run()
}
}