Struct EngineTarget 

pub struct EngineTarget {
    pub binary: PathBuf,
    pub env: Vec<(String, String)>,
    pub data_dir_env: Option<String>,
    pub data_dir_arg: Option<String>,
    pub forbidden_args: Vec<String>,
    pub allow_no_data_dir_for_refusal_probe: bool,
}

An engine binary and how to launch it without touching real user data.

Every launch (cli, run, start_sidecar) must name an isolated data directory, or it is refused before the process starts with a crate::userdata::UserDataViolation (rule no-isolated-data-dir). A directory is named when the caller’s CliOptions::data_dir reaches the engine through data_dir_env, or data_dir_env is set in the launch environment, or data_dir_arg appears in the arguments, and the directory passes crate::userdata::check_engine_dirs. A data_dir on a target that declares neither is not named: the engine would never see it.

Fields

binary: PathBuf

env: Vec<(String, String)>

Env applied to every engine process (e.g. GENRIGHT_NO_KEYCHAIN=1).

data_dir_env: Option<String>

Env var that receives the per-scenario data dir (e.g. GENRIGHT_APP_DATA).

data_dir_arg: Option<String>

Flag the engine takes its data dir under (e.g. --data-dir). Declaring it lets a caller name the isolated directory in the launch arguments; the harness only judges the value.

forbidden_args: Vec<String>

Flags that select the real library (e.g. --use-user-library). A launch whose arguments contain one, as flag or flag=value, is refused. Empty by default; the product declares its own, this crate hard-codes none.

allow_no_data_dir_for_refusal_probe: bool

Set by EngineTarget::allow_no_data_dir_for_refusal_probe; read it as a warning label.

Implementations

impl EngineTarget

pub fn new(binary: impl Into<PathBuf>) -> Self

A target with no environment, no data-dir variable or flag, and no deny list. Such a target cannot be launched until it declares data_dir_env or data_dir_arg (or the probe opt-out is used).

pub fn with_data_dir_env(self, var: &str) -> Self

Env var that receives the data dir (e.g. GENRIGHT_APP_DATA).

pub fn with_data_dir_arg(self, flag: &str) -> Self

Flag the engine takes its data dir under (e.g. --data-dir).

pub fn forbid_args(self, flags: &[&str]) -> Self

Flags that select the real library; a launch carrying one is refused.

pub fn allow_no_data_dir_for_refusal_probe(self) -> Self

Let launches of this target name no data directory, ONLY to assert the engine’s own refusal to run without one. Use it on a dedicated clone of the target, never on the target that runs real scenarios. A second barrier is still required: on Unix the launch environment (CliOptions::env or the target’s env) must redirect HOME into a QA directory, and XDG_DATA_HOME / XDG_CONFIG_HOME too wherever they are set or inherited; on Windows the Known Folder API ignores the environment, so the launch is refused whatever the caller passes. See crate::userdata::check_engine_launch.

pub fn cli( &self, args: &[String], opts: &CliOptions, tracker: &Tracker, ) -> Result<CliResult>

pub fn run( &self, words: &str, input: Value, opts: &CliOptions, tracker: &Tracker, ) -> Result<Value>

engine run('job status', input): words split, non-zero exit or json.error fails.

pub fn start_sidecar( &self, opts: &CliOptions, tracker: &Tracker, ) -> Result<Sidecar>

Trait Implementations

impl Clone for EngineTarget

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for EngineTarget

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.