Struct MontyReplExecutor 

pub struct MontyReplExecutor { /* private fields */ }

Available on crate feature code only.

REPL Monty executor: interpreter state (variables, function definitions, imports) persists across execute calls.

Build with MontyExecutorBuilder::build_repl. The executor stores the serialized interpreter between calls (Monty’s types are not designed to be held across await points), guarded by a Mutex so snippets apply in submission order. Monty preserves the session through Python-level exceptions, so a failed snippet does not destroy accumulated state.

Lifecycle

start initializes an empty interpreter session, stop drops it, and restart resets it. Calling execute() before start() lazily initializes the session.

Policy consistency

A session’s effective OS policy must not vary between calls (a mount visible in call 1 must not silently vanish in call 2 while state persists). The first call’s effective policy is captured; a later call whose policy differs is rejected with guidance to restart().

Example

use adk_code::{
    CodeExecutor, ExecutionLanguage, ExecutionPayload, ExecutionRequest,
    MontyExecutorBuilder, SandboxPolicy,
};

let executor = MontyExecutorBuilder::new().build_repl()?;
let request = |code: &str| ExecutionRequest {
    language: ExecutionLanguage::Python,
    payload: ExecutionPayload::Source { code: code.to_string() },
    argv: vec![],
    stdin: None,
    input: None,
    sandbox: SandboxPolicy::strict_python(),
    identity: None,
};
executor.execute(request("x = 41")).await?;
let result = executor.execute(request("x + 1")).await?;
assert_eq!(result.output, Some(serde_json::json!(42)));

Implementations

impl MontyReplExecutor

pub fn granted_policy(&self) -> SandboxPolicy

Available on crate feature embedded-python only.

A SandboxPolicy requesting exactly this executor’s grants.

See MontyOneShotExecutor::granted_policy.

Trait Implementations

impl CodeExecutor for MontyReplExecutor

fn start<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = Result<(), ExecutionError>> + Send + 'async_trait>>

where 'life0: 'async_trait, MontyReplExecutor: 'async_trait,

Initialize an empty interpreter session. A no-op if one exists.

fn stop<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = Result<(), ExecutionError>> + Send + 'async_trait>>

where 'life0: 'async_trait, MontyReplExecutor: 'async_trait,

Drop the interpreter session (variables, functions, imports).

fn name(&self) -> &str

Human-readable backend name.

fn capabilities(&self) -> BackendCapabilities

The capabilities this backend can enforce.

fn supports_language(&self, lang: &ExecutionLanguage) -> bool

Whether this backend supports the given language.

fn prompt_snippet(&self) -> Option<String>

A prose snippet describing this backend’s execution environment and built-in capabilities, suitable for appending to an LLM-facing tool description.

fn execute<'life0, 'async_trait>( &'life0 self, request: ExecutionRequest, ) -> Pin<Box<dyn Future<Output = Result<ExecutionResult, ExecutionError>> + Send + 'async_trait>>

where 'life0: 'async_trait, MontyReplExecutor: 'async_trait,

Execute a request and return a structured result.

fn is_running<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = bool> + Send + 'async_trait>>

where 'life0: 'async_trait, MontyReplExecutor: 'async_trait,

Whether the execution environment is currently running.

fn restart<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = Result<(), ExecutionError>> + Send + 'async_trait>>

where 'life0: 'async_trait, Self: 'async_trait,

Restart the execution environment.

impl Debug for MontyReplExecutor

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

Formats the value using the given formatter.