Skip to main content
Java compilation and validation tool using execute_in_sandbox.

​
react_agent.custom_tools.java_validator.JavaValidationInput Objects

View Source Pydantic input model for specifying Java validation parameters. Attributes:
  • source_code - The Java source code (either schema mapping or query) to compile and run.
  • framework - The specific Java target framework (e.g., Spring Data MongoDB, Spring Data Neo4j).
  • entry_type_name - The public class name of the entry point declared in the source code.

​
react_agent.custom_tools.java_validator.compile_and_run_java

View Source Compile and execute Java code within an isolated Daytona Sandbox container. This utility handles the complex orchestration of validating Java code:
  1. Fetches the required pom.xml configurations for Spring Data.
  2. Builds a bash script (run.sh) to compile via Maven and execute the specified entry point class.
  3. Provisions environment variables and database connection strings.
  4. Base64 encodes all payload files to safely transfer them over to the sandbox.
  5. Invokes the execute_in_sandbox tool.
  6. If a query output JSON was produced, it downloads it via download_file_from_sandbox.
Arguments:
  • source_code str - The raw Java source code to compile and run.
  • framework JavaFramework - The target Java framework (Spring Data MongoDB, Spring Data Neo4j).
  • entry_type_name str - The name of the main class containing the public static void main method.
  • runtime ToolRuntime[Context, State] - The LangGraph tool runtime containing context and state.
Returns: tuple[str, str | None]: A tuple containing the standard output/error log of the execution, and the raw JSON string of the query results (if applicable and successful). Raises:
  • RuntimeError - If fetching the framework config fails or the sandbox execution throws an exception.

​
react_agent.custom_tools.java_validator.validate_java_code

View Source LangChain Tool wrapper to execute and evaluate Java code inside a Sandbox. This tool acts as a bridge between the LangGraph state machine and the low-level sandbox execution. It calls compile_and_run_java and then parses the returned JSON results. Crucially, it determines whether it is currently validating the “source” query or the “target” query based on the tool call ID or state comparison, and dynamically issues a LangGraph Command to update the global state with the parsed QueryValidationResults. Arguments:
  • source_code str - The Java source code to validate.
  • framework JavaFramework - The target Java framework.
  • entry_type_name str - The entrypoint class name declared in the Java code.
  • runtime ToolRuntime[Context, State] - Automatically injected runtime context.
Returns: Command | str: A LangGraph state update command containing parsed results and the execution log, or just the error string if validation fails.