> ## Documentation Index
> Fetch the complete documentation index at: https://uom-demo.vercel.app/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# dotnet_validator

> [<span className="flex items-center gap-1"><Icon icon="external-link"/>View Source</span>](https://github.com/corovcam/Universal-Object-Mapping/tree/main/services/orchestrator/src/react_agent/custom_tools/dotnet_validator.py)

Dotnet compilation and validation tool using execute\_in\_sandbox.

## react\_agent.custom\_tools.dotnet\_validator.DotnetValidationInput Objects

```python theme={null}
class DotnetValidationInput(BaseModel)
```

[<span className="flex items-center gap-1"><Icon icon="external-link" />View Source</span>](https://github.com/corovcam/Universal-Object-Mapping/blob/main/services/orchestrator/src/react_agent/custom_tools/dotnet_validator.py#L44)

Pydantic input model for specifying .NET validation parameters.

**Attributes**:

* `source_code` - The C# source code (either schema mapping or query) to compile and run.
* `framework` - The specific .NET target framework (e.g., EF Core, Dapper, NHibernate).

#### react\_agent.custom\_tools.dotnet\_validator.compile\_and\_run\_dotnet

```python theme={null}
async def compile_and_run_dotnet(
        source_code: str, framework: DotnetFramework,
        runtime: ToolRuntime[Context, State]) -> tuple[str, str | None]
```

[<span className="flex items-center gap-1"><Icon icon="external-link" />View Source</span>](https://github.com/corovcam/Universal-Object-Mapping/blob/main/services/orchestrator/src/react_agent/custom_tools/dotnet_validator.py#L56)

Compile and execute C# code within an isolated Daytona Sandbox container.

This complex utility performs several steps to validate .NET code:

1. Generates the appropriate .csproj file based on the target framework.
2. Builds a comprehensive bash script (`run.sh`) to compile and execute the code.
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 C# source code to compile and run.
* `framework` *DotnetFramework* - The target .NET framework (EF Core, Dapper, etc.).
* `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.dotnet\_validator.validate\_dotnet\_code

```python theme={null}
@tool("validate_dotnet_code", args_schema=DotnetValidationInput)
async def validate_dotnet_code(
        source_code: str, framework: DotnetFramework,
        runtime: ToolRuntime[Context, State]) -> Command | str
```

[<span className="flex items-center gap-1"><Icon icon="external-link" />View Source</span>](https://github.com/corovcam/Universal-Object-Mapping/blob/main/services/orchestrator/src/react_agent/custom_tools/dotnet_validator.py#L265)

LangChain Tool wrapper to execute and evaluate .NET 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_dotnet` 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 C# source code to validate.
* `framework` *DotnetFramework* - The target .NET framework.
* `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.