cu29-export

Copper log export helpers and Python-facing log readers.

This crate is for offline analysis of .copper logs. That distinction matters: using Python here is fine because it does not put Python on the runtime hot path.

If you want to run task logic in Python, that is a different feature entirely: see cu-python-task.

What This Crate Provides

• the run_cli::<P>() logreader entrypoint used by Copper examples and templates
• structured log export helpers
• CopperList export helpers
• optional Python bindings for iterating logs without going through JSON first

Python Support

Python support lives behind the python feature and is not supported on macOS in this workspace.

There are two Python-facing patterns:

1. Generic structured log reading

libcu29_export can expose:

• struct_log_iterator_bare(...)
• struct_log_iterator_unified(...)
• runtime_lifecycle_iterator_unified(...)

This is useful when you want Python to inspect Copper's structured text logs or runtime lifecycle records.

The example script at examples/cu_standalone_structlog/readlog.py shows the basic import pattern.

2. App-specific typed CopperList reading

CopperLists are application-specific, so a Python module that reads them must know the generated tuple type for that application.

The intended pattern is:

1. call gen_cumsgs!("copperconfig.ron") in the application
2. expose a small #[pymodule] wrapper in that app
3. call copperlist_iterator_unified_typed_py::<YourGeneratedType>(...)

See examples/cu_flight_controller/src/python_module.rs and examples/cu_flight_controller/python/print_gnss_from_log.py for the reference implementation.

Feature Flags

• python: Rust-side helpers for embedding/exposing Python log readers
• python-extension-module: only for building the Python extension itself
• mcap: MCAP export support

Recommendation

Use Python here for post-processing, data mining, notebooks, and analysis scripts. That is a reasonable workflow.

Do not confuse that with putting Python inside a Copper control loop. Offline export and runtime Python tasks have very different tradeoffs.