MCAP Bag Parser

Parse MCAP rosbags into pandas dataframes using an adapter around rosx_introspection for speed.

Installation

pip install mcap-bag-parser

Note: The mcap-bag-parser is now a fully self-contained package using scikit-build-core, for automatically building/bundling rosx_introspection during pip install. Users now get a unified mcap_bag module with both legacy (BagFileParser) and modern (parse_mcap_file) APIs, with scikit-build-core handling C++ compilation and Python extension creation.

Development

# Clone with submodules
git clone --recurse-submodules https://gitlab.com/nealtanner/mcap-bag-parser.git
cd mcap-bag-parser

# Install in editable mode with test dependencies (pip >= 25.1)
pip install -e . --group test

# Run tests
pytest

For isolated development alongside other versions:

python -m venv .venv && source .venv/bin/activate
pip install -e . --group test

Rebuilding after C++ changes

# Update code and submodules
git pull --recurse-submodules  # Or: git pull && git submodule update --init

# Force rebuild (cleans build cache)
pip install --force-reinstall --no-deps -e .

# Or clean everything and rebuild
rm -rf build/ build_python/
pip install -e .

Publishing to PyPI

Important: PyPI rejects linux_x86_64 and linux_aarch64 wheels. Must use cibuildwheel for manylinux tags.

pip install build twine cibuildwheel
rm -rf dist/ wheelhouse/  # Clean old builds

# Linux - builds manylinux wheels in Docker (required for PyPI)
cibuildwheel --platform linux  # All architectures
# Or specific: --archs x86_64 or --archs aarch64

# macOS - builds universal2 wheels
cibuildwheel --platform macos

# Source distribution
python -m build --sdist

# Upload all
twine upload wheelhouse/*.whl dist/*.tar.gz  # Username: __token__

Supported platforms:

• Linux: x86_64, aarch64 (manylinux_2_28, glibc 2.28+)
• macOS: x86_64, arm64 (11.0+)
• Python: 3.10, 3.11, and 3.12+ (a single stable-ABI abi3 wheel covers 3.12 and later)