libxrk 0.12.0

Library for reading AIM XRK files from AIM automotive data loggers

libxrk

A Python library for reading AIM XRK and XRZ files from AIM automotive data loggers.

Features

• Read AIM XRK files (raw data logs)
• Read AIM XRZ files (zlib-compressed XRK files)
• Parse track data and telemetry channels
• GPS coordinate conversion and lap detection
• High-performance Cython implementation
• Supports Python 3.10 - 3.14

Installation

Install from PyPI

pip install libxrk

Install from Source

Prerequisites

On Ubuntu/Debian:

sudo apt install build-essential python3-dev

Install with uv

uv sync

The Cython extension will be automatically compiled during installation.

Usage

from libxrk import aim_xrk

# Read an XRK file
log = aim_xrk('path/to/file.xrk')

# Read an XRZ file (automatically decompressed)
log = aim_xrk('path/to/file.xrz')

# Access channels (each channel is a PyArrow table with 'timecodes' and value columns)
for channel_name, channel_table in log.channels.items():
    print(f"{channel_name}: {channel_table.num_rows} samples")

# Get all channels merged into a single PyArrow table
# (handles different sample rates with interpolation/forward-fill)
merged_table = log.get_channels_as_table()
print(merged_table.column_names)

# Convert to pandas DataFrame
df = merged_table.to_pandas()

# Access laps (PyArrow table with 'num', 'start_time', 'end_time' columns)
print(f"Laps: {log.laps.num_rows}")
for i in range(log.laps.num_rows):
    lap_num = log.laps.column("num")[i].as_py()
    start = log.laps.column("start_time")[i].as_py()
    end = log.laps.column("end_time")[i].as_py()
    print(f"Lap {lap_num}: {start} - {end}")

# Access metadata
print(log.metadata)
# Includes: Driver, Vehicle, Venue, Log Date/Time, Logger ID, Logger Model, Device Name, etc.

Filtering and Resampling

from libxrk import aim_xrk

log = aim_xrk('session.xrk')

# Select specific channels
gps_log = log.select_channels(['GPS Latitude', 'GPS Longitude', 'GPS Speed'])

# Filter to a time range (milliseconds, inclusive start, exclusive end)
segment = log.filter_by_time_range(60000, 120000)

# Filter to a specific lap
lap5 = log.filter_by_lap(5)

# Combine filtering and channel selection
lap5_gps = log.filter_by_lap(5, channel_names=['GPS Latitude', 'GPS Longitude'])

# Resample all channels to match a reference channel's timebase
aligned = log.resample_to_channel('GPS Speed')

# Resample to a custom timebase
import pyarrow as pa
target = pa.array(range(0, 100000, 100), type=pa.int64())  # 10 Hz
resampled = log.resample_to_timecodes(target)

# Chain operations for analysis workflows
df = (log
    .filter_by_lap(5)
    .select_channels(['Engine RPM', 'GPS Speed'])
    .resample_to_channel('GPS Speed')
    .get_channels_as_table()
    .to_pandas())

All filtering and resampling methods return new LogFile instances (immutable pattern), enabling method chaining for complex analysis workflows.

Channel Metadata

Each channel carries typed metadata accessible via ChannelMetadata:

from libxrk import aim_xrk, ChannelMetadata

log = aim_xrk('session.xrk')

# Extract typed metadata from a channel
meta = ChannelMetadata.from_channel_table(log.channels['Engine RPM'])
print(meta.units)        # "rpm"
print(meta.dec_pts)      # 0
print(meta.interpolate)  # True
print(meta.function)     # "Engine RPM"

# Or from a PyArrow field directly
field = log.channels['Engine RPM'].schema.field('Engine RPM')
meta = ChannelMetadata.from_field(field)

# Create metadata for custom channels
meta = ChannelMetadata(units="m/s", dec_pts=1, interpolate=True)
field = pa.field("speed", pa.float32(), metadata=meta.to_field_metadata())

Available fields: units, dec_pts, interpolate, function, source_type, source_channel_id, device_tag, cal_value_1, cal_value_2, display_range_min, display_range_max.

Development

Quick Check

# Run all quality checks (format check, type check, tests)
just check

Code Formatting

This project uses Black for code formatting.

# Format all Python files
just format

Type Checking

This project uses mypy for static type checking.

# Run type checker on all Python files
just typecheck

Running Tests

This project uses pytest for testing.

# Run all tests
just test

# Run specific test file
uv run pytest tests/test_xrk_loading.py

# Run tests with coverage
uv run pytest --cov=libxrk

Testing with Pyodide (WebAssembly)

You can test the library in a WebAssembly environment using Pyodide. This requires Node.js to be installed.

# Build and run tests in Pyodide 0.27.x (Python 3.12)
just pyodide-test

# Build and run tests in Pyodide 0.29.x (Python 3.13, requires pyenv)
just pyodide-test-0-29

Note: Pyodide tests for both 0.27.x and 0.29.x run automatically in CI via GitHub Actions.

Building

# Build CPython wheel and sdist
uv build

# Build all wheels (CPython, Pyodide/WebAssembly, and sdist)
just build-all

Clean Build

# Clean all build artifacts and rebuild
rm -rf build/ dist/ src/libxrk/*.so && uv sync

Testing

The project includes end-to-end tests that validate XRK and XRZ file loading and parsing.

Test files are located in tests/test_data/ and include real XRK and XRZ files for validation.

Credits

This project incorporates code from TrackDataAnalysis by Scott Smith, used under the MIT License.

License

MIT License - See LICENSE file for details.