Raster-threading

Multi-threaded GDAL raster read/write test harness for GeoTIFF and ERS grids.

Setup

# Install system GDAL (Ubuntu/Debian)
sudo apt-get install libgdal-dev gdal-bin

# Install Python dependencies
pip install -r requirements.txt

Running Tests

pytest test_raster_threading.py -v

Run tests with coverage:

pytest test_raster_threading.py --cov=raster_threading --cov-report=term

Overview

• raster_threading.py – helper functions for creating, reading, and writing rasters via GDAL using per-call dataset handles (thread-safe pattern).
• test_raster_threading.py – pytest harness that exercises:
  • Single-threaded create / read / write / multi-band round-trips
  • Concurrent reads of the same file from multiple threads
  • Concurrent reads of different files
  • Concurrent writes to separate files
  • Mixed concurrent readers and writers
  • Stress tests with 8 and 16 threads

All scenarios run against both GeoTIFF (.tif) and ERS (.ers) formats via parameterised fixtures.

Features

• ✅ Thread-safe GDAL operations using per-call dataset handles
• ✅ Support for GeoTIFF and ERS raster formats
• ✅ Comprehensive test suite covering concurrent reads, writes, and mixed operations
• ✅ Stress tests with 8 and 16 concurrent threads
• ✅ Multi-band raster support
• ✅ Continuous Integration via GitHub Actions

Installation

As a Library

pip install git+https://github.com/RichardScottOZ/Raster-threading.git

For Development

See CONTRIBUTING.md for detailed setup instructions.

Usage

from raster_threading import create_raster, read_raster, write_raster_band
import numpy as np

# Create a new GeoTIFF
path = "output.tif"
create_raster(path, fmt="GTiff", width=100, height=100, fill_value=0.0)

# Read raster data
data = read_raster(path)

# Write new data to a band
new_data = np.random.rand(100, 100).astype(np.float32)
write_raster_band(path, new_data, band_index=1)

Contributing

Contributions are welcome! Please see CONTRIBUTING.md for guidelines.

License

This project is licensed under the MIT License - see the LICENSE file for details.