native-fisher-py

Why native-fisher-py?

native-fisher-py is a self-contained alternative to the fisher-py reader. While fisher-py requires a local .NET runtime and pythonnet, this package bundles the necessary components using .NET NativeAOT and Rust to provide a consistent binary bridge.

Features

• Drop-in Compatible: Designed to match the fisher_py.RawFile API for simplified migration.
• Bundled .NET Components: No separate .NET runtime installation is required on the host system.
• Cross-Platform: Pre-built binaries for macOS (ARM64/x64), Linux (x64), and Windows (x64).
• Reliable Deployment: Easier integration into CI/CD pipelines and specialized Linux environments.

How it works

This project provides a native bridge to the official Thermo Fisher libraries using a three-layer approach:

1. Official DLLs: Uses the original .dll assemblies provided by Thermo Fisher Scientific.
2. C# NativeAOT Wrapper: A compiled transition layer (ThermoNativeReader) that interfaces with those DLLs.
3. Rust PyO3 Layer: A Rust bridge (native-fisher-py) that provides the final Python bindings.

This approach ensures stability and parity with the official reader while providing a dependency-free experience for Python users.

Quick Start

# Just change the import, the rest of your code stays the same!
from native_fisher_py import RawFile

with RawFile("data.raw") as raw:
    print(f"Number of scans: {raw.number_of_scans}")
    
    # Get spectral data as high-speed NumPy arrays
    m, i, c, meta = raw.get_scan_from_scan_number(1)
    print(f"First peak at {m[0]} m/z with intensity {i[0]}")

Migrating from fisher-py

If you are currently using fisher-py, migration is as simple as:

1. pip install native-fisher-py
2. Update your imports:

- from fisher_py import RawFile
+ from native_fisher_py import RawFile

3. (Optional) Uninstall fisher-py: pip uninstall fisher-py

All core methods (get_scan_from_scan_number, get_spectrum, get_chromatogram, get_ms2_scan_number_from_retention_time, etc.) are implemented with identical signatures and return types.

Quick Local Build

For convenience, you can run the included build.sh script to build both parts of the project:

./build.sh

Step-by-Step Manual Build

To build the project from source, you need .NET 8 SDK, Rust (cargo/maturin), and clang.

1. Build the C# NativeAOT Core

Navigate to the C# project and publish the NativeAOT shared library for your platform:

cd native/ThermoNativeReader

# Example for Apple Silicon (macOS arm64)
dotnet publish -r osx-arm64 -c Release -p:PublishAot=true

# Example for Linux (x64)
# dotnet publish -r linux-x64 -c Release -p:PublishAot=true

The output will be in publish/ThermoNativeReader.dylib (or .so / .dll).

2. Build the Rust Bridge

Navigate to the native_fisher_py folder and use maturin to build and install the Python package. You must point to the location of the C# library.

cd native_fisher_py

# Point to your build from Step 1
export THERMO_NATIVE_LIB=$(pwd)/../native/ThermoNativeReader/bin/Release/net8.0/osx-arm64/publish/ThermoNativeReader.dylib

maturin develop

Credits & Legal Notice

This project is powered by the Thermo Fisher Scientific RawFileReader (copyright © 2016-2026 Thermo Fisher Scientific, Inc.). All rights reserved.

The native-fisher-py package includes the official RawFileReader libraries, which remain the property of Thermo Fisher Scientific. By using this software, you agree to the terms specified in their license.

Releasing a New Version

Releases are fully automated via GitHub Actions (release.yml), but the workflow requires that the git tag strictly matches the version in native_fisher_py/Cargo.toml.

To trigger a new release and publish to PyPI:

1. Update the version in native_fisher_py/Cargo.toml.
2. Commit the version bump.
3. Create and push a matching git tag:

git tag vX.Y.Z
git push origin vX.Y.Z

The GitHub Action will automatically:

1. Verify that the git tag version matches the Cargo.toml version (failing otherwise).
2. Build the native wheels across all platforms (macOS, Linux, Windows).
3. Publish the final artifacts to PyPI.