Files
uLib/docs/python/installation.md
2026-03-06 10:45:33 +00:00

147 lines
3.6 KiB
Markdown
Raw Permalink Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

# Python Installation
The `uLib` Python package exposes the Core and Math C++ libraries via [pybind11](https://pybind11.readthedocs.io) bindings. There are two ways to install it: as an **end user** (pre-built wheel / pip) or as a **developer** (editable build from source).
---
## Prerequisites
`uLib` depends on native C++ libraries that must be compiled. Ensure the following are available in your environment before installing:
| Dependency | Minimum version | Notes |
|---|---|---|
| Python | 3.9 | |
| CMake | 3.12 | |
| pybind11 | 2.6.0 | |
| Conan | 2.x | for Eigen3 / Boost |
| micromamba / conda | any | recommended provides ROOT, VTK |
### Creating the `mutom` Conda/Micromamba Environment
A ready-to-use environment definition is provided as `condaenv.yml` at the repository root.
=== "Micromamba"
```bash
micromamba env create -f condaenv.yml
micromamba activate mutom
```
=== "Conda"
```bash
conda env create -f condaenv.yml
conda activate mutom
```
The environment installs CMake, Conan, ROOT, VTK, and the compiler toolchain.
> **CUDA (optional)**
> If you want GPU-accelerated voxel filtering, you also need NVCC inside the environment:
> ```bash
> micromamba install cuda-nvcc -c conda-forge
> ```
---
## User Installation (wheel / pip)
Once the native dependencies are present in your environment, install the package with Poetry or pip:
```bash
# Activate your environment first
micromamba activate mutom
# Build and install (CUDA disabled by default)
poetry install
# Build and install with CUDA support
USE_CUDA=ON poetry install
```
After installation the module is importable from anywhere in the environment:
```python
import uLib
print(dir(uLib.Core))
print(dir(uLib.Math))
```
---
## Developer Installation (editable / in-tree build)
For development you typically want to skip the packaging layer and work directly against the CMake build tree.
### Step 1 Install Conan dependencies
```bash
conan profile detect # first time only
conan install . --output-folder=build --build=missing
```
### Step 2 Configure and build
```bash
# Standard release build
cmake --preset conan-release
# …or manually
cmake -B build \
-DCMAKE_TOOLCHAIN_FILE=build/conan_toolchain.cmake \
-DCMAKE_BUILD_TYPE=Release \
-DUSE_CUDA=OFF # set to ON when a GPU is available
cmake --build build --target uLib_python -j$(nproc)
```
The shared library (`uLib_python*.so`) is written to `build/src/Python/`.
### Step 3 Make the module importable
Point `PYTHONPATH` at the build output **and** the Python source directory (the latter carries the `uLib/__init__.py` that stitches sub-modules together):
```bash
export PYTHONPATH="$(pwd)/build/src/Python:$(pwd)/src/Python:$PYTHONPATH"
python -c "import uLib; print(uLib.__version__)"
```
Or, for a one-shot check:
```bash
PYTHONPATH="build/src/Python:src/Python" python src/Python/testing/pybind_test.py
```
### Step 4 Run the tests
CMake registers the Python tests alongside the C++ ones; use `ctest` from the build directory:
```bash
cd build
ctest --output-on-failure -R pybind
```
Individual test scripts can also be run directly once `PYTHONPATH` is set:
```bash
python src/Python/testing/core_pybind_test.py
python src/Python/testing/math_pybind_test.py
python src/Python/testing/math_filters_test.py
```
---
## Verifying the Installation
```python
import uLib
# Core module
obj = uLib.Core.Object()
timer = uLib.Core.Timer()
timer.Start()
elapsed = timer.StopWatch() # float, seconds
# Math module
v3 = uLib.Math.Vector3f([1.0, 0.0, 0.0])
print(v3[0]) # 1.0
```