Installation

PC2Products runs on BERA Tools, with R and rpy2 added. Both environments below contain only R and rpy2 on the R side. Every R package comes from CRAN, installed by the tools into the environment's R the first time they need it (see How all the tools behave).

	`environment.yml` (conda)	`pixi.toml` (pixi)
Python	3.12	3.14
R	`r-base` 4.5.x	`r-base` 4.5.x
rpy2	From PyPI	From conda-forge on Linux, from PyPI on Windows
R packages	None	None
Platforms	Any	`win-64`, `linux-64`

R is held at 4.5.x because conda-forge builds rpy2 for R 4.5 on Linux only, and for nothing newer than R 4.4 on Windows, which is why rpy2 comes from PyPI on Windows. On Windows, CRAN provides prebuilt packages for R 4.5. On Linux, CRAN packages such as sf and terra are compiled from source, which needs the GDAL, GEOS, PROJ and udunits development libraries (for example libgdal-dev libgeos-dev libproj-dev libudunits2-dev on Ubuntu) and takes a while on the first run.

Using conda

git clone https://github.com/BERATool/pc2products.git
cd pc2products
conda env create -n PC2Products -f environment.yml
conda activate PC2Products
pip install --no-deps -e .

Using pixi

pixi install    # also installs beratools in editable mode, plus pytest and the docs tools
pixi run main   # launch the BERA Tools GUI

Verify the installation

To install every R package up front and confirm that they load, run the R environment test:

python -m pytest tests/test_r_environment.py -v

Then list the tools:

bt list-tools

Windows

Run the tools from an activated environment. If you call the environment's python.exe directly without activating it, GDAL can fail to load its DLLs.

Windows: rpy2 with a standard R install

If you use R from C:\Program Files\R with rpy2 installed by pip, run the tools and the R tests from PowerShell or cmd, not Git Bash. Under Git Bash, rpy2 3.6 crashes while locating R (IndexError in rpy2/situation), because Git Bash's sh makes R CMD config return empty output instead of failing.

Requirements

Python: rpy2 and psutil, in addition to the core BERA Tools dependencies. The example workflow also needs hydra-core. All three are in environment.yml and pixi.toml.

R: r-base 4.5.x. The tools install these packages into it the first time they need them:

R package	Purpose
`lidR`	LAScatalog processing, `rasterize_terrain`, `rasterize_canopy`
`rlas`	Reading LAS/LAZ files
`future`	Parallel processing of catalogs
`terra`	Raster output and mosaicking
`sf`, `units`	CRS, area and map-unit reporting
`RCSF`	Cloth Simulation Filter for ground classification
`gstat`	Kriging interpolation (DTM and water table)
`sp`, `raster`, `stars`	Spatial and raster support
`ggplot2`	Plots from Seismic Line Disturbance Analysis
`rpart`	Decision tree in Surface Water Mapping

Remove the environment

conda activate  # go to the base environment
conda env remove -n PC2Products