WNTR requires 64-bit Python (tested on versions 3.7, 3.8, 3.9, and 3.10) along with several Python package dependencies. See Requirements and Optional dependencies for more information. WNTR can be installed as a Python package as briefly described below. Detailed instructions are included in the following section.
Users can install the latest release of WNTR from PyPI or Anaconda using one of the following commands in a terminal, command line, or PowerShell prompt.
Detailed installation instructions are included below.
Step 1: Setup the Python environment¶
Python can be installed on Windows, Linux, and Mac OS X operating systems. WNTR requires 64-bit Python (tested on versions 3.7, 3.8, 3.9, and 3.10) along with several Python package dependencies. Python distributions, such as Anaconda, are recommended to manage the Python environment. Anaconda can be downloaded from https://www.anaconda.com/products/individual. General information on Python can be found at https://www.python.org/.
It is recommended to install Anaconda for a single user by selecting the ‘Just Me’ option during installation. If a user-writeable location is selected for installation (e.g., C:\Users\username\Anaconda3), then the ‘Just Me’ option does not require administrator privileges.
It is also recommended to add Anaconda to the PATH environment variable. This will facilitate access to Python from a command prompt without having to include the full path name. This can be done by either 1) selecting the ‘Add Anaconda to my PATH environment variable’ option during installation or 2) manually adding C:\Users\username\Anaconda3 to the environmental variables. Note that the first option is not recommended by Anaconda because it elevates the priority of Anaconda software over previously installed software. While the second option allows the user to define priority, this requires administrator privileges. If Anaconda is not added to the PATH environment variable, Python can be run by using the full path name (e.g., C:\Users\username\Anaconda3\python).
Anaconda includes the Python packages needed for WNTR, including NumPy, SciPy, NetworkX, pandas, and Matplotlib. For more information on Python package dependencies, see Requirements. If the Python installation does not include these dependencies, the user will need to install them. This is most commonly done using pip or conda.
Anaconda also comes with Spyder, an IDE, that includes enhanced editing and debugging features along with a graphical user interface. Debugging options are available from the toolbar. Code documentation is displayed in the object inspection window. Pop-up information on class structure and functions are displayed in the editor and console windows.
To open a Python console, open a command prompt (cmd.exe on Windows, terminal window on Linux and Mac OS X) and run ‘python’, as shown in Figure 3, or open a Python console using an IDE, like Spyder, as shown in Figure 4.
Step 2: Install WNTR¶
The installation process differs for users and developers. Installation instructions for both types are described below.
For users: Users can install WNTR using PyPI, Anaconda, or by downloading a zip file and building the source code.
If WNTR is installed using PyPI or Anaconda (Options 1 or 2 below), the examples folder will not be downloaded. The examples can be downloaded by going to https://github.com/USEPA/WNTR, select the “Clone or download” button and then select “Download ZIP.” Uncompress the zip file using standard software tools (e.g., unzip, WinZip) and store them in a folder.
Option 1: Users can install WNTR from PyPI using pip, which is a command line software tool used to install and manage Python packages. It can be downloaded from https://pypi.python.org/pypi/pip.
To install WNTR using pip, open a command line or PowerShell prompt and run:pip install wntr
This will install the latest release of WNTR from https://pypi.python.org/pypi/wntr.
Option 2: Users can install WNTR from Anaconda using conda, which is a command line software tool used to install and manage Python packages. It can be downloaded from https://www.anaconda.com/products/individual.
To install WNTR using conda, open a command line or PowerShell prompt and run:conda install -c conda-forge wntr
This will install the latest release of WNTR from https://anaconda.org/conda-forge/wntr.
Option 3: Users can download a zip file that includes source files and the examples folder from the US EPA GitHub organization.
To download the main branch, go to https://github.com/USEPA/WNTR, select the “Clone or download” button and then select “Download ZIP.” This downloads a zip file called WNTR-main.zip.
To download a specific release, go to https://github.com/USEPA/WNTR/releases and select a zip file.
Uncompress the zip file using standard software tools (e.g., unzip, WinZip) and store them in a folder. WNTR can then be installed by running a Python script, called setup.py, that is included in the source files. To build WNTR from the source files, open a command line or PowerShell prompt from within the folder that contains the files and run:python setup.py install
For developers: Developers can install and build WNTR from source files using git, which is a command line software tool for version control and software development. It can be downloaded from http://git-scm.com.
To build WNTR from source files using git, open a command line or PowerShell prompt and run:git clone https://github.com/USEPA/WNTR cd WNTR python setup.py develop
Step 3: Test installation¶
To test that WNTR is installed, open a Python console and run:import wntr
If WNTR is installed properly, Python proceeds to the next line. No other output is printed to the screen.
If WNTR is not installed properly, the user will see the following ImportError:ImportError: No module named wntr
See Getting started for a simple example that can also be used to test installation. A full set of software tests can also be run locally to ensure proper installation, see Software tests for more details.
Requirements for WNTR include 64-bit Python (tested on versions 3.7, 3.8, 3.9, and 3.10) along with several Python packages. Users should have experience using Python (https://www.python.org/), including the installation of additional Python packages. The following Python packages are required:
These packages are included in the Anaconda Python distribution.
The following Python packages are optional:
All of these packages except geopandas are included in the Anaconda Python distribution.
Proper installation of geopandas requires installing several geopandas dependencies, including fiona, pyproj, and shapely. See https://geopandas.org/en/stable/getting_started/install.html for more information. On Linux and Mac OS X, installing geopandas through the conda-forge channel will install the dependencies. On Windows, the dependencies must be installed manually, see https://geoffboeing.com/2014/09/using-geopandas-windows/ for more information.
Developers can clone and setup the main branch of WNTR from the GitHub
repository using the following commands in a terminal, command line, or PowerShell prompt.
WNTR includes C++ code that can be built into shared object files (e.g., pyd for Windows) using the optional
--build command line argument.
This requires that the developer has a C++ compiler on their path:
git clone https://github.com/USEPA/WNTR cd WNTR python setup.py develop --build
If the developer does NOT have a C++ compiler, or would rather use prebuilt wheels, the shared object files can be downloaded from WNTR GitHub Actions using the following steps:
Clone and setup the main branch of WNTR from the GitHub repository as shown above, but omit the
--buildcommand line argument
Select the latest GitHub Actions build_tests that uses the main branch from https://github.com/USEPA/WNTR/actions/workflows/build_tests.yml
Scroll down to “Artifacts”
Download the wheel that matches the desired operating system and Python version (for example, wntr_3.9_windows-latest.whl)
Unzip the wheel and locate the following files (which are named according to the operating system and Python version)
Copy these files into the matching directory in the cloned version of WNTR
Note that users installing WNTR through PyPI or conda do not need to compile code.
More information for developers can be found in the Software quality assurance section.