The babelizer is a utility for wrapping a library that exposes a Basic Model Interface (BMI) so that it can be imported as a Python package.
Supported languages include:
- C
- C++
- Fortran
- Python
The babelizer is an element of the CSDMS Workbench, an integrated system of software tools, technologies, and standards for building and coupling models. The Workbench provides two Python frameworks for model coupling, pymt and landlab. The babelizer was written to bring models written in other languages into these frameworks. However, as long as your model satisfies the requirements below, you can use the babelizer to bring your model into Python without having to use any of the other tools in the Workbench.
To determine if the babelizer is right for you, first be aware of a few requirements.
- Your model must be written in C, C++, Fortran, or Python
- Your model must provide a shared library
- Your model must expose a Basic Model Interface through this library
The most difficult of the three requirements is the last--implementing a BMI. This involves adding a series of functions with prescribed names, arguments, and return values for querying and controlling your model. We have created several resources to help you understand the BMI and to guide you through the implementation process.
- The Basic Model Interface documentation provides an overview of the BMI as well as a detailed description of all of the BMI functions.
- The following provide a BMI specification for each of the supported languages:
- The following give examples of a BMI implementation for each of the supported languages:
There are lots of other good reasons to create a BMI for your model--not just so you can bring it into Python with the babelizer! Read all about them in the Basic Model Interface documentation.
The babelizer requires Python >=3.10.
Apart from Python, the babelizer has a number of other requirements, all of which can be obtained through either pip or conda, that will be automatically installed when you install the babelizer.
To see a full listing of the requirements, have a look at the project's requirements.txt file.
If you are a developer of the babelizer you will also want to install additional dependencies for running the babelizer's tests to make sure that things are working as they should. These dependencies are listed in requirements-testing.txt.
To install the babelizer, first create a new environment. Although this isn't strictly necessary, it isolates the installation to avoid conflicts with your base Python installation. This can be done with conda:
$ conda create -n babelizer python=3
$ conda activate babelizerThe babelizer and its dependencies are best installed with conda:
$ conda install babelizer -c conda-forgeAfter downloading the the babelizer source code, run the following from babelizer's top-level directory (the one that contains setup.py) to install babelizer into the current environment:
$ pip install -e .or using conda:
$ conda install --file=requirements.txt -c conda-forgeThe babelizer requires a single toml-formatted input file that describes the library to wrap. This file is typically named babel.toml. An example of a blank babel.toml file:
[library]
[library."<name>"]
language = "c"
library = ""
header = ""
entry_point = ""
[build]
undef_macros = []
define_macros = []
libraries = []
library_dirs = []
include_dirs = []
extra_compile_args = []
[package]
name = ""
requirements = []
[info]
github_username = "pymt-lab"
package_author = "csdms"
package_author_email = "[email protected]"
package_license = "MIT"
summary = ""
[ci]
python_version = ["3.9"]
os = ["linux", "mac", "windows"]You can generate babel.toml files using the babelize generate command. For example, the above babel.toml was generated with:
$ babelize generate > babel.tomlThe library section specifies information about the library being babelized.
The name of the babelized class. This will be a Python class, so it should follow Python naming conventions such as camel-case typing.
The programming language of the library (possible values are "c", "c++", "fortran", and "python").
[library]
language = "c"The name of the BMI library to wrap. This is the text passed to the linker through the -l option; for example, use "foo" for a library libfoo.a.
The name of the header file (.h, .hxx) declaring the BMI class. This option is only needed when wrapping C and C++ libraries.
The name of the BMI entry point into the library. For object-oriented languages, this is typically the name of a class that implements the BMI. For procedural languages, this is typically a function.
An example of a C++ library (bmi_child), exposing a class BmiChild (which implements a BMI) might look like the following:
[library]
[library.Child]
language = "c++"
library = "bmi_child"
header = "bmi_child.hxx"
entry_point = "BmiChild"whereas a C library (bmi_cem), exposing a function register_bmi_cem (which implements a BMI) might look like:
[library]
[library.Cem]
language = "c"
library = "bmi_cem"
header = "bmi_cem.h"
entry_point = "register_bmi_cem"In the build section the user can specify flags to pass to the compiler when building the extension.
Name and extra requirements needed to build the babelized library.
Name to use for the wrapped package. This is used when creating the new package <package_name>. For example, the following will create a new package, pymt_foo.
[package]
name = "pymt_foo"List of packages required by the library being wrapped. For example, the following indicates that the packages foo and bar are dependencies for the package.
[package]
requirements = [ "foo", "bar",]Descriptive information about the package.
The GitHub username or organization where this package will be hosted. This is used in generating links to the CI, docs, etc.
Author of the wrapped package. Note that this is not the author of the library being wrapped, just the code generated by the babelizer.
Contact email to use for the wrapped package.
Specify the Open Source license for the wrapped package. Note that this is not the license for the library being wrapped, just for the code generated by the babelizer.
A short description of the wrapped library.
Information about how to set up continuous integration.
[ci]
python_version = ["3.7", "3.8", "3.9"]
os = ["linux", "mac", "windows"]A list of Python versions to build and test the generated project with.
A list of operating systems to build the generate project on. Supported values are linux, mac, and windows.
Below is an example of a babel.toml file that describes a shared library, written in C. In this example, the library, bmi_hydrotrend, exposes the function register_bmi_hydrotrend that implements a BMI for a component called hydrotrend.
[library]
[library.Hydrotrend]
language = "c"
library = "bmi_hydrotrend"
header = "bmi_hydrotrend.h"
entry_point = "register_bmi_hydrotrend"
[build]
undef_macros = []
define_macros = []
libraries = []
library_dirs = []
include_dirs = []
extra_compile_args = []
[package]
name = "pymt_hydrotrend"
requirements = ["hydrotrend"]
[info]
github_username = "pymt-lab"
package_author = "csdms"
package_author_email = "[email protected]"
package_license = "MIT"
summary = "PyMT plugin for hydrotrend"
[ci]
python_version = ["3.7", "3.8", "3.9"]
os = ["linux", "mac", "windows"]You can use the babelize sample-config command to generate
a sample babel.toml file to get you started. For example,
the above babel.toml can be generated with the following,
babelize sample-configGenerate Python bindings for a library that implements a BMI, sending output to the current directory
babelize init babel.tomlUpdate an existing repository
babelize updateFor a complete example of using the babelizer to wrap a C library exposing a BMI, see the User Guide of the documentation.