This project includes Python utilities related to the wasi-data specification.
WASI is a modular system interface for WebAssembly. Checkout this blog post for an excellent overview and introduction. This system interface securely and portability provides an interface to run WebAssembly modules outside of the Web.
A fantastic overview of the interface types proposal is this blog post.
Currently the primary use of the package is wrapping WASM functions with
callable objects compatible with pandas and Dask DataFrame apply methods.
This allows you to write functions in Rust, C, Go, or any other language
that compiles to WASM, and use those functions in your DataFrame workflows.
import witxcraft as wc
# Wrap all functions in a WASM module in Python wrappers.
# A WITX file in required to describe the function signatures.
# If the WITX file has the same basename as the WASM file,
# the witx= parameter does not need to be specified.
wasm_funcs = wc.fromwasmmod('mylib.wasm', witx='mylib.witx')
# Each function in the WASM file is now available as a
# callable object on the returned object. The function can be
# called standalone.
wasm_funcs.mult(10, 5)
# Or they can be used in DataFrame methods that apply functions.
df = pd.DataFrame([...], columns=['a', 'b'])
df['a'].apply(wasm_funcs.mult, args=[5])
df['a'].apply(wasm_funcs.mult, df['b'])
wasm_funcs.mult(df['a'], df['b'])The examples folder also contains examples in Jupyter notebook form. See the README.md file in that directory for information on how to run them.