This repo is for CUDA string processing code for data science applications. The modules here start with Pandas strings parity but also can include more advanced features.

This is a standalone library with no other dependencies but is expected to be used in RAPIDS applications. And some RAPIDS modules may depend on cuStrings and its bindings (nvStrings).

C/C++ source code including python bindings. See /cpp/README.md

Python modules and wrappers including test scripts. See /python/README.md

These are sample data files for testing the library features. There is no source code here but this is considered an active directory since many of the test cases rely on them.

Documentation of python interfaces generated from the python source code. This also includes some general documentation on capabilities and limitations.

Support files for deploying to conda environments.

See the C++ readme for instructions building and installing.

The Dockerfile can be used to setup a build and test environment. It is based on the cuDF docker image mainly to keep a consistent conda enviornment. The NVStrings library has no dependencies on cuDF.