xarray_extras: Advanced algorithms for xarray

This module offers several extensions to xarray, which could not be included into the main module because they fall into one or more of the following categories:

• They’re too experimental
• They’re too niche
• They introduce major new dependencies (e.g. numba or a C compiler)
• They would be better done by doing major rework on multiple packages, and then one would need to wait for said changes to reach a stable release of each package - in the right order.

The API of xarray-extras is unstable by definition, as features will be progressively migrated upwards towards xarray, dask, numpy, pandas, etc.

Features

csv

Multi-threaded CSV writer, much faster than pandas.DataFrame.to_csv(), with full support for dask and dask distributed.

cumulatives

Advanced cumulative sum/productory/mean functions

interpolate

dask-optimized n-dimensional spline interpolation

numba_extras

Additions to numba

recursive_diff

Recursively compare nested Python objects, with numpy/pandas/xarray support and tolerance for numerical comparisons

sort

Advanced sort/take functions

stack

Tools for stacking/unstacking dimensions

testing

Tools for unit tests

Command-line tools

bin/ncdiff.rst

Compare two NetCDF files or recursively find all NetCDF files within two paths and compare them

Index

• Installation
  • Required dependencies
  • Deployment
  • Testing
• What’s New
  • v0.3.0 (2018-12-13)
  • v0.2.2 (2018-07-24)
  • v0.2.1 (2018-07-22)
  • v0.2.0 (2018-07-15)
  • v0.1.0 (2018-05-19)
• csv
• cumulatives
• interpolate
• numba_extras
• recursive_diff
• sort
• stack
• testing
• ncdiff
• Usage
  • Chunking and RAM design
  • Further limitations

Credits

• recursive_diff, recursive_eq(), proper_unstack() and ncdiff were originally developed by Legal & General and released to the open source community in 2018.
• All boilerplate is from python_project_template, which in turn is from xarray.

License

xarray-extras is available under the open source Apache License.