emsarray.operations.cache

Operations for making cache keys based on dataset geometry.

Some operations such as triangulate_dataset() only depend on the dataset geometry and are expensive to compute. For applications that need to derive data from the dataset geometry it would be useful if the derived data could be reused between different runs of the same application or between multiple time slices of the same geometry distributed across multiple files. This module provides make_cache_key() to assist in this process by deriving a cache key from the important parts of a dataset geometry. Applications can use this cache key as part of a filename when save derived geometry data to disk or as a key to an in-memory cache of derived geometry.

The derived cache keys will be identical between different instances of an application, and between different files in multi-file datasets split over an unlimited dimension.

This module does not provide an actual cache implementation.

Functions

emsarray.operations.cache.make_cache_key(dataset, hash=None)

Derive a cache key from the geometry of a dataset.

Parameters:

• dataset (xarray.Dataset) – The dataset to generate a cache key from.

• hash (hashlib-compatible hash instance, optional) – An instance of a hashlib hash class. Defaults to hashlib.blake2b() with a digest size of 32, which is secure enough and fast enough for most purposes.

Returns:

cache_key (str) – A string suitable for use as a cache key. The string will be safe for use as part of a filename if data is to be cached to disk.

Examples

import emsarray
from emsarray.operations.cache import make_cache_key

# Make a cache key from the dataset
dataset = emsarray.tuorial.open_dataset("austen")
cache_key = make_cache_key(dataset)
>>> cache_key
'580853c44e732878937598e86d0b26cb81e18d986072c0790a122244e9d3f480'

Notes

The cache key will depend on the Convention class, the emsarray version, and a hash of the geometry of the dataset. The specific structure of the cache key may change between emsarray and python versions, and should not be relied upon.