Xarray Assets Extension class

[jsignell]

Related Issue(s):

• Convenience methods for converting STAC objects / linked assets to data containers #846

Description:

EDIT: I removed the open method, so this is just a regular extension class now.

This PR implements an extension class for xarray-assets extension with a special open method for extended pystac.Assets as described in #846 (comment)

import pystac
from pystac.extensions.xarray_assets import XarrayAssetsExtension

daymet_collection = pystac.Collection.from_file(
    "https://planetarycomputer.microsoft.com/api/stac/v1/collections/daymet-daily-hi"
)
daymet_asset = daymet_collection.assets["zarr-abfs"]
xr_ext = XarrayAssetsExtension.ext(daymet_asset, add_if_missing=True)
xr_ext.open()

PR Checklist:

• pre-commit hooks pass locally
• Tests pass (run scripts/test)
• Documentation has been updated to reflect changes, if applicable
• This PR maintains or improves overall codebase code coverage.
• Changes are added to the CHANGELOG. See the docs for information about adding to the changelog.

[jsignell]

Not super ergonomic with the current extension access system, but I'm optimistic that the extension API can be revamped a la #1051 without a substantial rewrite of the classes.

[codecov]

Codecov Report

Patch coverage: 94.66% and project coverage change: +0.02 🎉

Comparison is base (dd999cd) 91.96% compared to head (6bf65aa) 91.99%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1161      +/-   ##
==========================================
+ Coverage   91.96%   91.99%   +0.02%     
==========================================
  Files          50       51       +1     
  Lines        6720     6795      +75     
  Branches      994     1000       +6     
==========================================
+ Hits         6180     6251      +71     
- Misses        365      368       +3     
- Partials      175      176       +1     

Impacted Files	Coverage Δ
pystac/extensions/xarray_assets.py	94.59% <94.59%> (ø)
pystac/__init__.py	94.64% <100.00%> (+0.09%)	⬆️

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[gadomski]

Couple high level questions:

• Did you consider adding xpystac as a optional dependency (ala orjson)?
• How do you see this model working for "containers" that aren't tied to a specific extension? E.g. @TomAugspurger's proposed to_geopandas for an ItemCollection?
• I think this deserves more than just API docs, e.g. updating how-to-read-data-from-stac.ipynb to use the new extension interface?

[jsignell]

* Did you consider adding **xpystac** as a optional dependency (ala **orjson**)?

I guess that's a good idea in case we need to pin it later on?

* How do you see this model working for "containers" that aren't tied to a specific extension? E.g. @TomAugspurger's proposed `to_geopandas` for an `ItemCollection`?

Because ItemCollections can't have extensions or because there would be no need for a geopandas extension? I was kind of thinking the table extension could be the geopandas extension. I guess alternatively there could be a generic reader extension but that might be too far of a stretch from what extensions are meant to be.

* I think this deserves more than just API docs, e.g. updating how-to-read-data-from-stac.ipynb to use the new extension interface?

Yeah definitely. I wasn't sure if people would like this idea, so I didn't bother with that yet.

[gadomski]

I guess that's a good idea in case we need to pin it later on?

Yeah, and it makes it clearer that there's pystac functionality that uses xpystac, instead of just tests.

ItemCollections can't have extensions

Yup, this.

I was kind of thinking the table extension could be the geopandas extension.

But theoretically you could want to make a dataframe out of something that doesn't have the table extension in its stac_extensions list, correct?

Yeah definitely. I wasn't sure if people would like this idea, so I didn't bother with that yet.

👍🏼

[jsignell]

But theoretically you could want to make a dataframe out of something that doesn't have the table extension in its stac_extensions list, correct?

Yeah but couldn't you just add the table extension in that case?

[gadomski]

Yeah but couldn't you just add the table extension in that case?

If you don't need the table extension to put the STAC stuff into a data frame, IMO you shouldn't have to add the table extension. Otherwise, I could see a scenario where someone loads some STAC, adds the table extension to do some data frame work, then has to strip the table extension back out again before saving the STAC somewhere else -- feels hackey.