chore: uv support

[jjfrench]

Description

• adds uv package management
• updates and fixes pre-commit hooks

90% of the PR is whitespace by ruff-pre-commit

[github-actions]

PR Preview Action v1.8.1
🚀 View preview at https://NASA-IMPACT.github.io/veda-docs/pr-preview/pr-282/
Built to branch gh-pages at 2026-08-19 16:57 UTC. Preview will be ready when the GitHub Pages deployment is complete.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[jjfrench]

I recommend the Hide Whitespace setting on "Files changed" to view package management changes and ReviewNB for the notebooks.

[wildintellect]

• The switch to UV, pyproject.toml seems fine.
• The whitespace cleanup seems fine.
• In the notebooks themselves, I've checked a few, and the changes are non-trivial. I see changes to imports and some changes to the code based on those, and I'm not sure we want to make those changes, or that they don't break the code execution (we don't have auto testing).

[jjfrench]

Most of the import changes should be glob and os -> pathlib. I can comment on the other ones

[jjfrench]

Line #3.            f"{RASTER_API_URL}/collections/{item['collection']}/items/{item['id']}/WebMercatorQuad/tilejson.json?&assets=cog_default&color_formula=gamma+r+1.05&colormap_name=viridis&rescale={rescale_values['min']},{rescale_values['max']}",

This requests.get was located inside a for item in items loop but didn't use item and therefore was printing the same URL x times.

Reply via ReviewNB

[jjfrench]

Line #7.    cluster = gateway.connect(clusters[0].name) if clusters else GatewayCluster(shutdown_on_close=True)

functionally the same as

if clusters:
    cluster = gateway.connect(clusters[0].name)
else:
    cluster = GatewayCluster(shutdown_on_close=True)

Reply via ReviewNB

[jjfrench]

Several notebooks remove the pandas import as pd is not used when import pandas as pd

Reply via ReviewNB

[jjfrench]

Here's an example of some of the glob/os changes made.

Reply via ReviewNB

[jjfrench]

Line #3.    import contextlib

Main import change here. Uses contextlib in the case where there is a try / except -> pass

try:
    plt.colormaps.register(cmap_tornado)
except ValueError:
    pass 

with contextlib.suppress(ValueError):
    plt.colormaps.register(cmap_tornado)

Reply via ReviewNB

[jjfrench]

This is the only notebook I tried fixing, to no avail (it was previously broken and remains broken). I looked at the VEDA STAC catalog, and this collection did not have COG_DEFAULT variables for rescaling. Here, we're grabbing the render parameters from the listed collection.

Variables are now corrected, but it now seems to run into issues due to image sizes

Reply via ReviewNB

[jjfrench]

Line #6.    for _page in range(10):

page was not used, so we preface the variable with an underscore to indicate that it doesn't matter

Reply via ReviewNB

[jjfrench]

with (
    fs.open(SOURCE_URI) as fileobj,
    xr.open_dataset(fileobj, engine="h5netcdf") as ds,
):

Only change here is combining the two context managers into one with - contents underneath got shifted left

Reply via ReviewNB

[jjfrench]

One of the few install cells that isn't commented out - I've done this on MAAP docs before, but % is preferred to ! because % will guarantee installation in the kernel your notebook is running.

Reply via ReviewNB

[jjfrench]

import UTC, functionally the same as the timezone import usage of timezone.utc

datetime.now(tz=timezone.utc) -> datetime.now(tz=UTC)

Reply via ReviewNB

[wildintellect]

Noting it's unclear if an empty newline is a qmd format requirement.

[jjfrench]

I haven't checked but I think that it applied it to every file. Does it affect qmd's?

[wildintellect]

I'm not sure, we probably have to try building with quarto to see if it complains.

[jjfrench]

I don't see anything glaring in the build at the top of this PR

[jjfrench]

Line #4.    clusters = gateway.list_clusters()

has to skip this cell but I've added back the import that was flagged as "unused". Looks like it was also missing rioxarray.

Reply via ReviewNB

[jjfrench]

Line #5.            "query": {"datetime": {"eq": "2021-01-01T00:00:00"}},

changing this to use the top-level datatime search since it no longer goes through query

Reply via ReviewNB

[jjfrench]

I had to go into EDL and accept the GESDISC DATA ARCHIVE EULA

Reply via ReviewNB

[jjfrench]

start_date = datetime.datetime(2022, 4, 1)
end_date = datetime.datetime(2022, 5, 12)

this is a ton of data - I shortened the date range when testing

Reply via ReviewNB

[jjfrench]

added rioxarray

Reply via ReviewNB

[jjfrench]

I completed another self-review where I ran all the notebooks on the veda hub. I've added back hvplot and rioxarray imports with a noqa tag. Fixed one STAC search cell and added a comment about EDL EULAs for another.

[wildintellect]

@jjfrench if you finished testing the notebooks then I'm ok with the PR, but I think we need a @NASA-IMPACT/veda-data-services review of the publication notebooks.