Listing available collections from an API with rustac-py

[scottyhq]

For APIs with a lot of STAC collections, you might not know your desired collection ID in advance. pystac_client has a get_collections() function to facilitate this kind of workflow.

I'm increasingly using rustac to search, but find myself using pystac_client only to quickly discover collection IDs and metadata. I'm wondering if such a function could be added to the API here? For example a replacement for this bit of code to iterate over pages of collections:

catalog = pystac_client.Client.open("https://cmr.earthdata.nasa.gov/stac/ASF")
collections = list(catalog.get_collections())
print(f"Found {len(collections)} collections")
# Found 153 collections
df_collections = pd.DataFrame([x.to_dict() for x in collections])
df_collections.head()

[hrodmn]

I think having a /collections client in rustac makes sense but it begs the question: are rustac and pystac-client going to continue having overlapping features or should pystac-client migrate to use rustac for crawling catalogs?

[gadomski]

It's a good question. pystac-client has a lot of heuristics and other "smarts" that we (as a community) discard at our peril. But making a user decide which library to use for which type of search sucks.

IMO the best user experience is a pystac v2 that includes

• An API client (pystac-client)
• async (via rustac)
• stac-geoparquet (via rustac)

That way, folks could just install pystac and get everything. rustac and pystac-client could be optional dependencies so folks could still get the vanilla pystac experience?