use PotenitialAction to describe arrow dataset access

[valentinedwv]

Stac catalog has a database access element on the page. We need to find a way to replicate/communicate the same information.

Also, Collections may have assets, if so, then we render as a dataset with a potential action.
Datasets can have s3 links that are really better described with code, add a potenital action

So, for the s3 link, we might think about using potential actions rather than data distributions

Not sure if this is a createAction, or an consumeAction... and not really sure if that matters. we pick one.

[ylyangtw]

https://docs.mlcommons.org/croissant/docs/croissant-spec.html

[ylyangtw]

close as #12 is the same issue

[valentinedwv]

Partially, #12 makes it available.

Seems some parts of the croissant spec are missing, and @fils do you think we should follow it?

And noticing the first one also has a code snippet. Does the facet search also make that snippet accessible?

snippet from jsonld:

"distribution": 
[
{
"@type": 
"DataDownload",
"contentUrl": 
"https://sdsc.osn.xsede.org/bio230014-bucket01/challenges/metadata/model_id/cb_prophet.json",
"description": 
"Use `jsonlite::fromJSON()` to download the model metadata JSON file. This R code will return metadata provided during the model registration. ### R ```{r} # Use code below model_metadata <- jsonlite::fromJSON("https://sdsc.osn.xsede.org/bio230014-bucket01/challenges/metadata/model_id/cb_prophet.json") ",
"encodingFormat": 
"application/json",
"name": 
"Model Metadata"
},
{
"@type": 
"DataDownload",
"contentUrl": 
"https://github.com/cboettig/forecasts-darts-framework",
"description": 
"The link to the model code provided by the model submission team",
"encodingFormat": 
"text/html",
"name": 
"Link for Model Code"
},
{
"@type": 
"DataDownload",
"contentUrl": 
"s3://anonymous@bio230014-bucket01/challenges/forecasts/bundled-parquet//project_id=neon4cast/duration=P1D/variable=temperature/model_id=cb_prophet?endpoint_override=sdsc.osn.xsede.org",
"description": 
"Use `R` or `Python` code for remote access to the database. This code will return results for this variable and model combination. ### R ```{r} # Use code below all_results <- arrow::open_dataset("s3://anonymous@bio230014-bucket01/challenges/forecasts/bundled-parquet//project_id=neon4cast/duration=P1D/variable=temperature/model_id=cb_prophet?endpoint_override=sdsc.osn.xsede.org") df <- all_results |> dplyr::collect() ``` You can use dplyr operations before calling `dplyr::collect()` to `summarise`, `select` columns, and/or `filter` rows prior to pulling the data into a local `data.frame`. Reducing the data that is pulled locally will speed up the data download speed and reduce your memory usage. ### Python ```# Use code below import ibis con = ibis.duckdbf.connect() con.raw_sql(f''' CREATE OR REPLACE SECRET secret ( TYPE S3, ENDPOINT 'sdsc.osn.xsede.org', URL_STYLE 'path' );/n ''' path = "s3://bio230014-bucket01/challenges/forecasts/bundled-parquet//project_id=neon4cast/duration=P1D/variable=temperature/model_id=cb_prophet" con.read_parquet(path + "/**")",
"encodingFormat": 
"application/x-parquet",
"name": 
"Database Access for Daily Water_temperature"
}
],