More argopy/python export functions at: https://dataselection.euro-argo.eu

[gmaze]

The new https://dataselection.euro-argo.eu is awesome and powered by a fast and scalable server and db.
We should be able to use it with argopy, as we started to explore in #47

But the "export" functionality are still clunky, since it relies on a batch job launched in the Ifremer HPC Datarmor. Users receive data, many minutes after export, once the job has passed in the Datarmor queue.

I'd like to make the following 2 propositions, please vote or promote if you think this could be useful !

[gmaze]

💡 New export button in main page

What if users could simply get a small code snippet in the export popup window showing how to download their current data selection with argopy ?
This would let users decide where and how to get the data from.

[tcarval]

I like very much this idea.
We'll have to check that Argopy is able to take into account the rich filters of the Argo data selection.
The subsetting batch that generates NetCDF or CSV receives a list of profiles and generate data files from this list.
Argopy may have to be enhanced for this type of data selection.

We may also decide to limit to simple filters (spatial, temporal, parameters)

[gmaze]

Thanks @tcarval for looking at this
About data selection available in argopy:

• space/time, including max pressure (although the polygon is still ongoing work at New method to fetch data along a given trajectory #169 )
• Position/Data QC and Network: this is possible with the data fetchers options mode and ds selecting the standard/expert user modes and the dataset to use (called network here)
• parameters & deployment year: nope, we fetch data from what's available. But it would be manageable to add the code snippet to run after data fetching to enforce these criteria

[gmaze]

I guess the most problematic scenario is when data selection comes from the "everywhere" search box.
In this case, we could imagine to create a new data fetcher in argopy that would work in 2steps:

1. Get the list of profiles ID corresponding to the search criteria (using https://dataselection.euro-argo.eu/api/full-search-response)
2. Load profiles

With such a fetcher, it would be trivial to "translate" the user data selection into a code snippet.
It could look like this:

from argopy import DataFetcher
argo = DataFetcher(src='ea').selection(search="coatanoan", where="everywhere")
ds = argo.load().data

But I'm worry that such a fetcher would have poor performances

[gmaze]

There is also the possibility to simply add the new export button only for data selection that argopy can handle !

[cgourcuf]

This seems a very good idea from my side too. However, I would recommend to fix pending issues (number of cycles/profiles, etc.) as priorities (see https://gitlab.ifremer.fr/ea-data-selection/ea-data-selection/-/issues)

[gmaze]

💡 New export button in a profile visualisation page

When user click on a profile from the data selection map, they can open a "Show profile data" link that points to the profile visualisation/dashboard page.
On this page we could also add a new export button:

See the new upper right export button proposition.
Clicking this button will “copy” a code snippet to load this profile in python.

Once clicked, a popup window could be opened to confirm that a code snippet has been “copied” and is ready to be pasted in user code to load this profile