Adding a post_table

[juliomateoslangerak]

I have some code I'd like to share for posting a table into OMERO if you would like to. I would have to cleanup a bit and consider some posibilities before.
As for now, the signature is something like this:

def post_table(conn, target_id: int, name: str, column_names: list[str], column_descriptions: list[str], values: list[list], namespace=None, table_description=None): -> table_id

The values argument is a list of lists of types that should within logic map to the types supported by the OMERO tables. That's kind of fine but for the image_id, roi, etc references that are special in omero. How would you go about this?
One option would be to assume that if a column name is image_id, it is considered as an images column, etc

Also, ideally you would like to pass a pandas dataframe as an OMERO table. I could try to implement that.

[mellertd]

I like this, and I haven't looked into how the OMERO API deals with posting tables at all, so even better!

Regarding input, I do think it would be preferred to take a pandas dataframe, as we are aiming for taking and returning objects that are either core Python or part of the SciPy ecosystem. PyTables support might also make sense but possibly harder to implement?

[mellertd]

One other thought...I think tables are similar to other annotations in that you can attach them to several different omero model object types (images, datasets, etc). So that should be supported (similar to what we have done with post_map_annotation)

[juliomateoslangerak]

Indeed, a table is a 'special' file annotation that you may link to more than object.
You may construct the table through the omero's API. That is a bit of a pain but I already coded that. The problem may be that it is not very efficient of convenient if your data is on a proper pandas table.
however, as OMERO.tables is build on pyTables probably is easier to go that path

[erickmartins]

Closed with #58.