Multi-Partition parquet:kv tables do not display multiple partition drop-downs

[robbcamera]

Description

Multi-Partition parquet:kv tables do not display multiple partition drop-downs

Steps to reproduce

1. place a .parquet file in a multi-partition kv directory. for example, "/path/to/somewhere/date_partition=2023-08-01/id_partition=id-001/my.parquet"
2. emulate multiple partitions via soft-link; from "/path/to/somewhere" directory, ln -s date_partition=2023-08-01/id_partition=id-001 date_partition=2023-08-01/id_partition=id-002 and ln -s date_partition=2023-08-01 date_partition=2023-08-02. this will make it seem that there are 2 "date" partitions, each of which has 2 "id" partiions
3. import the table(s) in a query; `ParquetTools.readTable("/path/to/somewhere")
4. the "most recent" date-partion (2023-08-02) will be selected from the drop-down of available "date" partitions. note that there no is drop-down for the "id" partition

Expected results

4. a "date" partition and an "id" partition are visible as drop-downs. i guess this gets nasty if i want to select the "date=2023-08-01" and "ALL" of the "id" partitions ...

Actual results

3. there is no way to select a specific "id"

Additional details and attachments

Versions

• Engine Version: ...
• Web UI Version: ...
• Java Version: ...
• Barrage Version: ...
• OS: ...
• Browser: ...
• Docker: ...

[mofojed]

Web UI only reads one partition column:

web-client-ui/packages/iris-grid/src/IrisGrid.tsx

Line 841 in e3db6bf

const column =

I think it's all wired up through the JS API supporting multiple partition columns, so it's just the UI that needs to update. Parquet is the easiest way to reproduce.

[mofojed]

You can make a table with multiple partitions fairly easy in deephaven-core:

import deephaven.plot.express as dx
stocks = dx.data.stocks()
stocks_partitioned = stocks.partition_by(['sym', 'exchange', 'side'])

[mofojed]

Will need some guidance from @dsmmcken on layout for multiple partition selectors.

[mofojed]

An example of a multiple partition table that doesn't require #1143 :

from deephaven import empty_table

part = empty_table(4).update("II=ii")

from deephaven.parquet import write, read

write(part, "/tmp/pt-1438/x=0/y=0/part.parquet")
write(part, "/tmp/pt-1438/x=0/y=1/part.parquet")
write(part, "/tmp/pt-1438/x=1/y=0/part.parquet")
write(part, "/tmp/pt-1438/x=1/y=1/part.parquet")

partition_table = read("/tmp/pt-1438")