fix(chart-data-api): support numeric temporal columns

[villebro]

SUMMARY

Some database connectors (at least the native Pinot connector) return a numeric value (either as second or millisecord epoch) instead of a regular timestamp. This was working correctly on viz.py, but support was lacking on /api/v1/chart/data. This centralizes the logic in a util, simplifies the code and adds tests that were previously missing.

TEST PLAN

CI + new tests

ADDITIONAL INFORMATION

• Has associated issue:
• Changes UI
• Requires DB Migration.
• Confirm DB Migration upgrade and downgrade tested.
• Introduces new feature or API
• Removes existing feature or API

[codecov-io]

Codecov Report

Merging #13138 (5ba5a73) into master (2ce7982) will increase coverage by 13.73%.
The diff coverage is 42.36%.

@@             Coverage Diff             @@
##           master   #13138       +/-   ##
===========================================
+ Coverage   53.06%   66.79%   +13.73%     
===========================================
  Files         489      492        +3     
  Lines       17314    29026    +11712     
  Branches     4482        0     -4482     
===========================================
+ Hits         9187    19389    +10202     
- Misses       8127     9637     +1510     

Flag	Coverage Δ
cypress	?
python	66.79% <42.36%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
superset/examples/birth_names.py	73.19% <ø> (ø)
...43f2fdb_add_granularity_to_charts_where_missing.py	0.00% <0.00%> (ø)
...s/260bf0649a77_migrate_x_dateunit_in_time_range.py	0.00% <0.00%> (ø)
...ons/versions/41ce8799acc3_rename_pie_label_type.py	0.00% <0.00%> (ø)
superset/connectors/sqla/views.py	62.43% <4.34%> (ø)
superset/views/datasource.py	88.70% <16.66%> (ø)
superset/viz.py	58.61% <27.27%> (ø)
superset/charts/commands/exceptions.py	92.85% <77.77%> (ø)
superset/utils/core.py	88.26% <87.50%> (ø)
superset/db_engine_specs/trino.py	90.47% <90.47%> (ø)
... and 936 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2dbe92b...5ba5a73. Read the comment docs.

[amitmiran137]

left a comment.
also not sure if related or not but maybe this should be applied when importing a dataset?
https://github.com/apache/superset/blob/master/superset/datasets/commands/importers/v1/utils.py#L139
or it is only related to when loading the data into a viz

[robdiciuccio]

Is the copy operation necessary here? This could be very expensive on memory for large result sets.

[villebro]

Good point @robdiciuccio . I'm a bit allergic to functions that mutate their inputs (even Pandas is moving away from this pattern), but this could in fact have a noticeable performance hit. I'm going to be touching this code in the coming days, so I'll make sure to address this then (either confirm that the copy is shallow enough to not cause a perf hit or remove the copy operation).

[robdiciuccio]

Thanks, @villebro. DataFrame.copy() is deep by default.