datafusion-cli: results of group by aggregation query not showing

[Jefffrey]

Describe the bug

datafusion-cli isn't showing results for queries involving selecting count when group by.

To Reproduce

Given input file:

col1,col2
A,
B,b
C,c

Querying this file in datafusion-cli:

DataFusion CLI v34.0.0
❯ CREATE EXTERNAL TABLE kumachan STORED AS CSV WITH HEADER ROW LOCATION '/home/jeffrey/Downloads/kumachan.csv';
0 rows in set. Query took 0.029 seconds.

❯ select * from kumachan;
+------+------+
| col1 | col2 |
+------+------+
| A    |      |
| B    | b    |
| C    | c    |
+------+------+
3 rows in set. Query took 0.005 seconds.

❯ select count(*) from kumachan group by col1;
3 rows in set. Query took 0.010 seconds.

❯

Expected behavior

Should get results like:

+----------+
| COUNT(*) |
+----------+
| 1        |
| 1        |
| 1        |
+----------+

Additional context

No response

[alamb]

Maybe some other fallout from #8651 -- I had some work in progress to try and consolidate / clean up the output code more in https://github.com/alamb/arrow-datafusion/tree/alamb/cli-cleanup but I have not finished it

[alamb]

I think this is a regression and we should try and fix it before datafusion 35 release #8863 -- cc @andygrove

[alamb]

I can try to help debug this tomorrow if no one else beats me to it

[Jefffrey]

Problem is here:

https://github.com/apache/arrow-datafusion/blob/ffaa67904ed0ca454267ccc5832582bcb669a5c0/datafusion-cli/src/print_format.rs#L164

Specifically the batches[0].num_rows() == 0

It seems my query will have some empty RecordBatches initially, which causes this check to completely ignore printing the rest of the potentially data-filled RecordBatches

[alamb]

Thanks @Jefffrey -- I will make a patch

[alamb]

This appears to have been introduced in #8651

[alamb]

PR ready for review: #8895