`"UTMMedium"` field in the hits dataset causes a panic when performing SIMILAR TO.

[acking-you]

Describe the bug

The "UTMMedium" field in the hits dataset causes a panic when performing SIMILAR TO.

To Reproduce

Execute the SQL below:

select count(*) from hits where "UTMMedium" SIMILAR TO 'email|social|banner';

And you will got a panic,like this:

thread 'tokio-runtime-worker' panicked at xxx/src/expressions/binary.rs:616:27:
compute_utf8_flag_op_scalar failed to downcast array

Expected behavior

SQL should work or at least return errors properly.

Additional context

1. I tried to check the data and type of this field as follows:

> select arrow_typeof( "UTMMedium" )  from hits limit 1;
+------------------------------+
| arrow_typeof(hits.UTMMedium) |
+------------------------------+
| Utf8View                     |
+------------------------------+

> select "UTMMedium" from hits;
+---------------+
| UTMMedium     |
+---------------+
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
|               |
| cpc_marketing |
|               |
|               |
|               |
| cpc_marketing |
|               |
|               |
|               |
| cpc_marketing |
|               |
|               |
| .             |
| .             |
| .             |
+---------------+
99997497 row(s) fetched. (First 40 displayed. Use --maxrows to adjust)

2. I'm using datafusion-cli for testing, version 46.01.

[psiayn]

Hi, I was trying to reproduce this issue but I get a different error.

I am using this as the dataset: https://datasets.clickhouse.com/hits_compatible/athena_partitioned/[hits_1.parquet](https://datasets.clickhouse.com/hits_compatible/athena_partitioned/hits_1.parquet)
and using the latest 46.0.1 build from main.

This is the error I am getting.......

I hope I am using the correct dataset

[acking-you]

I am using this as the dataset: [https://datasets.clickhouse.com/hits_compatible/athena_partitioned/hits_1.parquet](https://datasets.clickhouse.com/hits_compatible/athena_partitioned/%5Bhits_1.parquet%5D(https://datasets.clickhouse.com/hits_compatible/athena_partitioned/hits_1.parquet))
and using the latest 46.0.1 build from main.

The hits file I used is approximately 14GB in size, automatically downloaded via a datafusion script. Let me find the corresponding download link for you.

May be that:https://datasets.clickhouse.com/hits_compatible/hits.parquet
And reference code: https://github.com/apache/datafusion/blob/main/benchmarks/bench.sh#L473

[psiayn]

Thank you, I will try with this

[psiayn]

Trying with this data I am not getting any error.... The dataset is 14gb

[acking-you]

Trying with this data I am not getting any error.... The dataset is 14gb

Thank you very much for trying. If so, then this panic may be related to my machine