Add missing serde/inputFormat combination for Hive

[homar]

Description

Support for this was accidentally removed in scope of #18556 (424)

Additional context and related issues

Release notes

(x) This is not user-visible or is docs only, and no release notes are required.
( ) Release notes are required. Please propose a release note for me.
( ) Release notes are required, with the following suggested text:

# Section
* Add back support for ParquetHiveSerDe with TextInputFormat. ({issue}`issuenumber`)

[findepi]

This combination was working before so we should keep supporting it.

add information which release or which code change affected it

( ) Release notes are required, with the following suggested text:

# Section
* Fix some things. ({issue}`issuenumber`)

please feel in

[electrum]

This works for reads, but what happens if you attempt to write to such a table? I'm worried that this could silently create corrupted data that is not readable. I'd like to understand writing behavior for such a table with

• Current Trino
• Older version of Trino (before the change that prompted this fix)
• Hive
• Spark

[homar]

This works for reads, but what happens if you attempt to write to such a table? I'm worried that this could silently create corrupted data that is not readable. I'd like to understand writing behavior for such a table with

• Current Trino
• Older version of Trino (before the change that prompted this fix)
• Hive
• Spark

Current trino/previous trino can't write to it as it is a non managed hive table.
Spark as you can see from tests can read and write.
Hive seems not to able to write to it.

[findepi]

Current trino/previous trino can't write to it as it is a non managed hive table.

there is a toggle to enable writes to non-managed tables.

[electrum]

When you write from Spark, are the files written in Parquet format?

[electrum]

The test seems to only read with Spark, not write. Can you also read and write with Hive?

[homar]

@electrum sorry, my bad.
I modified the test.
Now it writes and reads with spark and it validates that the created file is parquet.
Reading and writing with hive doesn't really work, I also added code to validate that.
Writing with trino doesn't work with new version and it didn't work with the old. It fails on this line

trino/plugin/trino-hive/src/main/java/io/trino/plugin/hive/HiveMetadata.java

Line 3582 in 029e1a0

throw new TrinoException(HIVE_UNSUPPORTED_FORMAT, format("Output format %s with SerDe %s is not supported", outputFormat, serde));

[electrum]

Thanks for confirming the behavior. This change looks good. Can you also add a test to verify that writing fails with Trino? I'm assuming something similar to the Hive insert would work.

[homar]

@electrum can you merge this ? I don't have the button

[findepi]

Release notes are required, with the following suggested text:

# Section
* Add back support for ParquetHiveSerDe with TextInputFormat. ({issue}`issuenumber`)

or maybe "[Hive] Fix reading from certain Parquet tables incorrectly created by Spark"

cc @colebow