pl.write_database with adbc not writing to a specified schema

[StuffbyYuki]

Checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of Polars.

Reproducible example

# my dataframe
df = pl.DataFrame({'brand': ['Volvo', 'BMW', 'Toyota'], 'model': ['p1800', 'M1', 'Celica'], 'year': [1968, 1978, 1975]})

# this results in writing the dataframe to the public schema
df.write_database(table_name="sandbox.cars_output", connection=uri, engine="adbc", if_exists='replace')

Log output

No response

Issue description

When writing a dataframe to a Postgres database with pl.write_database using adbc, the specified schema don't get reflected in the result. It writes to the public schema instead. And the whole string I specify in the table_name parameter becomes the table name created in the public schema. But with sqlalchemy, it writes to the specified scheme with no problems.

Expected behavior

I expect the dataframe written to the specified schema, "sandbox", instead of the "public".

Installed versions

--------Version info---------
Polars:               0.19.17
Index type:           UInt32
Platform:             macOS-14.1.1-arm64-arm-64bit
Python:               3.11.3 (main, Apr 15 2023, 22:27:28) [Clang 14.0.3 (clang-1403.0.22.14.1)]

----Optional dependencies----
adbc_driver_manager:  0.7.0
cloudpickle:          <not installed>
connectorx:           0.3.2
deltalake:            0.12.0
fsspec:               2023.10.0
gevent:               <not installed>
matplotlib:           3.8.0
numpy:                1.26.1
openpyxl:             <not installed>
pandas:               2.1.1
pyarrow:              12.0.0
pydantic:             2.4.2
pyiceberg:            0.5.0
pyxlsb:               <not installed>
sqlalchemy:           2.0.22
xlsx2csv:             <not installed>
xlsxwriter:           <not installed>

[deanm0000]

This is probably an issue for adbc rather than polars.

That said I believe you can do what you need with

df.write_database(table_name="cars_output", connection=f"{uri}?currentSchema=sandbox", engine="adbc", if_exists='replace')

Citing this for that hint.

If your uri already has parameters then change my ? to a & alternatively you could use a library to make it more robust like this

[deanm0000]

@stinodego might be a close, not sure if polars wants to take this on or leave to adbc and possibly connectorx or leave to user to augment their conn string.

[alexander-beedie]

I actually raised this issue with the adbc folks a while ago (end of September), and they have since both confirmed the bug and fixed it: apache/arrow-adbc#1109

If you install their recent 0.8.0 release (I see you're on 0.7.0) it'll work.

You should probably also update to the latest polars release 0.19.18, as I added some additional checks and affordances relating to adbc; for instance, if we see you relying on a db schema name with an adbc version below 0.8.0 we'll raise an error and tell you to go upgrade, eg:

ModuleNotFoundError("use of schema-qualified table names requires ADBC version >= 0.8, found 0.7.0")

[StuffbyYuki]

@alexander-beedie Updated both versions and it works like a charm. Thanks guys!