Polars 1.34.0 Panics on UHugeInt and Intervals

[paultiq]

What happens?

While looking at [#98], I noticed that 1.34.0 added support for UINT128. Which lead to three problems:

1. polars_io.py is missing UINT128 and INT128*
2. external/duckdb - arrow_converter.cpp is missing a case for UHUGEINT, which results in a panic (in both 1.33.1 and 1.34.0).
3. Interval types panic in 1.34.0 but not in 1.33.1.

Fix for 1

Add UINT128 and INT128 to the list of types in polars_io.py.

* I didn't come up with a test case / failure here.

Fix for 2

The fix for 2 is to insert this before the HUGEINT case:
case LogicalTypeId::UHUGEINT:

Generally, there seems to be a lack of "dynamic" tests that exercise all types. See the reproduction below for an idea of a test case that'll smoke-test every type.

Fix for 3

The offending commit seems to be - pola-rs/polars@0ed3499

Ironically, the related issue says this is meant to fix a DuckDB issue: pola-rs/polars#15969

The workaround is to set os.environ["POLARS_IMPORT_INTERVAL_AS_STRUCT"]="1".

Unsure of what the correct fix is.

1.33.1 behavior - panics on UHUGEINT only

Polars version: 1.33.1
duckdb version: 1.5.0.dev73
Trying type='INTERVAL'
INTERVAL: Failed - The datatype "tin" is still not supported in Rust implementation
Trying type='UHUGEINT'

thread '' (1586047) panicked at crates/polars-core/src/datatypes/field.rs:280:19:
Arrow datatype Extension(ExtensionType { name: "arrow.opaque", inner: FixedSizeBinary(16), metadata: Some("{"type_name":"uhugeint","vendor_name":"DuckDB"}") }) not supported by Polars. You probably need to activate that data-type feature.
note: run with RUST_BACKTRACE=1 environment variable to display a backtrace

1.34.0 behavior: Panics on both INTERVALand UHUGEINT

Trying type='INTERVAL'

thread '' (1591205) panicked at crates/polars-core/src/datatypes/field.rs:283:85:
called Result::unwrap() on an Err value: ComputeError(ErrString("could not import from month_day_nano_interval type. Hint: This can be imported by setting POLARS_IMPORT_INTERVAL_AS_STRUCT=1 in the environment. Note however that this is unstable functionality that may change at any time."))
note: run with RUST_BACKTRACE=1 environment variable to display a backtrace

Trying type='UHUGEINT'

thread '' (1591205) panicked at crates/polars-core/src/datatypes/field.rs:290:19:
Arrow datatype Extension(ExtensionType { name: "arrow.opaque", inner: FixedSizeBinary(16), metadata: Some("{"type_name":"uhugeint","vendor_name":"DuckDB"}") }) not supported by Polars. You probably need to activate that data-type feature.

To Reproduce

* I'd suggest adding a variation of the test below so that any new types are exercised

import duckdb.typing
import duckdb
import polars as pl

print(f"Polars version: {pl.__version__}")
print(f"duckdb version: {duckdb.__version__}")

types = [c for c in filter(lambda x: str(x)[0]!="_", dir(duckdb.typing))]

for type in types:
    with duckdb.connect() as con:
        print(f"Trying {type=}")
        try:
            rel = con.sql(f"select 1::{type} as a")

            comparison_literal = pl.lit(1)
            lazy_df = rel.pl(lazy=True)
            print(f"{type}: Passed")
        except Exception as e:
            print(f"{type}: Failed - {e}")

OS:

Linux

DuckDB Package Version:

1.5.x (source build)

Python Version:

3.13

Full Name:

Paul T

Affiliation:

Iqmo

What is the latest build you tested with? If possible, we recommend testing with the latest nightly build.

I have not tested with any build

Did you include all relevant data sets for reproducing the issue?

No - Other reason (please specify in the issue body)

Did you include all code required to reproduce the issue?

• Yes, I have

Did you include all relevant configuration to reproduce the issue?

• Yes, I have