feat: Support Bucket and Truncate transforms on write #1345
Conversation
sungwy
force-pushed
the
bucket-transforms
branch
from
560ba20 to
bd80f39
Compare
sungwy
changed the title
sungwy
changed the title
kevinjqliu
left a comment
kevinjqliu
left a comment
There was a problem hiding this comment.
LGTM! Great to have writes for all the different transformations!
| @pytest.mark.parametrize( | ||
| "spec, expected_rows", | ||
| [ | ||
| # none of non-identity is supported |
There was a problem hiding this comment.
| # none of non-identity is supported |
| source_type: PrimitiveType, | ||
| input_arr: Union[pa.Array, pa.ChunkedArray], | ||
| expected: Union[pa.Array, pa.ChunkedArray], | ||
| num_buckets: int, |
There was a problem hiding this comment.
nit: wydt of reordering these for readability? num_buckets, source_type and input_arr are configs of the BucketTransform; expected is the output
There was a problem hiding this comment.
Hmm I think I feel indifferent here - there’s something nice about having the input and expected arrays side by side
kevinjqliu
left a comment
kevinjqliu
left a comment
There was a problem hiding this comment.
LGTM, thanks for rebasing the PR! I left a few questions on the tests.
We're currently blocked by CI since spark 3.5.3 is removed from https://dlcdn.apache.org/spark/
| (PartitionSpec(PartitionField(source_id=4, field_id=1001, transform=TruncateTransform(2), name="int_trunc"))), | ||
| (PartitionSpec(PartitionField(source_id=5, field_id=1001, transform=TruncateTransform(2), name="long_trunc"))), | ||
| (PartitionSpec(PartitionField(source_id=2, field_id=1001, transform=TruncateTransform(2), name="string_trunc"))), | ||
| (PartitionSpec(PartitionField(source_id=11, field_id=1001, transform=TruncateTransform(2), name="binary_trunc"))), |
There was a problem hiding this comment.
should we include binary_trunc too?
There was a problem hiding this comment.
Yeah good question. Truncating binary isn't supported with iceberg-rust so I've excluded this test case for now: https://github.com/apache/iceberg-rust/blob/main/crates/iceberg/src/transform/truncate.rs#L132-L164
| # mixed with non-identity is not supported | ||
| ( | ||
| PartitionSpec( | ||
| PartitionField(source_id=4, field_id=1001, transform=BucketTransform(2), name="int_bucket"), | ||
| PartitionField(source_id=1, field_id=1002, transform=IdentityTransform(), name="bool"), | ||
| ) | ||
| ), |
There was a problem hiding this comment.
is this case supported now?
pyiceberg/transforms.py
Outdated
| def _pyiceberg_transform_wrapper( | ||
| self, transform_func: Callable[["ArrayLike", Any], "ArrayLike"], *args: Any | ||
| ) -> Callable[["ArrayLike"], "ArrayLike"]: | ||
| import pyarrow as pa |
There was a problem hiding this comment.
| import pyarrow as pa | |
| try: | |
| import pyarrow as pa | |
| except ModuleNotFoundError as e: | |
| raise ModuleNotFoundError("For bucket/truncate transforms, PyArrow needs to be installed") from e |
no problem @Fokko - I'm just coming back from holidays myself. And thank you for taking another round of reviews @kevinjqliu ! I'll take the nits and retrigger the CI now that the spark artifact issue has been fixed |
|
Thanks @sungwy and I hope you had some great time off :) |
3 participants
Getting the PR ready for when
pyiceberg_coreis released fromiceberg-rustPR to introduce python binding release: apache/iceberg-rust#705
Fixes: #1074
Consideration: we could replace the existing
pyarrowdependency onorder_preservingtransforms (Month,Year,Date) withpyiceberg_corefor consistency