Python: Integration tests #6398
Conversation
Fokko
force-pushed
the
fd-integration-tests
branch
from
ad9dae5
to
05b8aed
Compare
Fokko
force-pushed
the
fd-integration-tests
branch
from
e8fc9e1
to
79b8e36
Compare
| python-version: '3.9' | ||
| cache: poetry | ||
| cache-dependency-path: | | ||
| ./python/poetry.lock |
There was a problem hiding this comment.
Should there be more than just the lock file?
There was a problem hiding this comment.
If you change the dependencies, you need to regenerate the lock file. So that should be enough
python/dev/spark-defaults.conf
Outdated
|
|
||
| spark.sql.extensions org.apache.iceberg.spark.extensions.IcebergSparkSessionExtensions | ||
| spark.sql.catalog.demo org.apache.iceberg.spark.SparkCatalog | ||
| spark.sql.catalog.demo.catalog-impl org.apache.iceberg.rest.RESTCatalog |
There was a problem hiding this comment.
Less is more, thanks!
| - minio:minio | ||
| rest: | ||
| image: tabulario/iceberg-rest:0.2.0 | ||
| container_name: pyiceberg-rest |
There was a problem hiding this comment.
Where does this store the underlying catalog metadata?
| @@ -428,11 +428,11 @@ def visit_not_in(self, term: BoundTerm[pc.Expression], literals: Set[Any]) -> pc | |||
|
|
|||
| def visit_is_nan(self, term: BoundTerm[Any]) -> pc.Expression: | |||
| ref = pc.field(term.ref().field.name) | |||
| return ref.is_null(nan_is_null=True) & ref.is_valid() | |||
| return pc.is_nan(ref) | |||
There was a problem hiding this comment.
This probably shouldn't be in this PR right? Seems like an update with a new version of pyarrow?
There was a problem hiding this comment.
This is actually to make the CI pass. I've created a PR to allow ref.is_nan() as well, but this is not released yet.
| @@ -331,7 +331,7 @@ def __init__( | |||
| self, | |||
| table: Table, | |||
| row_filter: Union[str, BooleanExpression] = ALWAYS_TRUE, | |||
| selected_fields: Tuple[str] = ("*",), | |||
| selected_fields: Tuple[str, ...] = ("*",), | |||
There was a problem hiding this comment.
This also seems like a separate PR change, but good cleanup.
python/tests/test_integration.py
Outdated
| arrow_table = table_test_null_nan.scan(row_filter=IsNaN("col_numeric"), selected_fields=("idx", "col_numeric")).to_arrow() | ||
| assert len(arrow_table) == 1 | ||
| assert arrow_table[0][0].as_py() == 1 | ||
| assert math.isnan(arrow_table[1][0].as_py()) |
There was a problem hiding this comment.
I think it would be easier to read these tests if you called as_py() to produce rows and validated the rows. It looks like there's just one row, but the row/column indexes are backward because this is columnar?
There was a problem hiding this comment.
Let me rewrite those tests a bit
There was a problem hiding this comment.
I've changed it into assert math.isnan(arrow_table["col_numeric"][0].as_py())
python/tests/test_integration.py
Outdated
| def test_duckdb_nan(table_test_null_nan_rewritten: Table) -> None: | ||
| con = table_test_null_nan_rewritten.scan().to_duckdb("table_test_null_nan") | ||
| result = con.query("SELECT idx FROM table_test_null_nan WHERE isnan(col_numeric)").fetchone() | ||
| assert result == (1,) |
|
Thanks for the review @rdblue we can more tests later on |
* Integration tests * First version * Add caching * Add caching * Restore pyproject * WIP * NaN seems to be broken * WIP * Coming along * Cleanup * Install duckdb * Cleanup * Revert changes to poetry * Make it even nicer * Revert unneeded change * Update Spark version * Make test passing * comments
This is the first version of a framework to read Iceberg tables, produced by Spark, using PyIceberg. This makes it easier to run end-to-end tests and also validate the behavior of PyArrow and DuckDB.