Upgrade deps to datafusion 41

Cargo.toml

@@ -38,13 +38,13 @@ tokio = { version = "1.39", features = ["macros", "rt", "rt-multi-thread", "sync
 rand = "0.8"
 pyo3 = { version = "0.21", features = ["extension-module", "abi3", "abi3-py38"] }
 arrow = { version = "52", feature = ["pyarrow"] }
-datafusion = { version = "40.0.0", features = ["pyarrow", "avro", "unicode_expressions"] }
-datafusion-common = { version = "40.0.0", features = ["pyarrow"] }
-datafusion-expr = "40.0.0"
-datafusion-functions-array = "40.0.0"
-datafusion-optimizer = "40.0.0"
-datafusion-sql = "40.0.0"
-datafusion-substrait = { version = "40.0.0", optional = true }
+datafusion = { version = "41.0.0", features = ["pyarrow", "avro", "unicode_expressions"] }
+datafusion-common = { version = "41.0.0", features = ["pyarrow"] }
+datafusion-expr = { version = "41.0.0" }
+datafusion-functions-nested = { version = "41.0.0" }
+datafusion-optimizer = { version = "41.0.0" }
+datafusion-sql = { version = "41.0.0" }
+datafusion-substrait = { version = "41.0.0", optional = true }
 prost = "0.12" # keep in line with `datafusion-substrait`
 prost-types = "0.12" # keep in line with `datafusion-substrait`
 uuid = { version = "1.9", features = ["v4"] }
@@ -56,7 +56,6 @@ parking_lot = "0.12"
 regex-syntax = "0.8"
 syn = "2.0.68"
 url = "2"
-sqlparser = "0.47.0"
 
 [build-dependencies]
 pyo3-build-config = "0.21"

python/datafusion/tests/test_functions.py

@@ -808,7 +808,7 @@ def test_regr_funcs_sql(df):
 
     assert result[0].column(0) == pa.array([None], type=pa.float64())
     assert result[0].column(1) == pa.array([None], type=pa.float64())
-    assert result[0].column(2) == pa.array([1], type=pa.float64())
+    assert result[0].column(2) == pa.array([1], type=pa.uint64())
     assert result[0].column(3) == pa.array([None], type=pa.float64())
     assert result[0].column(4) == pa.array([1], type=pa.float64())
     assert result[0].column(5) == pa.array([1], type=pa.float64())
@@ -840,7 +840,7 @@ def test_regr_funcs_sql_2():
     # Assertions for SQL results
     assert result_sql[0].column(0) == pa.array([2], type=pa.float64())
     assert result_sql[0].column(1) == pa.array([0], type=pa.float64())
-    assert result_sql[0].column(2) == pa.array([3], type=pa.float64()) # todo: i would not expect this to be float
+    assert result_sql[0].column(2) == pa.array([3], type=pa.uint64())
     assert result_sql[0].column(3) == pa.array([1], type=pa.float64())
     assert result_sql[0].column(4) == pa.array([2], type=pa.float64())
     assert result_sql[0].column(5) == pa.array([4], type=pa.float64())
@@ -852,7 +852,7 @@ def test_regr_funcs_sql_2():
 @pytest.mark.parametrize("func, expected", [
     pytest.param(f.regr_slope, pa.array([2], type=pa.float64()), id="regr_slope"),
     pytest.param(f.regr_intercept, pa.array([0], type=pa.float64()), id="regr_intercept"),
-    pytest.param(f.regr_count, pa.array([3], type=pa.float64()), id="regr_count"), # TODO: I would expect this to return an int array
+    pytest.param(f.regr_count, pa.array([3], type=pa.uint64()), id="regr_count"),
     pytest.param(f.regr_r2, pa.array([1], type=pa.float64()), id="regr_r2"),
     pytest.param(f.regr_avgx, pa.array([2], type=pa.float64()), id="regr_avgx"),
     pytest.param(f.regr_avgy, pa.array([4], type=pa.float64()), id="regr_avgy"),

src/catalog.rs

@@ -25,7 +25,7 @@ use crate::errors::DataFusionError;
 use crate::utils::wait_for_future;
 use datafusion::{
     arrow::pyarrow::ToPyArrow,
-    catalog::{schema::SchemaProvider, CatalogProvider},
+    catalog::{CatalogProvider, SchemaProvider},
     datasource::{TableProvider, TableType},
 };
 

src/common/data_type.rs

@@ -18,6 +18,7 @@
 use datafusion::arrow::array::Array;
 use datafusion::arrow::datatypes::{DataType, IntervalUnit, TimeUnit};
 use datafusion_common::{DataFusionError, ScalarValue};
+use datafusion_expr::sqlparser::ast::NullTreatment as DFNullTreatment;
 use pyo3::{exceptions::PyValueError, prelude::*};
 
 use crate::errors::py_datafusion_err;
@@ -775,20 +776,20 @@ pub enum NullTreatment {
     RESPECT_NULLS,
 }
 
-impl From<NullTreatment> for sqlparser::ast::NullTreatment {
-    fn from(null_treatment: NullTreatment) -> sqlparser::ast::NullTreatment {
+impl From<NullTreatment> for DFNullTreatment {
+    fn from(null_treatment: NullTreatment) -> DFNullTreatment {
         match null_treatment {
-            NullTreatment::IGNORE_NULLS => sqlparser::ast::NullTreatment::IgnoreNulls,
-            NullTreatment::RESPECT_NULLS => sqlparser::ast::NullTreatment::RespectNulls,
+            NullTreatment::IGNORE_NULLS => DFNullTreatment::IgnoreNulls,
+            NullTreatment::RESPECT_NULLS => DFNullTreatment::RespectNulls,
         }
     }
 }
 
-impl From<sqlparser::ast::NullTreatment> for NullTreatment {
-    fn from(null_treatment: sqlparser::ast::NullTreatment) -> NullTreatment {
+impl From<DFNullTreatment> for NullTreatment {
+    fn from(null_treatment: DFNullTreatment) -> NullTreatment {
         match null_treatment {
-            sqlparser::ast::NullTreatment::IgnoreNulls => NullTreatment::IGNORE_NULLS,
-            sqlparser::ast::NullTreatment::RespectNulls => NullTreatment::RESPECT_NULLS,
+            DFNullTreatment::IgnoreNulls => NullTreatment::IGNORE_NULLS,
+            DFNullTreatment::RespectNulls => NullTreatment::RESPECT_NULLS,
         }
     }
 }