alamb commented on code in PR #9222:
URL: https://github.com/apache/arrow-datafusion/pull/9222#discussion_r1488587841
##########
datafusion/physical-expr/src/array_expressions.rs:
##########
@@ -229,121 +228,6 @@ fn check_datatypes(name: &str, args: &[&ArrayRef]) ->
Result<()> {
Ok(())
}
-/// Convert one or more [`ArrayRef`] of the same type into a
Review Comment:
I moved the code from here to the datafusion-function-array crate
##########
datafusion/functions-array/src/kernels.rs:
##########
@@ -1,254 +0,0 @@
-// Licensed to the Apache Software Foundation (ASF) under one
Review Comment:
I moved this code to `to_string.rs` to align with keeping the ScalarFunction
definitions inline with their implementations
##########
datafusion/functions-array/src/udf.rs:
##########
@@ -1,85 +0,0 @@
-// Licensed to the Apache Software Foundation (ASF) under one
-// or more contributor license agreements. See the NOTICE file
-// distributed with this work for additional information
-// regarding copyright ownership. The ASF licenses this file
-// to you under the Apache License, Version 2.0 (the
-// "License"); you may not use this file except in compliance
-// with the License. You may obtain a copy of the License at
-//
-// http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing,
-// software distributed under the License is distributed on an
-// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
-// KIND, either express or implied. See the License for the
-// specific language governing permissions and limitations
-// under the License.
-
-//! [`ScalarUDFImpl`] definitions for array functions.
-
-use arrow::datatypes::DataType;
-use datafusion_common::{plan_err, DataFusionError};
-use datafusion_expr::expr::ScalarFunction;
-use datafusion_expr::Expr;
-use datafusion_expr::{ColumnarValue, ScalarUDFImpl, Signature, Volatility};
-use std::any::Any;
-
-// Create static instances of ScalarUDFs for each function
-make_udf_function!(ArrayToString,
Review Comment:
moved to `to_string.rs`
##########
datafusion/optimizer/src/analyzer/rewrite_expr.rs:
##########
@@ -154,103 +154,35 @@ fn rewrite_array_concat_operator_to_func(
return None;
}
- match (left, right) {
+ fn fn_name(expr: &Expr) -> Option<&str> {
+ if let Expr::ScalarFunction(ScalarFunction { func_def, .. }) = expr {
+ Some(func_def.name())
+ } else {
+ None
+ }
+ }
+
+ // TODO figure out how to generalize this so it isn't hard coded by name
+ match (fn_name(left), fn_name(right)) {
Review Comment:
I am not sure why we have specific rewrites here for functions (that can't
be done by constant folding or some other method). I don't like special casing
the names
I need to look into it
##########
datafusion/functions-array/src/make_array.rs:
##########
@@ -0,0 +1,311 @@
+// Licensed to the Apache Software Foundation (ASF) under one
+// or more contributor license agreements. See the NOTICE file
+// distributed with this work for additional information
+// regarding copyright ownership. The ASF licenses this file
+// to you under the Apache License, Version 2.0 (the
+// "License"); you may not use this file except in compliance
+// with the License. You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing,
+// software distributed under the License is distributed on an
+// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+// KIND, either express or implied. See the License for the
+// specific language governing permissions and limitations
+// under the License.
+
+//! implementation of make_array function
+
+use arrow::array::{
+ new_null_array, Array, ArrayData, ArrayRef, Capacities, GenericListArray,
+ MutableArrayData, NullArray, OffsetSizeTrait,
+};
+use arrow::buffer::OffsetBuffer;
+use arrow::datatypes::{DataType, Field};
+use datafusion_common::utils::array_into_list_array;
+use datafusion_common::{plan_err, DataFusionError, Result};
+use datafusion_expr::{
+ ColumnarValue, Expr, ScalarUDFImpl, Signature, TypeSignature, Volatility,
+};
+use std::any::Any;
+use std::sync::Arc;
+
+// Create static instances of ScalarUDFs for each function
+make_udf_function!(
+ MakeArray,
+ make_array,
+ array,
+ "returns an Arrow array using the specified input expressions.",
+ udf
+);
+
+/// Create a new array from a list of expressions
+pub fn array(args: Vec<Expr>) -> Expr {
+ udf().call(args)
+}
+
+#[derive(Debug)]
+pub(super) struct MakeArray {
+ signature: Signature,
+ aliases: Vec<String>,
+}
+
+impl MakeArray {
+ pub fn new() -> Self {
+ Self {
+ signature: // 0 or more arguments of arbitrary type
+ Signature::one_of(vec![TypeSignature::VariadicEqual,
TypeSignature::Any(0)],
+ Volatility::Immutable),
+ aliases: vec![
+ "make_list".to_string(),
+ ],
+ }
+ }
+}
+
+impl ScalarUDFImpl for MakeArray {
+ fn as_any(&self) -> &dyn Any {
+ self
+ }
+ fn name(&self) -> &str {
+ "make_array"
+ }
+
+ fn signature(&self) -> &Signature {
+ &self.signature
+ }
+
+ fn return_type(&self, arg_types: &[DataType]) -> Result<DataType> {
+ use DataType::*;
+ match arg_types.len() {
+ 0 => Ok(List(Arc::new(Field::new("item", Null, true)))),
+ _ => {
+ let mut expr_type = Null;
+ for input_expr_type in arg_types {
+ if !input_expr_type.equals_datatype(&Null) {
+ expr_type = input_expr_type.clone();
+ break;
+ }
+ }
+
+ Ok(List(Arc::new(Field::new("item", expr_type, true))))
+ }
+ }
+ }
+
+ fn invoke(&self, args: &[ColumnarValue]) -> Result<ColumnarValue> {
+ make_array_inner(&ColumnarValue::values_to_arrays(args)?)
+ .map(ColumnarValue::Array)
+ }
+
+ fn aliases(&self) -> &[String] {
+ &self.aliases
+ }
+}
+
+/// `make_array` SQL function
+pub(crate) fn make_array_inner(arrays: &[ArrayRef]) -> Result<ArrayRef> {
+ let mut data_type = DataType::Null;
+ for arg in arrays {
+ let arg_data_type = arg.data_type();
+ if !arg_data_type.equals_datatype(&DataType::Null) {
+ data_type = arg_data_type.clone();
+ break;
+ }
+ }
+
+ match data_type {
+ // Either an empty array or all nulls:
+ DataType::Null => {
+ let array =
+ new_null_array(&DataType::Null, arrays.iter().map(|a|
a.len()).sum());
+ Ok(Arc::new(array_into_list_array(array)))
+ }
+ DataType::LargeList(..) => array_array::<i64>(arrays, data_type),
+ _ => array_array::<i32>(arrays, data_type),
+ }
+}
+
+/// Convert one or more [`ArrayRef`] of the same type into a
+/// `ListArray` or 'LargeListArray' depending on the offset size.
+///
+/// # Example (non nested)
+///
+/// Calling `array(col1, col2)` where col1 and col2 are non nested
+/// would return a single new `ListArray`, where each row was a list
+/// of 2 elements:
+///
+/// ```text
+/// ┌─────────┐ ┌─────────┐ ┌──────────────┐
+/// │ ┌─────┐ │ │ ┌─────┐ │ │ ┌──────────┐ │
+/// │ │ A │ │ │ │ X │ │ │ │ [A, X] │ │
+/// │ ├─────┤ │ │ ├─────┤ │ │ ├──────────┤ │
+/// │ │NULL │ │ │ │ Y │ │──────────▶│ │[NULL, Y] │ │
+/// │ ├─────┤ │ │ ├─────┤ │ │ ├──────────┤ │
+/// │ │ C │ │ │ │ Z │ │ │ │ [C, Z] │ │
+/// │ └─────┘ │ │ └─────┘ │ │ └──────────┘ │
+/// └─────────┘ └─────────┘ └──────────────┘
+/// col1 col2 output
+/// ```
+///
+/// # Example (nested)
+///
+/// Calling `array(col1, col2)` where col1 and col2 are lists
+/// would return a single new `ListArray`, where each row was a list
+/// of the corresponding elements of col1 and col2.
+///
+/// ``` text
+/// ┌──────────────┐ ┌──────────────┐ ┌─────────────────────────────┐
+/// │ ┌──────────┐ │ │ ┌──────────┐ │ │ ┌────────────────────────┐ │
+/// │ │ [A, X] │ │ │ │ [] │ │ │ │ [[A, X], []] │ │
+/// │ ├──────────┤ │ │ ├──────────┤ │ │ ├────────────────────────┤ │
+/// │ │[NULL, Y] │ │ │ │[Q, R, S] │ │───────▶│ │ [[NULL, Y], [Q, R, S]] │ │
+/// │ ├──────────┤ │ │ ├──────────┤ │ │ ├────────────────────────│ │
+/// │ │ [C, Z] │ │ │ │ NULL │ │ │ │ [[C, Z], NULL] │ │
+/// │ └──────────┘ │ │ └──────────┘ │ │ └────────────────────────┘ │
+/// └──────────────┘ └──────────────┘ └─────────────────────────────┘
+/// col1 col2 output
+/// ```
+fn array_array<O: OffsetSizeTrait>(
+ args: &[ArrayRef],
+ data_type: DataType,
+) -> Result<ArrayRef> {
+ // do not accept 0 arguments.
+ if args.is_empty() {
+ return plan_err!("Array requires at least one argument");
+ }
+
+ let mut data = vec![];
+ let mut total_len = 0;
+ for arg in args {
+ let arg_data = if arg.as_any().is::<NullArray>() {
+ ArrayData::new_empty(&data_type)
+ } else {
+ arg.to_data()
+ };
+ total_len += arg_data.len();
+ data.push(arg_data);
+ }
+
+ let mut offsets: Vec<O> = Vec::with_capacity(total_len);
+ offsets.push(O::usize_as(0));
+
+ let capacity = Capacities::Array(total_len);
+ let data_ref = data.iter().collect::<Vec<_>>();
+ let mut mutable = MutableArrayData::with_capacities(data_ref, true,
capacity);
+
+ let num_rows = args[0].len();
+ for row_idx in 0..num_rows {
+ for (arr_idx, arg) in args.iter().enumerate() {
+ if !arg.as_any().is::<NullArray>()
+ && !arg.is_null(row_idx)
+ && arg.is_valid(row_idx)
+ {
+ mutable.extend(arr_idx, row_idx, row_idx + 1);
+ } else {
+ mutable.extend_nulls(1);
+ }
+ }
+ offsets.push(O::usize_as(mutable.len()));
+ }
+ let data = mutable.freeze();
+
+ Ok(Arc::new(GenericListArray::<O>::try_new(
+ Arc::new(Field::new("item", data_type, true)),
+ OffsetBuffer::new(offsets.into()),
+ arrow::array::make_array(data),
+ None,
+ )?))
+}
+
+#[cfg(test)]
+mod tests {
+ use super::*;
+ use arrow::array::ListArray;
+ use arrow::datatypes::Int64Type;
+ use datafusion_common::cast::as_list_array;
+
+ /// Only test internal functions, array-related sql functions will be
tested in sqllogictest `array.slt`
+ #[test]
+ fn test_align_array_dimensions() {
+ let array1d_1 =
+ Arc::new(ListArray::from_iter_primitive::<Int64Type, _, _>(vec![
+ Some(vec![Some(1), Some(2), Some(3)]),
+ Some(vec![Some(4), Some(5)]),
+ ]));
+ let array1d_2 =
+ Arc::new(ListArray::from_iter_primitive::<Int64Type, _, _>(vec![
+ Some(vec![Some(6), Some(7), Some(8)]),
+ ]));
+
+ let array2d_1 = Arc::new(array_into_list_array(array1d_1.clone())) as
ArrayRef;
+ let array2d_2 = Arc::new(array_into_list_array(array1d_2.clone())) as
ArrayRef;
+
+ let res = align_array_dimensions::<i32>(vec![
+ array1d_1.to_owned(),
+ array2d_2.to_owned(),
+ ])
+ .unwrap();
+
+ let expected = as_list_array(&array2d_1).unwrap();
+ let expected_dim =
datafusion_common::utils::list_ndims(array2d_1.data_type());
+ assert_ne!(as_list_array(&res[0]).unwrap(), expected);
+ assert_eq!(
+ datafusion_common::utils::list_ndims(res[0].data_type()),
+ expected_dim
+ );
+
+ let array3d_1 = Arc::new(array_into_list_array(array2d_1)) as ArrayRef;
+ let array3d_2 = array_into_list_array(array2d_2.to_owned());
+ let res =
+ align_array_dimensions::<i32>(vec![array1d_1,
Arc::new(array3d_2.clone())])
+ .unwrap();
+
+ let expected = as_list_array(&array3d_1).unwrap();
+ let expected_dim =
datafusion_common::utils::list_ndims(array3d_1.data_type());
+ assert_ne!(as_list_array(&res[0]).unwrap(), expected);
+ assert_eq!(
+ datafusion_common::utils::list_ndims(res[0].data_type()),
+ expected_dim
+ );
+ }
+
+ fn align_array_dimensions<O: OffsetSizeTrait>(
Review Comment:
I copied this over from array_expressions where it is used in one other
place. I'll try and figure out how to remove it shortly
##########
datafusion/proto/proto/datafusion.proto:
##########
@@ -564,7 +564,7 @@ enum ScalarFunction {
Sqrt = 17;
Tan = 18;
Trunc = 19;
- Array = 20;
Review Comment:
array is an alias for make_array it seems
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
