datafusion_comet_spark_expr/array_funcs/
array_insert.rs

1// Licensed to the Apache Software Foundation (ASF) under one
2// or more contributor license agreements.  See the NOTICE file
3// distributed with this work for additional information
4// regarding copyright ownership.  The ASF licenses this file
5// to you under the Apache License, Version 2.0 (the
6// "License"); you may not use this file except in compliance
7// with the License.  You may obtain a copy of the License at
8//
9//   http://www.apache.org/licenses/LICENSE-2.0
10//
11// Unless required by applicable law or agreed to in writing,
12// software distributed under the License is distributed on an
13// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14// KIND, either express or implied.  See the License for the
15// specific language governing permissions and limitations
16// under the License.
17
18use arrow::array::{make_array, Array, ArrayRef, GenericListArray, Int32Array, OffsetSizeTrait};
19use arrow::datatypes::{DataType, Field, Schema};
20use arrow::{
21    array::{as_primitive_array, Capacities, MutableArrayData},
22    buffer::{NullBuffer, OffsetBuffer},
23    datatypes::ArrowNativeType,
24    record_batch::RecordBatch,
25};
26use datafusion::common::{
27    cast::{as_large_list_array, as_list_array},
28    internal_err, DataFusionError, Result as DataFusionResult,
29};
30use datafusion::logical_expr::ColumnarValue;
31use datafusion::physical_expr::PhysicalExpr;
32use std::hash::Hash;
33use std::{
34    any::Any,
35    fmt::{Debug, Display, Formatter},
36    sync::Arc,
37};
38
39// 2147483632 == java.lang.Integer.MAX_VALUE - 15
40// It is a value of ByteArrayUtils.MAX_ROUNDED_ARRAY_LENGTH
41// https://github.com/apache/spark/blob/master/common/utils/src/main/java/org/apache/spark/unsafe/array/ByteArrayUtils.java
42const MAX_ROUNDED_ARRAY_LENGTH: usize = 2147483632;
43
44#[derive(Debug, Eq)]
45pub struct ArrayInsert {
46    src_array_expr: Arc<dyn PhysicalExpr>,
47    pos_expr: Arc<dyn PhysicalExpr>,
48    item_expr: Arc<dyn PhysicalExpr>,
49    legacy_negative_index: bool,
50}
51
52impl Hash for ArrayInsert {
53    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
54        self.src_array_expr.hash(state);
55        self.pos_expr.hash(state);
56        self.item_expr.hash(state);
57        self.legacy_negative_index.hash(state);
58    }
59}
60impl PartialEq for ArrayInsert {
61    fn eq(&self, other: &Self) -> bool {
62        self.src_array_expr.eq(&other.src_array_expr)
63            && self.pos_expr.eq(&other.pos_expr)
64            && self.item_expr.eq(&other.item_expr)
65            && self.legacy_negative_index.eq(&other.legacy_negative_index)
66    }
67}
68
69impl ArrayInsert {
70    pub fn new(
71        src_array_expr: Arc<dyn PhysicalExpr>,
72        pos_expr: Arc<dyn PhysicalExpr>,
73        item_expr: Arc<dyn PhysicalExpr>,
74        legacy_negative_index: bool,
75    ) -> Self {
76        Self {
77            src_array_expr,
78            pos_expr,
79            item_expr,
80            legacy_negative_index,
81        }
82    }
83
84    pub fn array_type(&self, data_type: &DataType) -> DataFusionResult<DataType> {
85        match data_type {
86            DataType::List(field) => Ok(DataType::List(Arc::clone(field))),
87            DataType::LargeList(field) => Ok(DataType::LargeList(Arc::clone(field))),
88            data_type => Err(DataFusionError::Internal(format!(
89                "Unexpected src array type in ArrayInsert: {data_type:?}"
90            ))),
91        }
92    }
93}
94
95impl PhysicalExpr for ArrayInsert {
96    fn as_any(&self) -> &dyn Any {
97        self
98    }
99
100    fn fmt_sql(&self, _: &mut Formatter<'_>) -> std::fmt::Result {
101        unimplemented!()
102    }
103
104    fn data_type(&self, input_schema: &Schema) -> DataFusionResult<DataType> {
105        self.array_type(&self.src_array_expr.data_type(input_schema)?)
106    }
107
108    fn nullable(&self, input_schema: &Schema) -> DataFusionResult<bool> {
109        self.src_array_expr.nullable(input_schema)
110    }
111
112    fn evaluate(&self, batch: &RecordBatch) -> DataFusionResult<ColumnarValue> {
113        let pos_value = self
114            .pos_expr
115            .evaluate(batch)?
116            .into_array(batch.num_rows())?;
117
118        // Spark supports only IntegerType (Int32):
119        // https://github.com/apache/spark/blob/branch-3.5/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/collectionOperations.scala#L4737
120        if !matches!(pos_value.data_type(), DataType::Int32) {
121            return Err(DataFusionError::Internal(format!(
122                "Unexpected index data type in ArrayInsert: {:?}, expected type is Int32",
123                pos_value.data_type()
124            )));
125        }
126
127        // Check that src array is actually an array and get it's value type
128        let src_value = self
129            .src_array_expr
130            .evaluate(batch)?
131            .into_array(batch.num_rows())?;
132
133        let src_element_type = match self.array_type(src_value.data_type())? {
134            DataType::List(field) => &field.data_type().clone(),
135            DataType::LargeList(field) => &field.data_type().clone(),
136            _ => unreachable!(),
137        };
138
139        // Check that inserted value has the same type as an array
140        let item_value = self
141            .item_expr
142            .evaluate(batch)?
143            .into_array(batch.num_rows())?;
144        if item_value.data_type() != src_element_type {
145            return Err(DataFusionError::Internal(format!(
146                "Type mismatch in ArrayInsert: array type is {:?} but item type is {:?}",
147                src_element_type,
148                item_value.data_type()
149            )));
150        }
151
152        match src_value.data_type() {
153            DataType::List(_) => {
154                let list_array = as_list_array(&src_value)?;
155                array_insert(
156                    list_array,
157                    &item_value,
158                    &pos_value,
159                    self.legacy_negative_index,
160                )
161            }
162            DataType::LargeList(_) => {
163                let list_array = as_large_list_array(&src_value)?;
164                array_insert(
165                    list_array,
166                    &item_value,
167                    &pos_value,
168                    self.legacy_negative_index,
169                )
170            }
171            _ => unreachable!(), // This case is checked already
172        }
173    }
174
175    fn children(&self) -> Vec<&Arc<dyn PhysicalExpr>> {
176        vec![&self.src_array_expr, &self.pos_expr, &self.item_expr]
177    }
178
179    fn with_new_children(
180        self: Arc<Self>,
181        children: Vec<Arc<dyn PhysicalExpr>>,
182    ) -> DataFusionResult<Arc<dyn PhysicalExpr>> {
183        match children.len() {
184            3 => Ok(Arc::new(ArrayInsert::new(
185                Arc::clone(&children[0]),
186                Arc::clone(&children[1]),
187                Arc::clone(&children[2]),
188                self.legacy_negative_index,
189            ))),
190            _ => internal_err!("ArrayInsert should have exactly three childrens"),
191        }
192    }
193}
194
195fn array_insert<O: OffsetSizeTrait>(
196    list_array: &GenericListArray<O>,
197    items_array: &ArrayRef,
198    pos_array: &ArrayRef,
199    legacy_mode: bool,
200) -> DataFusionResult<ColumnarValue> {
201    // The code is based on the implementation of the array_append from the Apache DataFusion
202    // https://github.com/apache/datafusion/blob/main/datafusion/functions-nested/src/concat.rs#L513
203    //
204    // This code is also based on the implementation of the array_insert from the Apache Spark
205    // https://github.com/apache/spark/blob/branch-3.5/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/collectionOperations.scala#L4713
206
207    let values = list_array.values();
208    let offsets = list_array.offsets();
209    let values_data = values.to_data();
210    let item_data = items_array.to_data();
211    let new_capacity = Capacities::Array(values_data.len() + item_data.len());
212
213    let mut mutable_values =
214        MutableArrayData::with_capacities(vec![&values_data, &item_data], true, new_capacity);
215
216    let mut new_offsets = vec![O::usize_as(0)];
217    let mut new_nulls = Vec::<bool>::with_capacity(list_array.len());
218
219    let pos_data: &Int32Array = as_primitive_array(&pos_array); // Spark supports only i32 for positions
220
221    for (row_index, offset_window) in offsets.windows(2).enumerate() {
222        let pos = pos_data.values()[row_index];
223        let start = offset_window[0].as_usize();
224        let end = offset_window[1].as_usize();
225        let is_item_null = items_array.is_null(row_index);
226
227        if list_array.is_null(row_index) {
228            // In Spark if value of the array is NULL than nothing happens
229            mutable_values.extend_nulls(1);
230            new_offsets.push(new_offsets[row_index] + O::one());
231            new_nulls.push(false);
232            continue;
233        }
234
235        if pos == 0 {
236            return Err(DataFusionError::Internal(
237                "Position for array_insert should be greter or less than zero".to_string(),
238            ));
239        }
240
241        if (pos > 0) || ((-pos).as_usize() < (end - start + 1)) {
242            let corrected_pos = if pos > 0 {
243                (pos - 1).as_usize()
244            } else {
245                end - start - (-pos).as_usize() + if legacy_mode { 0 } else { 1 }
246            };
247            let new_array_len = std::cmp::max(end - start + 1, corrected_pos);
248            if new_array_len > MAX_ROUNDED_ARRAY_LENGTH {
249                return Err(DataFusionError::Internal(format!(
250                    "Max array length in Spark is {MAX_ROUNDED_ARRAY_LENGTH:?}, but got {new_array_len:?}"
251                )));
252            }
253
254            if (start + corrected_pos) <= end {
255                mutable_values.extend(0, start, start + corrected_pos);
256                mutable_values.extend(1, row_index, row_index + 1);
257                mutable_values.extend(0, start + corrected_pos, end);
258                new_offsets.push(new_offsets[row_index] + O::usize_as(new_array_len));
259            } else {
260                mutable_values.extend(0, start, end);
261                mutable_values.extend_nulls(new_array_len - (end - start));
262                mutable_values.extend(1, row_index, row_index + 1);
263                // In that case spark actualy makes array longer than expected;
264                // For example, if pos is equal to 5, len is eq to 3, than resulted len will be 5
265                new_offsets.push(new_offsets[row_index] + O::usize_as(new_array_len) + O::one());
266            }
267        } else {
268            // This comment is takes from the Apache Spark source code as is:
269            // special case- if the new position is negative but larger than the current array size
270            // place the new item at start of array, place the current array contents at the end
271            // and fill the newly created array elements inbetween with a null
272            let base_offset = if legacy_mode { 1 } else { 0 };
273            let new_array_len = (-pos + base_offset).as_usize();
274            if new_array_len > MAX_ROUNDED_ARRAY_LENGTH {
275                return Err(DataFusionError::Internal(format!(
276                    "Max array length in Spark is {MAX_ROUNDED_ARRAY_LENGTH:?}, but got {new_array_len:?}"
277                )));
278            }
279            mutable_values.extend(1, row_index, row_index + 1);
280            mutable_values.extend_nulls(new_array_len - (end - start + 1));
281            mutable_values.extend(0, start, end);
282            new_offsets.push(new_offsets[row_index] + O::usize_as(new_array_len));
283        }
284        if is_item_null {
285            if (start == end) || (values.is_null(row_index)) {
286                new_nulls.push(false)
287            } else {
288                new_nulls.push(true)
289            }
290        } else {
291            new_nulls.push(true)
292        }
293    }
294
295    let data = make_array(mutable_values.freeze());
296    let data_type = match list_array.data_type() {
297        DataType::List(field) => field.data_type(),
298        DataType::LargeList(field) => field.data_type(),
299        _ => unreachable!(),
300    };
301    let new_array = GenericListArray::<O>::try_new(
302        Arc::new(Field::new("item", data_type.clone(), true)),
303        OffsetBuffer::new(new_offsets.into()),
304        data,
305        Some(NullBuffer::new(new_nulls.into())),
306    )?;
307
308    Ok(ColumnarValue::Array(Arc::new(new_array)))
309}
310
311impl Display for ArrayInsert {
312    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
313        write!(
314            f,
315            "ArrayInsert [array: {:?}, pos: {:?}, item: {:?}]",
316            self.src_array_expr, self.pos_expr, self.item_expr
317        )
318    }
319}
320
321#[cfg(test)]
322mod test {
323    use super::*;
324    use arrow::array::{Array, ArrayRef, Int32Array, ListArray};
325    use arrow::datatypes::Int32Type;
326    use datafusion::common::Result;
327    use datafusion::physical_plan::ColumnarValue;
328    use std::sync::Arc;
329
330    #[test]
331    fn test_array_insert() -> Result<()> {
332        // Test inserting an item into a list array
333        // Inputs and expected values are taken from the Spark results
334        let list = ListArray::from_iter_primitive::<Int32Type, _, _>(vec![
335            Some(vec![Some(1), Some(2), Some(3)]),
336            Some(vec![Some(4), Some(5)]),
337            Some(vec![None]),
338            Some(vec![Some(1), Some(2), Some(3)]),
339            Some(vec![Some(1), Some(2), Some(3)]),
340            None,
341        ]);
342
343        let positions = Int32Array::from(vec![2, 1, 1, 5, 6, 1]);
344        let items = Int32Array::from(vec![
345            Some(10),
346            Some(20),
347            Some(30),
348            Some(100),
349            Some(100),
350            Some(40),
351        ]);
352
353        let ColumnarValue::Array(result) = array_insert(
354            &list,
355            &(Arc::new(items) as ArrayRef),
356            &(Arc::new(positions) as ArrayRef),
357            false,
358        )?
359        else {
360            unreachable!()
361        };
362
363        let expected = ListArray::from_iter_primitive::<Int32Type, _, _>(vec![
364            Some(vec![Some(1), Some(10), Some(2), Some(3)]),
365            Some(vec![Some(20), Some(4), Some(5)]),
366            Some(vec![Some(30), None]),
367            Some(vec![Some(1), Some(2), Some(3), None, Some(100)]),
368            Some(vec![Some(1), Some(2), Some(3), None, None, Some(100)]),
369            None,
370        ]);
371
372        assert_eq!(&result.to_data(), &expected.to_data());
373
374        Ok(())
375    }
376
377    #[test]
378    fn test_array_insert_negative_index() -> Result<()> {
379        // Test insert with negative index
380        // Inputs and expected values are taken from the Spark results
381        let list = ListArray::from_iter_primitive::<Int32Type, _, _>(vec![
382            Some(vec![Some(1), Some(2), Some(3)]),
383            Some(vec![Some(4), Some(5)]),
384            Some(vec![Some(1)]),
385            None,
386        ]);
387
388        let positions = Int32Array::from(vec![-2, -1, -3, -1]);
389        let items = Int32Array::from(vec![Some(10), Some(20), Some(100), Some(30)]);
390
391        let ColumnarValue::Array(result) = array_insert(
392            &list,
393            &(Arc::new(items) as ArrayRef),
394            &(Arc::new(positions) as ArrayRef),
395            false,
396        )?
397        else {
398            unreachable!()
399        };
400
401        let expected = ListArray::from_iter_primitive::<Int32Type, _, _>(vec![
402            Some(vec![Some(1), Some(2), Some(10), Some(3)]),
403            Some(vec![Some(4), Some(5), Some(20)]),
404            Some(vec![Some(100), None, Some(1)]),
405            None,
406        ]);
407
408        assert_eq!(&result.to_data(), &expected.to_data());
409
410        Ok(())
411    }
412
413    #[test]
414    fn test_array_insert_legacy_mode() -> Result<()> {
415        // Test the so-called "legacy" mode exisiting in the Spark
416        let list = ListArray::from_iter_primitive::<Int32Type, _, _>(vec![
417            Some(vec![Some(1), Some(2), Some(3)]),
418            Some(vec![Some(4), Some(5)]),
419            None,
420        ]);
421
422        let positions = Int32Array::from(vec![-1, -1, -1]);
423        let items = Int32Array::from(vec![Some(10), Some(20), Some(30)]);
424
425        let ColumnarValue::Array(result) = array_insert(
426            &list,
427            &(Arc::new(items) as ArrayRef),
428            &(Arc::new(positions) as ArrayRef),
429            true,
430        )?
431        else {
432            unreachable!()
433        };
434
435        let expected = ListArray::from_iter_primitive::<Int32Type, _, _>(vec![
436            Some(vec![Some(1), Some(2), Some(10), Some(3)]),
437            Some(vec![Some(4), Some(20), Some(5)]),
438            None,
439        ]);
440
441        assert_eq!(&result.to_data(), &expected.to_data());
442
443        Ok(())
444    }
445}
datafusion_comet_spark_expr/array_funcs/array_insert.rs

datafusion_comet_spark_expr/array_funcs/
array_insert.rs
