arrow_array/array/
map_array.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 crate::array::{get_offsets, print_long_array};
19use crate::iterator::MapArrayIter;
20use crate::{make_array, Array, ArrayAccessor, ArrayRef, ListArray, StringArray, StructArray};
21use arrow_buffer::{ArrowNativeType, Buffer, NullBuffer, OffsetBuffer, ToByteSlice};
22use arrow_data::{ArrayData, ArrayDataBuilder};
23use arrow_schema::{ArrowError, DataType, Field, FieldRef};
24use std::any::Any;
25use std::sync::Arc;
26
27/// An array of key-value maps
28///
29/// Keys should always be non-null, but values can be null.
30///
31/// [`MapArray`] is physically a [`ListArray`] of key values pairs stored as an `entries`
32/// [`StructArray`] with 2 child fields.
33///
34/// See [`MapBuilder`](crate::builder::MapBuilder) for how to construct a [`MapArray`]
35#[derive(Clone)]
36pub struct MapArray {
37    data_type: DataType,
38    nulls: Option<NullBuffer>,
39    /// The [`StructArray`] that is the direct child of this array
40    entries: StructArray,
41    /// The start and end offsets of each entry
42    value_offsets: OffsetBuffer<i32>,
43}
44
45impl MapArray {
46    /// Create a new [`MapArray`] from the provided parts
47    ///
48    /// See [`MapBuilder`](crate::builder::MapBuilder) for a higher-level interface
49    /// to construct a [`MapArray`]
50    ///
51    /// # Errors
52    ///
53    /// Errors if
54    ///
55    /// * `offsets.len() - 1 != nulls.len()`
56    /// * `offsets.last() > entries.len()`
57    /// * `field.is_nullable()`
58    /// * `entries.null_count() != 0`
59    /// * `entries.columns().len() != 2`
60    /// * `field.data_type() != entries.data_type()`
61    pub fn try_new(
62        field: FieldRef,
63        offsets: OffsetBuffer<i32>,
64        entries: StructArray,
65        nulls: Option<NullBuffer>,
66        ordered: bool,
67    ) -> Result<Self, ArrowError> {
68        let len = offsets.len() - 1; // Offsets guaranteed to not be empty
69        let end_offset = offsets.last().unwrap().as_usize();
70        // don't need to check other values of `offsets` because they are checked
71        // during construction of `OffsetBuffer`
72        if end_offset > entries.len() {
73            return Err(ArrowError::InvalidArgumentError(format!(
74                "Max offset of {end_offset} exceeds length of entries {}",
75                entries.len()
76            )));
77        }
78
79        if let Some(n) = nulls.as_ref() {
80            if n.len() != len {
81                return Err(ArrowError::InvalidArgumentError(format!(
82                    "Incorrect length of null buffer for MapArray, expected {len} got {}",
83                    n.len(),
84                )));
85            }
86        }
87        if field.is_nullable() || entries.null_count() != 0 {
88            return Err(ArrowError::InvalidArgumentError(
89                "MapArray entries cannot contain nulls".to_string(),
90            ));
91        }
92
93        if field.data_type() != entries.data_type() {
94            return Err(ArrowError::InvalidArgumentError(format!(
95                "MapArray expected data type {} got {} for {:?}",
96                field.data_type(),
97                entries.data_type(),
98                field.name()
99            )));
100        }
101
102        if entries.columns().len() != 2 {
103            return Err(ArrowError::InvalidArgumentError(format!(
104                "MapArray entries must contain two children, got {}",
105                entries.columns().len()
106            )));
107        }
108
109        Ok(Self {
110            data_type: DataType::Map(field, ordered),
111            nulls,
112            entries,
113            value_offsets: offsets,
114        })
115    }
116
117    /// Create a new [`MapArray`] from the provided parts
118    ///
119    /// See [`MapBuilder`](crate::builder::MapBuilder) for a higher-level interface
120    /// to construct a [`MapArray`]
121    ///
122    /// # Panics
123    ///
124    /// Panics if [`Self::try_new`] returns an error
125    pub fn new(
126        field: FieldRef,
127        offsets: OffsetBuffer<i32>,
128        entries: StructArray,
129        nulls: Option<NullBuffer>,
130        ordered: bool,
131    ) -> Self {
132        Self::try_new(field, offsets, entries, nulls, ordered).unwrap()
133    }
134
135    /// Deconstruct this array into its constituent parts
136    pub fn into_parts(
137        self,
138    ) -> (
139        FieldRef,
140        OffsetBuffer<i32>,
141        StructArray,
142        Option<NullBuffer>,
143        bool,
144    ) {
145        let (f, ordered) = match self.data_type {
146            DataType::Map(f, ordered) => (f, ordered),
147            _ => unreachable!(),
148        };
149        (f, self.value_offsets, self.entries, self.nulls, ordered)
150    }
151
152    /// Returns a reference to the offsets of this map
153    ///
154    /// Unlike [`Self::value_offsets`] this returns the [`OffsetBuffer`]
155    /// allowing for zero-copy cloning
156    #[inline]
157    pub fn offsets(&self) -> &OffsetBuffer<i32> {
158        &self.value_offsets
159    }
160
161    /// Returns a reference to the keys of this map
162    pub fn keys(&self) -> &ArrayRef {
163        self.entries.column(0)
164    }
165
166    /// Returns a reference to the values of this map
167    pub fn values(&self) -> &ArrayRef {
168        self.entries.column(1)
169    }
170
171    /// Returns a reference to the [`StructArray`] entries of this map
172    pub fn entries(&self) -> &StructArray {
173        &self.entries
174    }
175
176    /// Returns the data type of the map's keys.
177    pub fn key_type(&self) -> &DataType {
178        self.keys().data_type()
179    }
180
181    /// Returns the data type of the map's values.
182    pub fn value_type(&self) -> &DataType {
183        self.values().data_type()
184    }
185
186    /// Returns ith value of this map array.
187    ///
188    /// Note: This method does not check for nulls and the value is arbitrary
189    /// if [`is_null`](Self::is_null) returns true for the index.
190    ///
191    /// # Safety
192    /// Caller must ensure that the index is within the array bounds
193    pub unsafe fn value_unchecked(&self, i: usize) -> StructArray {
194        let end = *self.value_offsets().get_unchecked(i + 1);
195        let start = *self.value_offsets().get_unchecked(i);
196        self.entries
197            .slice(start.to_usize().unwrap(), (end - start).to_usize().unwrap())
198    }
199
200    /// Returns ith value of this map array.
201    ///
202    /// This is a [`StructArray`] containing two fields
203    ///
204    /// Note: This method does not check for nulls and the value is arbitrary
205    /// (but still well-defined) if [`is_null`](Self::is_null) returns true for the index.
206    ///
207    /// # Panics
208    /// Panics if index `i` is out of bounds
209    pub fn value(&self, i: usize) -> StructArray {
210        let end = self.value_offsets()[i + 1] as usize;
211        let start = self.value_offsets()[i] as usize;
212        self.entries.slice(start, end - start)
213    }
214
215    /// Returns the offset values in the offsets buffer
216    #[inline]
217    pub fn value_offsets(&self) -> &[i32] {
218        &self.value_offsets
219    }
220
221    /// Returns the length for value at index `i`.
222    #[inline]
223    pub fn value_length(&self, i: usize) -> i32 {
224        let offsets = self.value_offsets();
225        offsets[i + 1] - offsets[i]
226    }
227
228    /// Returns a zero-copy slice of this array with the indicated offset and length.
229    pub fn slice(&self, offset: usize, length: usize) -> Self {
230        Self {
231            data_type: self.data_type.clone(),
232            nulls: self.nulls.as_ref().map(|n| n.slice(offset, length)),
233            entries: self.entries.clone(),
234            value_offsets: self.value_offsets.slice(offset, length),
235        }
236    }
237
238    /// constructs a new iterator
239    pub fn iter(&self) -> MapArrayIter<'_> {
240        MapArrayIter::new(self)
241    }
242}
243
244impl From<ArrayData> for MapArray {
245    fn from(data: ArrayData) -> Self {
246        Self::try_new_from_array_data(data)
247            .expect("Expected infallible creation of MapArray from ArrayData failed")
248    }
249}
250
251impl From<MapArray> for ArrayData {
252    fn from(array: MapArray) -> Self {
253        let len = array.len();
254        let builder = ArrayDataBuilder::new(array.data_type)
255            .len(len)
256            .nulls(array.nulls)
257            .buffers(vec![array.value_offsets.into_inner().into_inner()])
258            .child_data(vec![array.entries.to_data()]);
259
260        unsafe { builder.build_unchecked() }
261    }
262}
263
264impl MapArray {
265    fn try_new_from_array_data(data: ArrayData) -> Result<Self, ArrowError> {
266        if !matches!(data.data_type(), DataType::Map(_, _)) {
267            return Err(ArrowError::InvalidArgumentError(format!(
268                "MapArray expected ArrayData with DataType::Map got {}",
269                data.data_type()
270            )));
271        }
272
273        if data.buffers().len() != 1 {
274            return Err(ArrowError::InvalidArgumentError(format!(
275                "MapArray data should contain a single buffer only (value offsets), had {}",
276                data.len()
277            )));
278        }
279
280        if data.child_data().len() != 1 {
281            return Err(ArrowError::InvalidArgumentError(format!(
282                "MapArray should contain a single child array (values array), had {}",
283                data.child_data().len()
284            )));
285        }
286
287        let entries = data.child_data()[0].clone();
288
289        if let DataType::Struct(fields) = entries.data_type() {
290            if fields.len() != 2 {
291                return Err(ArrowError::InvalidArgumentError(format!(
292                    "MapArray should contain a struct array with 2 fields, have {} fields",
293                    fields.len()
294                )));
295            }
296        } else {
297            return Err(ArrowError::InvalidArgumentError(format!(
298                "MapArray should contain a struct array child, found {:?}",
299                entries.data_type()
300            )));
301        }
302        let entries = entries.into();
303
304        // SAFETY:
305        // ArrayData is valid, and verified type above
306        let value_offsets = unsafe { get_offsets(&data) };
307
308        Ok(Self {
309            data_type: data.data_type().clone(),
310            nulls: data.nulls().cloned(),
311            entries,
312            value_offsets,
313        })
314    }
315
316    /// Creates map array from provided keys, values and entry_offsets.
317    pub fn new_from_strings<'a>(
318        keys: impl Iterator<Item = &'a str>,
319        values: &dyn Array,
320        entry_offsets: &[u32],
321    ) -> Result<Self, ArrowError> {
322        let entry_offsets_buffer = Buffer::from(entry_offsets.to_byte_slice());
323        let keys_data = StringArray::from_iter_values(keys);
324
325        let keys_field = Arc::new(Field::new("keys", DataType::Utf8, false));
326        let values_field = Arc::new(Field::new(
327            "values",
328            values.data_type().clone(),
329            values.null_count() > 0,
330        ));
331
332        let entry_struct = StructArray::from(vec![
333            (keys_field, Arc::new(keys_data) as ArrayRef),
334            (values_field, make_array(values.to_data())),
335        ]);
336
337        let map_data_type = DataType::Map(
338            Arc::new(Field::new(
339                "entries",
340                entry_struct.data_type().clone(),
341                false,
342            )),
343            false,
344        );
345        let map_data = ArrayData::builder(map_data_type)
346            .len(entry_offsets.len() - 1)
347            .add_buffer(entry_offsets_buffer)
348            .add_child_data(entry_struct.into_data())
349            .build()?;
350
351        Ok(MapArray::from(map_data))
352    }
353}
354
355impl Array for MapArray {
356    fn as_any(&self) -> &dyn Any {
357        self
358    }
359
360    fn to_data(&self) -> ArrayData {
361        self.clone().into_data()
362    }
363
364    fn into_data(self) -> ArrayData {
365        self.into()
366    }
367
368    fn data_type(&self) -> &DataType {
369        &self.data_type
370    }
371
372    fn slice(&self, offset: usize, length: usize) -> ArrayRef {
373        Arc::new(self.slice(offset, length))
374    }
375
376    fn len(&self) -> usize {
377        self.value_offsets.len() - 1
378    }
379
380    fn is_empty(&self) -> bool {
381        self.value_offsets.len() <= 1
382    }
383
384    fn shrink_to_fit(&mut self) {
385        if let Some(nulls) = &mut self.nulls {
386            nulls.shrink_to_fit();
387        }
388        self.entries.shrink_to_fit();
389        self.value_offsets.shrink_to_fit();
390    }
391
392    fn offset(&self) -> usize {
393        0
394    }
395
396    fn nulls(&self) -> Option<&NullBuffer> {
397        self.nulls.as_ref()
398    }
399
400    fn logical_null_count(&self) -> usize {
401        // More efficient that the default implementation
402        self.null_count()
403    }
404
405    fn get_buffer_memory_size(&self) -> usize {
406        let mut size = self.entries.get_buffer_memory_size();
407        size += self.value_offsets.inner().inner().capacity();
408        if let Some(n) = self.nulls.as_ref() {
409            size += n.buffer().capacity();
410        }
411        size
412    }
413
414    fn get_array_memory_size(&self) -> usize {
415        let mut size = std::mem::size_of::<Self>() + self.entries.get_array_memory_size();
416        size += self.value_offsets.inner().inner().capacity();
417        if let Some(n) = self.nulls.as_ref() {
418            size += n.buffer().capacity();
419        }
420        size
421    }
422}
423
424impl ArrayAccessor for &MapArray {
425    type Item = StructArray;
426
427    fn value(&self, index: usize) -> Self::Item {
428        MapArray::value(self, index)
429    }
430
431    unsafe fn value_unchecked(&self, index: usize) -> Self::Item {
432        MapArray::value(self, index)
433    }
434}
435
436impl std::fmt::Debug for MapArray {
437    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
438        write!(f, "MapArray\n[\n")?;
439        print_long_array(self, f, |array, index, f| {
440            std::fmt::Debug::fmt(&array.value(index), f)
441        })?;
442        write!(f, "]")
443    }
444}
445
446impl From<MapArray> for ListArray {
447    fn from(value: MapArray) -> Self {
448        let field = match value.data_type() {
449            DataType::Map(field, _) => field,
450            _ => unreachable!("This should be a map type."),
451        };
452        let data_type = DataType::List(field.clone());
453        let builder = value.into_data().into_builder().data_type(data_type);
454        let array_data = unsafe { builder.build_unchecked() };
455
456        ListArray::from(array_data)
457    }
458}
459
460#[cfg(test)]
461mod tests {
462    use crate::cast::AsArray;
463    use crate::types::UInt32Type;
464    use crate::{Int32Array, UInt32Array};
465    use arrow_schema::Fields;
466
467    use super::*;
468
469    fn create_from_buffers() -> MapArray {
470        // Construct key and values
471        let keys_data = ArrayData::builder(DataType::Int32)
472            .len(8)
473            .add_buffer(Buffer::from([0, 1, 2, 3, 4, 5, 6, 7].to_byte_slice()))
474            .build()
475            .unwrap();
476        let values_data = ArrayData::builder(DataType::UInt32)
477            .len(8)
478            .add_buffer(Buffer::from(
479                [0u32, 10, 20, 30, 40, 50, 60, 70].to_byte_slice(),
480            ))
481            .build()
482            .unwrap();
483
484        // Construct a buffer for value offsets, for the nested array:
485        //  [[0, 1, 2], [3, 4, 5], [6, 7]]
486        let entry_offsets = Buffer::from([0, 3, 6, 8].to_byte_slice());
487
488        let keys = Arc::new(Field::new("keys", DataType::Int32, false));
489        let values = Arc::new(Field::new("values", DataType::UInt32, false));
490        let entry_struct = StructArray::from(vec![
491            (keys, make_array(keys_data)),
492            (values, make_array(values_data)),
493        ]);
494
495        // Construct a map array from the above two
496        let map_data_type = DataType::Map(
497            Arc::new(Field::new(
498                "entries",
499                entry_struct.data_type().clone(),
500                false,
501            )),
502            false,
503        );
504        let map_data = ArrayData::builder(map_data_type)
505            .len(3)
506            .add_buffer(entry_offsets)
507            .add_child_data(entry_struct.into_data())
508            .build()
509            .unwrap();
510        MapArray::from(map_data)
511    }
512
513    #[test]
514    fn test_map_array() {
515        // Construct key and values
516        let key_data = ArrayData::builder(DataType::Int32)
517            .len(8)
518            .add_buffer(Buffer::from([0, 1, 2, 3, 4, 5, 6, 7].to_byte_slice()))
519            .build()
520            .unwrap();
521        let value_data = ArrayData::builder(DataType::UInt32)
522            .len(8)
523            .add_buffer(Buffer::from(
524                [0u32, 10, 20, 0, 40, 0, 60, 70].to_byte_slice(),
525            ))
526            .null_bit_buffer(Some(Buffer::from(&[0b11010110])))
527            .build()
528            .unwrap();
529
530        // Construct a buffer for value offsets, for the nested array:
531        //  [[0, 1, 2], [3, 4, 5], [6, 7]]
532        let entry_offsets = Buffer::from([0, 3, 6, 8].to_byte_slice());
533
534        let keys_field = Arc::new(Field::new("keys", DataType::Int32, false));
535        let values_field = Arc::new(Field::new("values", DataType::UInt32, true));
536        let entry_struct = StructArray::from(vec![
537            (keys_field.clone(), make_array(key_data)),
538            (values_field.clone(), make_array(value_data.clone())),
539        ]);
540
541        // Construct a map array from the above two
542        let map_data_type = DataType::Map(
543            Arc::new(Field::new(
544                "entries",
545                entry_struct.data_type().clone(),
546                false,
547            )),
548            false,
549        );
550        let map_data = ArrayData::builder(map_data_type)
551            .len(3)
552            .add_buffer(entry_offsets)
553            .add_child_data(entry_struct.into_data())
554            .build()
555            .unwrap();
556        let map_array = MapArray::from(map_data);
557
558        assert_eq!(value_data, map_array.values().to_data());
559        assert_eq!(&DataType::UInt32, map_array.value_type());
560        assert_eq!(3, map_array.len());
561        assert_eq!(0, map_array.null_count());
562        assert_eq!(6, map_array.value_offsets()[2]);
563        assert_eq!(2, map_array.value_length(2));
564
565        let key_array = Arc::new(Int32Array::from(vec![0, 1, 2])) as ArrayRef;
566        let value_array =
567            Arc::new(UInt32Array::from(vec![None, Some(10u32), Some(20)])) as ArrayRef;
568        let struct_array = StructArray::from(vec![
569            (keys_field.clone(), key_array),
570            (values_field.clone(), value_array),
571        ]);
572        assert_eq!(
573            struct_array,
574            StructArray::from(map_array.value(0).into_data())
575        );
576        assert_eq!(
577            &struct_array,
578            unsafe { map_array.value_unchecked(0) }
579                .as_any()
580                .downcast_ref::<StructArray>()
581                .unwrap()
582        );
583        for i in 0..3 {
584            assert!(map_array.is_valid(i));
585            assert!(!map_array.is_null(i));
586        }
587
588        // Now test with a non-zero offset
589        let map_array = map_array.slice(1, 2);
590
591        assert_eq!(value_data, map_array.values().to_data());
592        assert_eq!(&DataType::UInt32, map_array.value_type());
593        assert_eq!(2, map_array.len());
594        assert_eq!(0, map_array.null_count());
595        assert_eq!(6, map_array.value_offsets()[1]);
596        assert_eq!(2, map_array.value_length(1));
597
598        let key_array = Arc::new(Int32Array::from(vec![3, 4, 5])) as ArrayRef;
599        let value_array = Arc::new(UInt32Array::from(vec![None, Some(40), None])) as ArrayRef;
600        let struct_array =
601            StructArray::from(vec![(keys_field, key_array), (values_field, value_array)]);
602        assert_eq!(
603            &struct_array,
604            map_array
605                .value(0)
606                .as_any()
607                .downcast_ref::<StructArray>()
608                .unwrap()
609        );
610        assert_eq!(
611            &struct_array,
612            unsafe { map_array.value_unchecked(0) }
613                .as_any()
614                .downcast_ref::<StructArray>()
615                .unwrap()
616        );
617    }
618
619    #[test]
620    #[ignore = "Test fails because slice of <list<struct>> is still buggy"]
621    fn test_map_array_slice() {
622        let map_array = create_from_buffers();
623
624        let sliced_array = map_array.slice(1, 2);
625        assert_eq!(2, sliced_array.len());
626        assert_eq!(1, sliced_array.offset());
627        let sliced_array_data = sliced_array.to_data();
628        for array_data in sliced_array_data.child_data() {
629            assert_eq!(array_data.offset(), 1);
630        }
631
632        // Check offset and length for each non-null value.
633        let sliced_map_array = sliced_array.as_any().downcast_ref::<MapArray>().unwrap();
634        assert_eq!(3, sliced_map_array.value_offsets()[0]);
635        assert_eq!(3, sliced_map_array.value_length(0));
636        assert_eq!(6, sliced_map_array.value_offsets()[1]);
637        assert_eq!(2, sliced_map_array.value_length(1));
638
639        // Construct key and values
640        let keys_data = ArrayData::builder(DataType::Int32)
641            .len(5)
642            .add_buffer(Buffer::from([3, 4, 5, 6, 7].to_byte_slice()))
643            .build()
644            .unwrap();
645        let values_data = ArrayData::builder(DataType::UInt32)
646            .len(5)
647            .add_buffer(Buffer::from([30u32, 40, 50, 60, 70].to_byte_slice()))
648            .build()
649            .unwrap();
650
651        // Construct a buffer for value offsets, for the nested array:
652        //  [[3, 4, 5], [6, 7]]
653        let entry_offsets = Buffer::from([0, 3, 5].to_byte_slice());
654
655        let keys = Arc::new(Field::new("keys", DataType::Int32, false));
656        let values = Arc::new(Field::new("values", DataType::UInt32, false));
657        let entry_struct = StructArray::from(vec![
658            (keys, make_array(keys_data)),
659            (values, make_array(values_data)),
660        ]);
661
662        // Construct a map array from the above two
663        let map_data_type = DataType::Map(
664            Arc::new(Field::new(
665                "entries",
666                entry_struct.data_type().clone(),
667                false,
668            )),
669            false,
670        );
671        let expected_map_data = ArrayData::builder(map_data_type)
672            .len(2)
673            .add_buffer(entry_offsets)
674            .add_child_data(entry_struct.into_data())
675            .build()
676            .unwrap();
677        let expected_map_array = MapArray::from(expected_map_data);
678
679        assert_eq!(&expected_map_array, sliced_map_array)
680    }
681
682    #[test]
683    #[should_panic(expected = "index out of bounds: the len is ")]
684    fn test_map_array_index_out_of_bound() {
685        let map_array = create_from_buffers();
686
687        map_array.value(map_array.len());
688    }
689
690    #[test]
691    #[should_panic(expected = "MapArray expected ArrayData with DataType::Map got Dictionary")]
692    fn test_from_array_data_validation() {
693        // A DictionaryArray has similar buffer layout to a MapArray
694        // but the meaning of the values differs
695        let struct_t = DataType::Struct(Fields::from(vec![
696            Field::new("keys", DataType::Int32, true),
697            Field::new("values", DataType::UInt32, true),
698        ]));
699        let dict_t = DataType::Dictionary(Box::new(DataType::Int32), Box::new(struct_t));
700        let _ = MapArray::from(ArrayData::new_empty(&dict_t));
701    }
702
703    #[test]
704    fn test_new_from_strings() {
705        let keys = vec!["a", "b", "c", "d", "e", "f", "g", "h"];
706        let values_data = UInt32Array::from(vec![0u32, 10, 20, 30, 40, 50, 60, 70]);
707
708        // Construct a buffer for value offsets, for the nested array:
709        //  [[a, b, c], [d, e, f], [g, h]]
710        let entry_offsets = [0, 3, 6, 8];
711
712        let map_array =
713            MapArray::new_from_strings(keys.clone().into_iter(), &values_data, &entry_offsets)
714                .unwrap();
715
716        assert_eq!(
717            &values_data,
718            map_array.values().as_primitive::<UInt32Type>()
719        );
720        assert_eq!(&DataType::UInt32, map_array.value_type());
721        assert_eq!(3, map_array.len());
722        assert_eq!(0, map_array.null_count());
723        assert_eq!(6, map_array.value_offsets()[2]);
724        assert_eq!(2, map_array.value_length(2));
725
726        let key_array = Arc::new(StringArray::from(vec!["a", "b", "c"])) as ArrayRef;
727        let value_array = Arc::new(UInt32Array::from(vec![0u32, 10, 20])) as ArrayRef;
728        let keys_field = Arc::new(Field::new("keys", DataType::Utf8, false));
729        let values_field = Arc::new(Field::new("values", DataType::UInt32, false));
730        let struct_array =
731            StructArray::from(vec![(keys_field, key_array), (values_field, value_array)]);
732        assert_eq!(
733            struct_array,
734            StructArray::from(map_array.value(0).into_data())
735        );
736        assert_eq!(
737            &struct_array,
738            unsafe { map_array.value_unchecked(0) }
739                .as_any()
740                .downcast_ref::<StructArray>()
741                .unwrap()
742        );
743        for i in 0..3 {
744            assert!(map_array.is_valid(i));
745            assert!(!map_array.is_null(i));
746        }
747    }
748
749    #[test]
750    fn test_try_new() {
751        let offsets = OffsetBuffer::new(vec![0, 1, 4, 5].into());
752        let fields = Fields::from(vec![
753            Field::new("key", DataType::Int32, false),
754            Field::new("values", DataType::Int32, false),
755        ]);
756        let columns = vec![
757            Arc::new(Int32Array::from(vec![1, 2, 3, 4, 5])) as _,
758            Arc::new(Int32Array::from(vec![1, 2, 3, 4, 5])) as _,
759        ];
760
761        let entries = StructArray::new(fields.clone(), columns, None);
762        let field = Arc::new(Field::new("entries", DataType::Struct(fields), false));
763
764        MapArray::new(field.clone(), offsets.clone(), entries.clone(), None, false);
765
766        let nulls = NullBuffer::new_null(3);
767        MapArray::new(field.clone(), offsets, entries.clone(), Some(nulls), false);
768
769        let nulls = NullBuffer::new_null(3);
770        let offsets = OffsetBuffer::new(vec![0, 1, 2, 4, 5].into());
771        let err = MapArray::try_new(
772            field.clone(),
773            offsets.clone(),
774            entries.clone(),
775            Some(nulls),
776            false,
777        )
778        .unwrap_err();
779
780        assert_eq!(
781            err.to_string(),
782            "Invalid argument error: Incorrect length of null buffer for MapArray, expected 4 got 3"
783        );
784
785        let err = MapArray::try_new(field, offsets.clone(), entries.slice(0, 2), None, false)
786            .unwrap_err();
787
788        assert_eq!(
789            err.to_string(),
790            "Invalid argument error: Max offset of 5 exceeds length of entries 2"
791        );
792
793        let field = Arc::new(Field::new("element", DataType::Int64, false));
794        let err = MapArray::try_new(field, offsets.clone(), entries, None, false)
795            .unwrap_err()
796            .to_string();
797
798        assert!(
799            err.starts_with("Invalid argument error: MapArray expected data type Int64 got Struct"),
800            "{err}"
801        );
802
803        let fields = Fields::from(vec![
804            Field::new("a", DataType::Int32, false),
805            Field::new("b", DataType::Int32, false),
806            Field::new("c", DataType::Int32, false),
807        ]);
808        let columns = vec![
809            Arc::new(Int32Array::from(vec![1, 2, 3, 4, 5])) as _,
810            Arc::new(Int32Array::from(vec![1, 2, 3, 4, 5])) as _,
811            Arc::new(Int32Array::from(vec![1, 2, 3, 4, 5])) as _,
812        ];
813
814        let s = StructArray::new(fields.clone(), columns, None);
815        let field = Arc::new(Field::new("entries", DataType::Struct(fields), false));
816        let err = MapArray::try_new(field, offsets, s, None, false).unwrap_err();
817
818        assert_eq!(
819            err.to_string(),
820            "Invalid argument error: MapArray entries must contain two children, got 3"
821        );
822    }
823}
arrow_array/array/map_array.rs

arrow_array/array/
map_array.rs
