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::{Array, ArrayAccessor, ArrayRef, ListArray, StringArray, StructArray, make_array};
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 a reference to the fields of the [`StructArray`] that backs this map.
177    pub fn entries_fields(&self) -> (&Field, &Field) {
178        let fields = self.entries.fields().iter().collect::<Vec<_>>();
179        let fields = TryInto::<[&FieldRef; 2]>::try_into(fields)
180            .expect("Every map has a key and value field");
181
182        (fields[0].as_ref(), fields[1].as_ref())
183    }
184
185    /// Returns the data type of the map's keys.
186    pub fn key_type(&self) -> &DataType {
187        self.keys().data_type()
188    }
189
190    /// Returns the data type of the map's values.
191    pub fn value_type(&self) -> &DataType {
192        self.values().data_type()
193    }
194
195    /// Returns ith value of this map array.
196    ///
197    /// Note: This method does not check for nulls and the value is arbitrary
198    /// if [`is_null`](Self::is_null) returns true for the index.
199    ///
200    /// # Safety
201    /// Caller must ensure that the index is within the array bounds
202    pub unsafe fn value_unchecked(&self, i: usize) -> StructArray {
203        let end = *unsafe { self.value_offsets().get_unchecked(i + 1) };
204        let start = *unsafe { self.value_offsets().get_unchecked(i) };
205        self.entries
206            .slice(start.to_usize().unwrap(), (end - start).to_usize().unwrap())
207    }
208
209    /// Returns ith value of this map array.
210    ///
211    /// This is a [`StructArray`] containing two fields
212    ///
213    /// Note: This method does not check for nulls and the value is arbitrary
214    /// (but still well-defined) if [`is_null`](Self::is_null) returns true for the index.
215    ///
216    /// # Panics
217    /// Panics if index `i` is out of bounds
218    pub fn value(&self, i: usize) -> StructArray {
219        let end = self.value_offsets()[i + 1] as usize;
220        let start = self.value_offsets()[i] as usize;
221        self.entries.slice(start, end - start)
222    }
223
224    /// Returns the offset values in the offsets buffer
225    #[inline]
226    pub fn value_offsets(&self) -> &[i32] {
227        &self.value_offsets
228    }
229
230    /// Returns the length for value at index `i`.
231    #[inline]
232    pub fn value_length(&self, i: usize) -> i32 {
233        let offsets = self.value_offsets();
234        offsets[i + 1] - offsets[i]
235    }
236
237    /// Returns a zero-copy slice of this array with the indicated offset and length.
238    pub fn slice(&self, offset: usize, length: usize) -> Self {
239        Self {
240            data_type: self.data_type.clone(),
241            nulls: self.nulls.as_ref().map(|n| n.slice(offset, length)),
242            entries: self.entries.clone(),
243            value_offsets: self.value_offsets.slice(offset, length),
244        }
245    }
246
247    /// constructs a new iterator
248    pub fn iter(&self) -> MapArrayIter<'_> {
249        MapArrayIter::new(self)
250    }
251}
252
253impl From<ArrayData> for MapArray {
254    fn from(data: ArrayData) -> Self {
255        Self::try_new_from_array_data(data)
256            .expect("Expected infallible creation of MapArray from ArrayData failed")
257    }
258}
259
260impl From<MapArray> for ArrayData {
261    fn from(array: MapArray) -> Self {
262        let len = array.len();
263        let builder = ArrayDataBuilder::new(array.data_type)
264            .len(len)
265            .nulls(array.nulls)
266            .buffers(vec![array.value_offsets.into_inner().into_inner()])
267            .child_data(vec![array.entries.to_data()]);
268
269        unsafe { builder.build_unchecked() }
270    }
271}
272
273impl MapArray {
274    fn try_new_from_array_data(data: ArrayData) -> Result<Self, ArrowError> {
275        if !matches!(data.data_type(), DataType::Map(_, _)) {
276            return Err(ArrowError::InvalidArgumentError(format!(
277                "MapArray expected ArrayData with DataType::Map got {}",
278                data.data_type()
279            )));
280        }
281
282        if data.buffers().len() != 1 {
283            return Err(ArrowError::InvalidArgumentError(format!(
284                "MapArray data should contain a single buffer only (value offsets), had {}",
285                data.len()
286            )));
287        }
288
289        if data.child_data().len() != 1 {
290            return Err(ArrowError::InvalidArgumentError(format!(
291                "MapArray should contain a single child array (values array), had {}",
292                data.child_data().len()
293            )));
294        }
295
296        let entries = data.child_data()[0].clone();
297
298        if let DataType::Struct(fields) = entries.data_type() {
299            if fields.len() != 2 {
300                return Err(ArrowError::InvalidArgumentError(format!(
301                    "MapArray should contain a struct array with 2 fields, have {} fields",
302                    fields.len()
303                )));
304            }
305        } else {
306            return Err(ArrowError::InvalidArgumentError(format!(
307                "MapArray should contain a struct array child, found {:?}",
308                entries.data_type()
309            )));
310        }
311        let entries = entries.into();
312
313        // SAFETY:
314        // ArrayData is valid, and verified type above
315        let value_offsets = unsafe { get_offsets(&data) };
316
317        Ok(Self {
318            data_type: data.data_type().clone(),
319            nulls: data.nulls().cloned(),
320            entries,
321            value_offsets,
322        })
323    }
324
325    /// Creates map array from provided keys, values and entry_offsets.
326    pub fn new_from_strings<'a>(
327        keys: impl Iterator<Item = &'a str>,
328        values: &dyn Array,
329        entry_offsets: &[u32],
330    ) -> Result<Self, ArrowError> {
331        let entry_offsets_buffer = Buffer::from(entry_offsets.to_byte_slice());
332        let keys_data = StringArray::from_iter_values(keys);
333
334        let keys_field = Arc::new(Field::new("keys", DataType::Utf8, false));
335        let values_field = Arc::new(Field::new(
336            "values",
337            values.data_type().clone(),
338            values.null_count() > 0,
339        ));
340
341        let entry_struct = StructArray::from(vec![
342            (keys_field, Arc::new(keys_data) as ArrayRef),
343            (values_field, make_array(values.to_data())),
344        ]);
345
346        let map_data_type = DataType::Map(
347            Arc::new(Field::new(
348                "entries",
349                entry_struct.data_type().clone(),
350                false,
351            )),
352            false,
353        );
354        let map_data = ArrayData::builder(map_data_type)
355            .len(entry_offsets.len() - 1)
356            .add_buffer(entry_offsets_buffer)
357            .add_child_data(entry_struct.into_data())
358            .build()?;
359
360        Ok(MapArray::from(map_data))
361    }
362}
363
364impl super::private::Sealed for MapArray {}
365
366impl Array for MapArray {
367    fn as_any(&self) -> &dyn Any {
368        self
369    }
370
371    fn to_data(&self) -> ArrayData {
372        self.clone().into_data()
373    }
374
375    fn into_data(self) -> ArrayData {
376        self.into()
377    }
378
379    fn data_type(&self) -> &DataType {
380        &self.data_type
381    }
382
383    fn slice(&self, offset: usize, length: usize) -> ArrayRef {
384        Arc::new(self.slice(offset, length))
385    }
386
387    fn len(&self) -> usize {
388        self.value_offsets.len() - 1
389    }
390
391    fn is_empty(&self) -> bool {
392        self.value_offsets.len() <= 1
393    }
394
395    fn shrink_to_fit(&mut self) {
396        if let Some(nulls) = &mut self.nulls {
397            nulls.shrink_to_fit();
398        }
399        self.entries.shrink_to_fit();
400        self.value_offsets.shrink_to_fit();
401    }
402
403    fn offset(&self) -> usize {
404        0
405    }
406
407    fn nulls(&self) -> Option<&NullBuffer> {
408        self.nulls.as_ref()
409    }
410
411    fn logical_null_count(&self) -> usize {
412        // More efficient that the default implementation
413        self.null_count()
414    }
415
416    fn get_buffer_memory_size(&self) -> usize {
417        let mut size = self.entries.get_buffer_memory_size();
418        size += self.value_offsets.inner().inner().capacity();
419        if let Some(n) = self.nulls.as_ref() {
420            size += n.buffer().capacity();
421        }
422        size
423    }
424
425    fn get_array_memory_size(&self) -> usize {
426        let mut size = std::mem::size_of::<Self>() + self.entries.get_array_memory_size();
427        size += self.value_offsets.inner().inner().capacity();
428        if let Some(n) = self.nulls.as_ref() {
429            size += n.buffer().capacity();
430        }
431        size
432    }
433}
434
435impl ArrayAccessor for &MapArray {
436    type Item = StructArray;
437
438    fn value(&self, index: usize) -> Self::Item {
439        MapArray::value(self, index)
440    }
441
442    unsafe fn value_unchecked(&self, index: usize) -> Self::Item {
443        MapArray::value(self, index)
444    }
445}
446
447impl std::fmt::Debug for MapArray {
448    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
449        write!(f, "MapArray\n[\n")?;
450        print_long_array(self, f, |array, index, f| {
451            std::fmt::Debug::fmt(&array.value(index), f)
452        })?;
453        write!(f, "]")
454    }
455}
456
457impl From<MapArray> for ListArray {
458    fn from(value: MapArray) -> Self {
459        let field = match value.data_type() {
460            DataType::Map(field, _) => field,
461            _ => unreachable!("This should be a map type."),
462        };
463        let data_type = DataType::List(field.clone());
464        let builder = value.into_data().into_builder().data_type(data_type);
465        let array_data = unsafe { builder.build_unchecked() };
466
467        ListArray::from(array_data)
468    }
469}
470
471#[cfg(test)]
472mod tests {
473    use crate::cast::AsArray;
474    use crate::types::UInt32Type;
475    use crate::{Int32Array, UInt32Array};
476    use arrow_schema::Fields;
477
478    use super::*;
479
480    fn create_from_buffers() -> MapArray {
481        // Construct key and values
482        let keys_data = ArrayData::builder(DataType::Int32)
483            .len(8)
484            .add_buffer(Buffer::from([0, 1, 2, 3, 4, 5, 6, 7].to_byte_slice()))
485            .build()
486            .unwrap();
487        let values_data = ArrayData::builder(DataType::UInt32)
488            .len(8)
489            .add_buffer(Buffer::from(
490                [0u32, 10, 20, 30, 40, 50, 60, 70].to_byte_slice(),
491            ))
492            .build()
493            .unwrap();
494
495        // Construct a buffer for value offsets, for the nested array:
496        //  [[0, 1, 2], [3, 4, 5], [6, 7]]
497        let entry_offsets = Buffer::from([0, 3, 6, 8].to_byte_slice());
498
499        let keys = Arc::new(Field::new("keys", DataType::Int32, false));
500        let values = Arc::new(Field::new("values", DataType::UInt32, false));
501        let entry_struct = StructArray::from(vec![
502            (keys, make_array(keys_data)),
503            (values, make_array(values_data)),
504        ]);
505
506        // Construct a map array from the above two
507        let map_data_type = DataType::Map(
508            Arc::new(Field::new(
509                "entries",
510                entry_struct.data_type().clone(),
511                false,
512            )),
513            false,
514        );
515        let map_data = ArrayData::builder(map_data_type)
516            .len(3)
517            .add_buffer(entry_offsets)
518            .add_child_data(entry_struct.into_data())
519            .build()
520            .unwrap();
521        MapArray::from(map_data)
522    }
523
524    #[test]
525    fn test_map_array() {
526        // Construct key and values
527        let key_data = ArrayData::builder(DataType::Int32)
528            .len(8)
529            .add_buffer(Buffer::from([0, 1, 2, 3, 4, 5, 6, 7].to_byte_slice()))
530            .build()
531            .unwrap();
532        let value_data = ArrayData::builder(DataType::UInt32)
533            .len(8)
534            .add_buffer(Buffer::from(
535                [0u32, 10, 20, 0, 40, 0, 60, 70].to_byte_slice(),
536            ))
537            .null_bit_buffer(Some(Buffer::from(&[0b11010110])))
538            .build()
539            .unwrap();
540
541        // Construct a buffer for value offsets, for the nested array:
542        //  [[0, 1, 2], [3, 4, 5], [6, 7]]
543        let entry_offsets = Buffer::from([0, 3, 6, 8].to_byte_slice());
544
545        let keys_field = Arc::new(Field::new("keys", DataType::Int32, false));
546        let values_field = Arc::new(Field::new("values", DataType::UInt32, true));
547        let entry_struct = StructArray::from(vec![
548            (keys_field.clone(), make_array(key_data)),
549            (values_field.clone(), make_array(value_data.clone())),
550        ]);
551
552        // Construct a map array from the above two
553        let map_data_type = DataType::Map(
554            Arc::new(Field::new(
555                "entries",
556                entry_struct.data_type().clone(),
557                false,
558            )),
559            false,
560        );
561        let map_data = ArrayData::builder(map_data_type)
562            .len(3)
563            .add_buffer(entry_offsets)
564            .add_child_data(entry_struct.into_data())
565            .build()
566            .unwrap();
567        let map_array = MapArray::from(map_data);
568
569        assert_eq!(value_data, map_array.values().to_data());
570        assert_eq!(&DataType::UInt32, map_array.value_type());
571        assert_eq!(3, map_array.len());
572        assert_eq!(0, map_array.null_count());
573        assert_eq!(6, map_array.value_offsets()[2]);
574        assert_eq!(2, map_array.value_length(2));
575
576        let key_array = Arc::new(Int32Array::from(vec![0, 1, 2])) as ArrayRef;
577        let value_array =
578            Arc::new(UInt32Array::from(vec![None, Some(10u32), Some(20)])) as ArrayRef;
579        let struct_array = StructArray::from(vec![
580            (keys_field.clone(), key_array),
581            (values_field.clone(), value_array),
582        ]);
583        assert_eq!(
584            struct_array,
585            StructArray::from(map_array.value(0).into_data())
586        );
587        assert_eq!(
588            &struct_array,
589            unsafe { map_array.value_unchecked(0) }
590                .as_any()
591                .downcast_ref::<StructArray>()
592                .unwrap()
593        );
594        for i in 0..3 {
595            assert!(map_array.is_valid(i));
596            assert!(!map_array.is_null(i));
597        }
598
599        // Now test with a non-zero offset
600        let map_array = map_array.slice(1, 2);
601
602        assert_eq!(value_data, map_array.values().to_data());
603        assert_eq!(&DataType::UInt32, map_array.value_type());
604        assert_eq!(2, map_array.len());
605        assert_eq!(0, map_array.null_count());
606        assert_eq!(6, map_array.value_offsets()[1]);
607        assert_eq!(2, map_array.value_length(1));
608
609        let key_array = Arc::new(Int32Array::from(vec![3, 4, 5])) as ArrayRef;
610        let value_array = Arc::new(UInt32Array::from(vec![None, Some(40), None])) as ArrayRef;
611        let struct_array =
612            StructArray::from(vec![(keys_field, key_array), (values_field, value_array)]);
613        assert_eq!(
614            &struct_array,
615            map_array
616                .value(0)
617                .as_any()
618                .downcast_ref::<StructArray>()
619                .unwrap()
620        );
621        assert_eq!(
622            &struct_array,
623            unsafe { map_array.value_unchecked(0) }
624                .as_any()
625                .downcast_ref::<StructArray>()
626                .unwrap()
627        );
628    }
629
630    #[test]
631    #[ignore = "Test fails because slice of <list<struct>> is still buggy"]
632    fn test_map_array_slice() {
633        let map_array = create_from_buffers();
634
635        let sliced_array = map_array.slice(1, 2);
636        assert_eq!(2, sliced_array.len());
637        assert_eq!(1, sliced_array.offset());
638        let sliced_array_data = sliced_array.to_data();
639        for array_data in sliced_array_data.child_data() {
640            assert_eq!(array_data.offset(), 1);
641        }
642
643        // Check offset and length for each non-null value.
644        let sliced_map_array = sliced_array.as_any().downcast_ref::<MapArray>().unwrap();
645        assert_eq!(3, sliced_map_array.value_offsets()[0]);
646        assert_eq!(3, sliced_map_array.value_length(0));
647        assert_eq!(6, sliced_map_array.value_offsets()[1]);
648        assert_eq!(2, sliced_map_array.value_length(1));
649
650        // Construct key and values
651        let keys_data = ArrayData::builder(DataType::Int32)
652            .len(5)
653            .add_buffer(Buffer::from([3, 4, 5, 6, 7].to_byte_slice()))
654            .build()
655            .unwrap();
656        let values_data = ArrayData::builder(DataType::UInt32)
657            .len(5)
658            .add_buffer(Buffer::from([30u32, 40, 50, 60, 70].to_byte_slice()))
659            .build()
660            .unwrap();
661
662        // Construct a buffer for value offsets, for the nested array:
663        //  [[3, 4, 5], [6, 7]]
664        let entry_offsets = Buffer::from([0, 3, 5].to_byte_slice());
665
666        let keys = Arc::new(Field::new("keys", DataType::Int32, false));
667        let values = Arc::new(Field::new("values", DataType::UInt32, false));
668        let entry_struct = StructArray::from(vec![
669            (keys, make_array(keys_data)),
670            (values, make_array(values_data)),
671        ]);
672
673        // Construct a map array from the above two
674        let map_data_type = DataType::Map(
675            Arc::new(Field::new(
676                "entries",
677                entry_struct.data_type().clone(),
678                false,
679            )),
680            false,
681        );
682        let expected_map_data = ArrayData::builder(map_data_type)
683            .len(2)
684            .add_buffer(entry_offsets)
685            .add_child_data(entry_struct.into_data())
686            .build()
687            .unwrap();
688        let expected_map_array = MapArray::from(expected_map_data);
689
690        assert_eq!(&expected_map_array, sliced_map_array)
691    }
692
693    #[test]
694    #[should_panic(expected = "index out of bounds: the len is ")]
695    fn test_map_array_index_out_of_bound() {
696        let map_array = create_from_buffers();
697
698        map_array.value(map_array.len());
699    }
700
701    #[test]
702    #[should_panic(expected = "MapArray expected ArrayData with DataType::Map got Dictionary")]
703    fn test_from_array_data_validation() {
704        // A DictionaryArray has similar buffer layout to a MapArray
705        // but the meaning of the values differs
706        let struct_t = DataType::Struct(Fields::from(vec![
707            Field::new("keys", DataType::Int32, true),
708            Field::new("values", DataType::UInt32, true),
709        ]));
710        let dict_t = DataType::Dictionary(Box::new(DataType::Int32), Box::new(struct_t));
711        let _ = MapArray::from(ArrayData::new_empty(&dict_t));
712    }
713
714    #[test]
715    fn test_new_from_strings() {
716        let keys = vec!["a", "b", "c", "d", "e", "f", "g", "h"];
717        let values_data = UInt32Array::from(vec![0u32, 10, 20, 30, 40, 50, 60, 70]);
718
719        // Construct a buffer for value offsets, for the nested array:
720        //  [[a, b, c], [d, e, f], [g, h]]
721        let entry_offsets = [0, 3, 6, 8];
722
723        let map_array =
724            MapArray::new_from_strings(keys.clone().into_iter(), &values_data, &entry_offsets)
725                .unwrap();
726
727        assert_eq!(
728            &values_data,
729            map_array.values().as_primitive::<UInt32Type>()
730        );
731        assert_eq!(&DataType::UInt32, map_array.value_type());
732        assert_eq!(3, map_array.len());
733        assert_eq!(0, map_array.null_count());
734        assert_eq!(6, map_array.value_offsets()[2]);
735        assert_eq!(2, map_array.value_length(2));
736
737        let key_array = Arc::new(StringArray::from(vec!["a", "b", "c"])) as ArrayRef;
738        let value_array = Arc::new(UInt32Array::from(vec![0u32, 10, 20])) as ArrayRef;
739        let keys_field = Arc::new(Field::new("keys", DataType::Utf8, false));
740        let values_field = Arc::new(Field::new("values", DataType::UInt32, false));
741        let struct_array =
742            StructArray::from(vec![(keys_field, key_array), (values_field, value_array)]);
743        assert_eq!(
744            struct_array,
745            StructArray::from(map_array.value(0).into_data())
746        );
747        assert_eq!(
748            &struct_array,
749            unsafe { map_array.value_unchecked(0) }
750                .as_any()
751                .downcast_ref::<StructArray>()
752                .unwrap()
753        );
754        for i in 0..3 {
755            assert!(map_array.is_valid(i));
756            assert!(!map_array.is_null(i));
757        }
758    }
759
760    #[test]
761    fn test_try_new() {
762        let offsets = OffsetBuffer::new(vec![0, 1, 4, 5].into());
763        let fields = Fields::from(vec![
764            Field::new("key", DataType::Int32, false),
765            Field::new("values", DataType::Int32, false),
766        ]);
767        let columns = vec![
768            Arc::new(Int32Array::from(vec![1, 2, 3, 4, 5])) as _,
769            Arc::new(Int32Array::from(vec![1, 2, 3, 4, 5])) as _,
770        ];
771
772        let entries = StructArray::new(fields.clone(), columns, None);
773        let field = Arc::new(Field::new("entries", DataType::Struct(fields), false));
774
775        MapArray::new(field.clone(), offsets.clone(), entries.clone(), None, false);
776
777        let nulls = NullBuffer::new_null(3);
778        MapArray::new(field.clone(), offsets, entries.clone(), Some(nulls), false);
779
780        let nulls = NullBuffer::new_null(3);
781        let offsets = OffsetBuffer::new(vec![0, 1, 2, 4, 5].into());
782        let err = MapArray::try_new(
783            field.clone(),
784            offsets.clone(),
785            entries.clone(),
786            Some(nulls),
787            false,
788        )
789        .unwrap_err();
790
791        assert_eq!(
792            err.to_string(),
793            "Invalid argument error: Incorrect length of null buffer for MapArray, expected 4 got 3"
794        );
795
796        let err = MapArray::try_new(field, offsets.clone(), entries.slice(0, 2), None, false)
797            .unwrap_err();
798
799        assert_eq!(
800            err.to_string(),
801            "Invalid argument error: Max offset of 5 exceeds length of entries 2"
802        );
803
804        let field = Arc::new(Field::new("element", DataType::Int64, false));
805        let err = MapArray::try_new(field, offsets.clone(), entries, None, false)
806            .unwrap_err()
807            .to_string();
808
809        assert!(
810            err.starts_with("Invalid argument error: MapArray expected data type Int64 got Struct"),
811            "{err}"
812        );
813
814        let fields = Fields::from(vec![
815            Field::new("a", DataType::Int32, false),
816            Field::new("b", DataType::Int32, false),
817            Field::new("c", DataType::Int32, false),
818        ]);
819        let columns = vec![
820            Arc::new(Int32Array::from(vec![1, 2, 3, 4, 5])) as _,
821            Arc::new(Int32Array::from(vec![1, 2, 3, 4, 5])) as _,
822            Arc::new(Int32Array::from(vec![1, 2, 3, 4, 5])) as _,
823        ];
824
825        let s = StructArray::new(fields.clone(), columns, None);
826        let field = Arc::new(Field::new("entries", DataType::Struct(fields), false));
827        let err = MapArray::try_new(field, offsets, s, None, false).unwrap_err();
828
829        assert_eq!(
830            err.to_string(),
831            "Invalid argument error: MapArray entries must contain two children, got 3"
832        );
833    }
834}
arrow_array/array/map_array.rs

arrow_array/array/
map_array.rs
