1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
//! Encodings that enable zero-copy sharing of data with Arrow.

use std::sync::Arc;

use arrow_array::types::*;
use arrow_array::{
    Array, ArrayRef, ArrowPrimitiveType, BooleanArray as ArrowBoolArray, Date32Array, Date64Array,
    NullArray as ArrowNullArray, PrimitiveArray as ArrowPrimitiveArray,
    StructArray as ArrowStructArray, Time32MillisecondArray, Time32SecondArray,
    Time64MicrosecondArray, Time64NanosecondArray, TimestampMicrosecondArray,
    TimestampMillisecondArray, TimestampNanosecondArray, TimestampSecondArray,
};
use arrow_buffer::ScalarBuffer;
use arrow_schema::{Field, FieldRef, Fields};
use vortex_datetime_dtype::{is_temporal_ext_type, TemporalMetadata, TimeUnit};
use vortex_dtype::{DType, NativePType, PType};
use vortex_error::{vortex_bail, VortexError, VortexResult};

use crate::array::{
    varbinview_as_arrow, BoolArray, ExtensionArray, ListArray, NullArray, PrimitiveArray,
    StructArray, TemporalArray, VarBinViewArray,
};
use crate::arrow::wrappers::as_offset_buffer;
use crate::arrow::{infer_data_type, FromArrowArray};
use crate::compute::try_cast;
use crate::encoding::Encoding;
use crate::stats::ArrayStatistics;
use crate::validity::ArrayValidity;
use crate::variants::{PrimitiveArrayTrait, StructArrayTrait};
use crate::{ArrayDType, ArrayData, ArrayLen, IntoArrayData, ToArrayData};

/// The set of canonical array encodings, also the set of encodings that can be transferred to
/// Arrow with zero-copy.
///
/// Note that a canonical form is not recursive, i.e. a StructArray may contain non-canonical
/// child arrays, which may themselves need to be [canonicalized](IntoCanonical).
///
/// # Logical vs. Physical encodings
///
/// Vortex separates logical and physical types, however this creates ambiguity with Arrow, there is
/// no separation. Thus, if you receive an Arrow array, compress it using Vortex, and then
/// decompress it later to pass to a compute kernel, there are multiple suitable Arrow array
/// variants to hold the data.
///
/// To disambiguate, we choose a canonical physical encoding for every Vortex [`DType`], which
/// will correspond to an arrow-rs [`arrow_schema::DataType`].
///
/// # Views support
///
/// Binary and String views, also known as "German strings" are a better encoding format for
/// nearly all use-cases. Variable-length binary views are part of the Apache Arrow spec, and are
/// fully supported by the Datafusion query engine. We use them as our canonical string encoding
/// for all `Utf8` and `Binary` typed arrays in Vortex.
///
#[derive(Debug, Clone)]
pub enum Canonical {
    Null(NullArray),
    Bool(BoolArray),
    Primitive(PrimitiveArray),
    Struct(StructArray),
    // TODO(joe): maybe this should be a ListView, however this will be annoying in spiral
    List(ListArray),
    VarBinView(VarBinViewArray),
    Extension(ExtensionArray),
}

impl Canonical {
    /// Convert a canonical array into its equivalent [ArrayRef](Arrow array).
    ///
    /// Scalar arrays such as Bool and Primitive canonical arrays should convert with
    /// zero copies, while more complex variants such as Struct may require allocations if its child
    /// arrays require decompression.
    pub fn into_arrow(self) -> VortexResult<ArrayRef> {
        Ok(match self {
            Canonical::Null(a) => null_to_arrow(a)?,
            Canonical::Bool(a) => bool_to_arrow(a)?,
            Canonical::Primitive(a) => primitive_to_arrow(a)?,
            Canonical::Struct(a) => struct_to_arrow(a)?,
            Canonical::List(a) => list_to_arrow(a)?,
            Canonical::VarBinView(a) => varbinview_as_arrow(&a),
            Canonical::Extension(a) => {
                if is_temporal_ext_type(a.id()) {
                    temporal_to_arrow(TemporalArray::try_from(a.into_array())?)?
                } else {
                    // Convert storage array directly into arrow, losing type information
                    // that will let us round-trip.
                    // TODO(aduffy): https://github.com/spiraldb/vortex/issues/1167
                    a.storage().into_arrow()?
                }
            }
        })
    }
}

impl Canonical {
    // Create an empty canonical array of the given dtype.
    pub fn empty(dtype: &DType) -> VortexResult<Canonical> {
        let arrow_dtype = infer_data_type(dtype)?;
        ArrayData::from_arrow(
            arrow_array::new_empty_array(&arrow_dtype),
            dtype.is_nullable(),
        )
        .into_canonical()
    }
}

// Unwrap canonical type back down to specialized type.
impl Canonical {
    pub fn into_null(self) -> VortexResult<NullArray> {
        match self {
            Canonical::Null(a) => Ok(a),
            _ => vortex_bail!("Cannot unwrap NullArray from {:?}", &self),
        }
    }

    pub fn into_bool(self) -> VortexResult<BoolArray> {
        match self {
            Canonical::Bool(a) => Ok(a),
            _ => vortex_bail!("Cannot unwrap BoolArray from {:?}", &self),
        }
    }

    pub fn into_primitive(self) -> VortexResult<PrimitiveArray> {
        match self {
            Canonical::Primitive(a) => Ok(a),
            _ => vortex_bail!("Cannot unwrap PrimitiveArray from {:?}", &self),
        }
    }

    pub fn into_struct(self) -> VortexResult<StructArray> {
        match self {
            Canonical::Struct(a) => Ok(a),
            _ => vortex_bail!("Cannot unwrap StructArray from {:?}", &self),
        }
    }

    pub fn into_list(self) -> VortexResult<ListArray> {
        match self {
            Canonical::List(a) => Ok(a),
            _ => vortex_bail!("Cannot unwrap StructArray from {:?}", &self),
        }
    }

    pub fn into_varbinview(self) -> VortexResult<VarBinViewArray> {
        match self {
            Canonical::VarBinView(a) => Ok(a),
            _ => vortex_bail!("Cannot unwrap VarBinViewArray from {:?}", &self),
        }
    }

    pub fn into_extension(self) -> VortexResult<ExtensionArray> {
        match self {
            Canonical::Extension(a) => Ok(a),
            _ => vortex_bail!("Cannot unwrap ExtensionArray from {:?}", &self),
        }
    }
}

fn null_to_arrow(null_array: NullArray) -> VortexResult<ArrayRef> {
    Ok(Arc::new(ArrowNullArray::new(null_array.len())))
}

fn bool_to_arrow(bool_array: BoolArray) -> VortexResult<ArrayRef> {
    Ok(Arc::new(ArrowBoolArray::new(
        bool_array.boolean_buffer(),
        bool_array.logical_validity().to_null_buffer()?,
    )))
}

fn primitive_to_arrow(primitive_array: PrimitiveArray) -> VortexResult<ArrayRef> {
    fn as_arrow_array_primitive<T: ArrowPrimitiveType>(
        array: &PrimitiveArray,
    ) -> VortexResult<Arc<ArrowPrimitiveArray<T>>> {
        Ok(Arc::new(ArrowPrimitiveArray::new(
            ScalarBuffer::<T::Native>::new(array.buffer().clone().into_arrow(), 0, array.len()),
            array.logical_validity().to_null_buffer()?,
        )))
    }

    Ok(match primitive_array.ptype() {
        PType::U8 => as_arrow_array_primitive::<UInt8Type>(&primitive_array)?,
        PType::U16 => as_arrow_array_primitive::<UInt16Type>(&primitive_array)?,
        PType::U32 => as_arrow_array_primitive::<UInt32Type>(&primitive_array)?,
        PType::U64 => as_arrow_array_primitive::<UInt64Type>(&primitive_array)?,
        PType::I8 => as_arrow_array_primitive::<Int8Type>(&primitive_array)?,
        PType::I16 => as_arrow_array_primitive::<Int16Type>(&primitive_array)?,
        PType::I32 => as_arrow_array_primitive::<Int32Type>(&primitive_array)?,
        PType::I64 => as_arrow_array_primitive::<Int64Type>(&primitive_array)?,
        PType::F16 => as_arrow_array_primitive::<Float16Type>(&primitive_array)?,
        PType::F32 => as_arrow_array_primitive::<Float32Type>(&primitive_array)?,
        PType::F64 => as_arrow_array_primitive::<Float64Type>(&primitive_array)?,
    })
}

fn struct_to_arrow(struct_array: StructArray) -> VortexResult<ArrayRef> {
    let field_arrays = struct_array
        .names()
        .iter()
        .zip(struct_array.children())
        .map(|(name, f)| {
            f.into_canonical()
                .map_err(|err| err.with_context(format!("Failed to canonicalize field {}", name)))
                .and_then(|c| c.into_arrow())
        })
        .collect::<VortexResult<Vec<_>>>()?;

    let nulls = struct_array.logical_validity().to_null_buffer()?;

    if field_arrays.is_empty() {
        Ok(Arc::new(ArrowStructArray::new_empty_fields(
            struct_array.len(),
            nulls,
        )))
    } else {
        let arrow_fields = struct_array
            .names()
            .iter()
            .zip(field_arrays.iter())
            .zip(struct_array.dtypes().iter())
            .map(|((name, arrow_field), vortex_field)| {
                Field::new(
                    &**name,
                    arrow_field.data_type().clone(),
                    vortex_field.is_nullable(),
                )
            })
            .map(Arc::new)
            .collect::<Fields>();

        Ok(Arc::new(ArrowStructArray::try_new(
            arrow_fields,
            field_arrays,
            nulls,
        )?))
    }
}

// TODO(joe): unify with varbin
fn list_to_arrow(list: ListArray) -> VortexResult<ArrayRef> {
    let offsets = list
        .offsets()
        .into_primitive()
        .map_err(|err| err.with_context("Failed to canonicalize offsets"))?;

    let offsets = match offsets.ptype() {
        PType::I32 | PType::I64 => offsets,
        PType::U64 => try_cast(offsets, PType::I64.into())?.into_primitive()?,
        PType::U32 => try_cast(offsets, PType::I32.into())?.into_primitive()?,

        // Unless it's u64, everything else can be converted into an i32.
        _ => try_cast(offsets.to_array(), PType::I32.into())
            .and_then(|a| a.into_primitive())
            .map_err(|err| err.with_context("Failed to cast offsets to PrimitiveArray of i32"))?,
    };

    let values = list.elements().into_arrow()?;

    let field_ref = FieldRef::new(Field::new_list_field(
        values.data_type().clone(),
        list.validity().nullability().into(),
    ));

    let nulls = list.logical_validity().to_null_buffer()?;

    Ok(match offsets.ptype() {
        PType::I32 => Arc::new(arrow_array::ListArray::try_new(
            field_ref,
            as_offset_buffer::<i32>(list.offsets().into_primitive()?),
            values,
            nulls,
        )?),
        PType::I64 => Arc::new(arrow_array::LargeListArray::try_new(
            field_ref,
            as_offset_buffer::<i64>(list.offsets().into_primitive()?),
            values,
            nulls,
        )?),
        _ => vortex_bail!("Invalid offsets type {}", offsets.ptype()),
    })
}

fn temporal_to_arrow(temporal_array: TemporalArray) -> VortexResult<ArrayRef> {
    macro_rules! extract_temporal_values {
        ($values:expr, $prim:ty) => {{
            let temporal_values = try_cast(
                $values,
                &DType::Primitive(<$prim as NativePType>::PTYPE, $values.dtype().nullability()),
            )?
            .into_primitive()?;
            let len = temporal_values.len();
            let nulls = temporal_values.logical_validity().to_null_buffer()?;
            let scalars =
                ScalarBuffer::<$prim>::new(temporal_values.into_buffer().into_arrow(), 0, len);

            (scalars, nulls)
        }};
    }

    Ok(match temporal_array.temporal_metadata() {
        TemporalMetadata::Date(time_unit) => match time_unit {
            TimeUnit::D => {
                let (scalars, nulls) =
                    extract_temporal_values!(&temporal_array.temporal_values(), i32);
                Arc::new(Date32Array::new(scalars, nulls))
            }
            TimeUnit::Ms => {
                let (scalars, nulls) =
                    extract_temporal_values!(&temporal_array.temporal_values(), i64);
                Arc::new(Date64Array::new(scalars, nulls))
            }
            _ => vortex_bail!(
                "Invalid TimeUnit {time_unit} for {}",
                temporal_array.ext_dtype().id()
            ),
        },
        TemporalMetadata::Time(time_unit) => match time_unit {
            TimeUnit::S => {
                let (scalars, nulls) =
                    extract_temporal_values!(&temporal_array.temporal_values(), i32);
                Arc::new(Time32SecondArray::new(scalars, nulls))
            }
            TimeUnit::Ms => {
                let (scalars, nulls) =
                    extract_temporal_values!(&temporal_array.temporal_values(), i32);
                Arc::new(Time32MillisecondArray::new(scalars, nulls))
            }
            TimeUnit::Us => {
                let (scalars, nulls) =
                    extract_temporal_values!(&temporal_array.temporal_values(), i64);
                Arc::new(Time64MicrosecondArray::new(scalars, nulls))
            }
            TimeUnit::Ns => {
                let (scalars, nulls) =
                    extract_temporal_values!(&temporal_array.temporal_values(), i64);
                Arc::new(Time64NanosecondArray::new(scalars, nulls))
            }
            _ => vortex_bail!(
                "Invalid TimeUnit {time_unit} for {}",
                temporal_array.ext_dtype().id()
            ),
        },
        TemporalMetadata::Timestamp(time_unit, _) => {
            let (scalars, nulls) = extract_temporal_values!(&temporal_array.temporal_values(), i64);
            match time_unit {
                TimeUnit::Ns => Arc::new(TimestampNanosecondArray::new(scalars, nulls)),
                TimeUnit::Us => Arc::new(TimestampMicrosecondArray::new(scalars, nulls)),
                TimeUnit::Ms => Arc::new(TimestampMillisecondArray::new(scalars, nulls)),
                TimeUnit::S => Arc::new(TimestampSecondArray::new(scalars, nulls)),
                _ => vortex_bail!(
                    "Invalid TimeUnit {time_unit} for {}",
                    temporal_array.ext_dtype().id()
                ),
            }
        }
    })
}

/// Support trait for transmuting an array into the canonical encoding for its [vortex_dtype::DType].
///
/// This conversion ensures that the array's encoding matches one of the builtin canonical
/// encodings, each of which has a corresponding [Canonical] variant.
///
/// # Invariants
///
/// The DType of the array will be unchanged by canonicalization.
pub trait IntoCanonical {
    fn into_canonical(self) -> VortexResult<Canonical>;

    fn into_arrow(self) -> VortexResult<ArrayRef>
    where
        Self: Sized,
    {
        self.into_canonical()?.into_arrow()
    }
}

/// Encoding VTable for canonicalizing an array.
#[allow(clippy::wrong_self_convention)]
pub trait IntoCanonicalVTable {
    fn into_canonical(&self, array: ArrayData) -> VortexResult<Canonical>;

    fn into_arrow(&self, array: ArrayData) -> VortexResult<ArrayRef>;
}

/// Implement the [IntoCanonicalVTable] for all encodings with arrays implementing [IntoCanonical].
impl<E: Encoding> IntoCanonicalVTable for E
where
    E::Array: IntoCanonical,
    E::Array: TryFrom<ArrayData, Error = VortexError>,
{
    fn into_canonical(&self, data: ArrayData) -> VortexResult<Canonical> {
        #[cfg(feature = "canonical_counter")]
        data.inc_canonical_counter();
        let canonical = E::Array::try_from(data.clone())?.into_canonical()?;
        canonical.inherit_statistics(data.statistics());
        Ok(canonical)
    }

    fn into_arrow(&self, array: ArrayData) -> VortexResult<ArrayRef> {
        E::Array::try_from(array)?.into_arrow()
    }
}

/// Trait for types that can be converted from an owned type into an owned array variant.
///
/// # Canonicalization
///
/// This trait has a blanket implementation for all types implementing [IntoCanonical].
pub trait IntoArrayVariant {
    fn into_null(self) -> VortexResult<NullArray>;

    fn into_bool(self) -> VortexResult<BoolArray>;

    fn into_primitive(self) -> VortexResult<PrimitiveArray>;

    fn into_struct(self) -> VortexResult<StructArray>;

    fn into_list(self) -> VortexResult<ListArray>;

    fn into_varbinview(self) -> VortexResult<VarBinViewArray>;

    fn into_extension(self) -> VortexResult<ExtensionArray>;
}

impl<T> IntoArrayVariant for T
where
    T: IntoCanonical,
{
    fn into_null(self) -> VortexResult<NullArray> {
        self.into_canonical()?.into_null()
    }

    fn into_bool(self) -> VortexResult<BoolArray> {
        self.into_canonical()?.into_bool()
    }

    fn into_primitive(self) -> VortexResult<PrimitiveArray> {
        self.into_canonical()?.into_primitive()
    }

    fn into_struct(self) -> VortexResult<StructArray> {
        self.into_canonical()?.into_struct()
    }

    fn into_list(self) -> VortexResult<ListArray> {
        self.into_canonical()?.into_list()
    }

    fn into_varbinview(self) -> VortexResult<VarBinViewArray> {
        self.into_canonical()?.into_varbinview()
    }

    fn into_extension(self) -> VortexResult<ExtensionArray> {
        self.into_canonical()?.into_extension()
    }
}

/// IntoCanonical implementation for Array.
///
/// Canonicalizing an array requires potentially decompressing, so this requires a roundtrip through
/// the array's internal codec.
impl IntoCanonical for ArrayData {
    fn into_canonical(self) -> VortexResult<Canonical> {
        // We only care to know when we canonicalize something non-trivial.
        if !self.is_canonical() && self.len() > 1 {
            log::trace!("Canonicalizing array with encoding {:?}", self.encoding());
        }
        self.encoding().into_canonical(self)
    }

    fn into_arrow(self) -> VortexResult<ArrayRef> {
        self.encoding().into_arrow(self)
    }
}

/// This conversion is always "free" and should not touch underlying data. All it does is create an
/// owned pointer to the underlying concrete array type.
///
/// This combined with the above [IntoCanonical] impl for [ArrayData] allows simple two-way conversions
/// between arbitrary Vortex encodings and canonical Arrow-compatible encodings.
impl From<Canonical> for ArrayData {
    fn from(value: Canonical) -> Self {
        match value {
            Canonical::Null(a) => a.into_array(),
            Canonical::Bool(a) => a.into_array(),
            Canonical::Primitive(a) => a.into_array(),
            Canonical::Struct(a) => a.into_array(),
            Canonical::List(a) => a.into_array(),
            Canonical::VarBinView(a) => a.into_array(),
            Canonical::Extension(a) => a.into_array(),
        }
    }
}

impl AsRef<ArrayData> for Canonical {
    fn as_ref(&self) -> &ArrayData {
        match self {
            Canonical::Null(a) => a.as_ref(),
            Canonical::Bool(a) => a.as_ref(),
            Canonical::Primitive(a) => a.as_ref(),
            Canonical::Struct(a) => a.as_ref(),
            Canonical::List(a) => a.as_ref(),
            Canonical::VarBinView(a) => a.as_ref(),
            Canonical::Extension(a) => a.as_ref(),
        }
    }
}

impl IntoArrayData for Canonical {
    fn into_array(self) -> ArrayData {
        match self {
            Canonical::Null(a) => a.into_array(),
            Canonical::Bool(a) => a.into_array(),
            Canonical::Primitive(a) => a.into_array(),
            Canonical::Struct(a) => a.into_array(),
            Canonical::List(a) => a.into_array(),
            Canonical::VarBinView(a) => a.into_array(),
            Canonical::Extension(a) => a.into_array(),
        }
    }
}

#[cfg(test)]
mod test {
    use std::sync::Arc;

    use arrow_array::cast::AsArray;
    use arrow_array::types::{Int32Type, Int64Type, UInt64Type};
    use arrow_array::{
        ListArray as ArrowListArray, PrimitiveArray as ArrowPrimitiveArray, StringArray,
        StringViewArray, StructArray as ArrowStructArray,
    };
    use arrow_buffer::{NullBufferBuilder, OffsetBuffer};
    use arrow_schema::{DataType, Field};

    use crate::array::{PrimitiveArray, SparseArray, StructArray};
    use crate::arrow::FromArrowArray;
    use crate::validity::Validity;
    use crate::{ArrayData, IntoArrayData, IntoCanonical};

    #[test]
    fn test_canonicalize_nested_struct() {
        // Create a struct array with multiple internal components.
        let nested_struct_array = StructArray::from_fields(&[
            (
                "a",
                PrimitiveArray::from_vec(vec![1u64], Validity::NonNullable).into_array(),
            ),
            (
                "b",
                StructArray::from_fields(&[(
                    "inner_a",
                    // The nested struct contains a SparseArray representing the primitive array
                    //   [100i64, 100i64, 100i64]
                    // SparseArray is not a canonical type, so converting `into_arrow()` should map
                    // this to the nearest canonical type (PrimitiveArray).
                    SparseArray::try_new(
                        PrimitiveArray::from_vec(vec![0u64; 1], Validity::NonNullable).into_array(),
                        PrimitiveArray::from_vec(vec![100i64], Validity::NonNullable).into_array(),
                        1,
                        0i64.into(),
                    )
                    .unwrap()
                    .into_array(),
                )])
                .unwrap()
                .into_array(),
            ),
        ])
        .unwrap();

        let arrow_struct = nested_struct_array
            .into_arrow()
            .unwrap()
            .as_any()
            .downcast_ref::<ArrowStructArray>()
            .cloned()
            .unwrap();

        assert!(arrow_struct
            .column(0)
            .as_any()
            .downcast_ref::<ArrowPrimitiveArray<UInt64Type>>()
            .is_some());

        let inner_struct = arrow_struct
            .column(1)
            .clone()
            .as_any()
            .downcast_ref::<ArrowStructArray>()
            .cloned()
            .unwrap();

        let inner_a = inner_struct
            .column(0)
            .as_any()
            .downcast_ref::<ArrowPrimitiveArray<Int64Type>>();
        assert!(inner_a.is_some());

        assert_eq!(
            inner_a.cloned().unwrap(),
            ArrowPrimitiveArray::from(vec![100i64]),
        );
    }

    #[test]
    fn roundtrip_struct() {
        let mut nulls = NullBufferBuilder::new(6);
        nulls.append_n_non_nulls(4);
        nulls.append_null();
        nulls.append_non_null();
        let names = Arc::new(StringViewArray::from_iter(vec![
            Some("Joseph"),
            None,
            Some("Angela"),
            Some("Mikhail"),
            None,
            None,
        ]));
        let ages = Arc::new(ArrowPrimitiveArray::<Int32Type>::from(vec![
            Some(25),
            Some(31),
            None,
            Some(57),
            None,
            None,
        ]));

        let arrow_struct = ArrowStructArray::new(
            vec![
                Arc::new(Field::new("name", DataType::Utf8View, true)),
                Arc::new(Field::new("age", DataType::Int32, true)),
            ]
            .into(),
            vec![names, ages],
            nulls.finish(),
        );

        let vortex_struct = ArrayData::from_arrow(&arrow_struct, true);

        assert_eq!(
            &arrow_struct,
            vortex_struct.into_arrow().unwrap().as_struct()
        );
    }

    #[test]
    fn roundtrip_list() {
        let names = Arc::new(StringArray::from_iter(vec![
            Some("Joseph"),
            Some("Angela"),
            Some("Mikhail"),
        ]));

        let arrow_list = ArrowListArray::new(
            Arc::new(Field::new_list_field(DataType::Utf8, true)),
            OffsetBuffer::from_lengths(vec![0, 2, 1]),
            names,
            None,
        );

        let vortex_list = ArrayData::from_arrow(&arrow_list, true);

        assert_eq!(&arrow_list, vortex_list.into_arrow().unwrap().as_list());
    }
}