vortex_array/arrays/chunked/
array.rs1use std::fmt::Debug;
9use std::fmt::Display;
10use std::fmt::Formatter;
11
12use futures::stream;
13use vortex_buffer::BufferMut;
14use vortex_error::VortexExpect;
15use vortex_error::VortexResult;
16use vortex_error::vortex_bail;
17
18use crate::ArrayRef;
19use crate::ArraySlots;
20use crate::Canonical;
21use crate::ExecutionCtx;
22use crate::IntoArray;
23use crate::array::Array;
24use crate::array::ArrayParts;
25use crate::array::TypedArrayRef;
26use crate::array_slots;
27use crate::arrays::Chunked;
28use crate::arrays::PrimitiveArray;
29use crate::dtype::DType;
30use crate::iter::ArrayIterator;
31use crate::iter::ArrayIteratorAdapter;
32use crate::search_sorted::SearchSorted;
33use crate::search_sorted::SearchSortedSide;
34use crate::stream::ArrayStream;
35use crate::stream::ArrayStreamAdapter;
36use crate::validity::Validity;
37
38#[array_slots(Chunked)]
40pub struct ChunkedSlots {
41 #[slot(0)]
43 pub chunk_offsets: ArrayRef,
44 #[slot(1..)]
46 pub chunks: Vec<ArrayRef>,
47}
48
49#[derive(Clone, Debug)]
50pub struct ChunkedData {
51 pub(super) chunk_offsets: Vec<usize>,
52 pub(super) next_builder_slot: usize,
54}
55
56impl Display for ChunkedData {
57 fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
58 write!(f, "nchunks: {}", self.chunk_offsets.len().saturating_sub(1))
59 }
60}
61
62pub trait ChunkedArrayExt: TypedArrayRef<Chunked> {
63 fn chunk_offsets_array(&self) -> &ArrayRef {
64 self.as_ref().slots()[ChunkedSlots::CHUNK_OFFSETS]
65 .as_ref()
66 .vortex_expect("validated chunk offsets slot")
67 }
68
69 fn nchunks(&self) -> usize {
70 self.as_ref()
71 .slots()
72 .len()
73 .saturating_sub(ChunkedSlots::CHUNKS_OFFSET)
74 }
75
76 fn chunk(&self, idx: usize) -> &ArrayRef {
77 self.as_ref().slots()[ChunkedSlots::CHUNKS_OFFSET + idx]
78 .as_ref()
79 .vortex_expect("validated chunk slot")
80 }
81
82 fn iter_chunks<'a>(&'a self) -> Box<dyn Iterator<Item = &'a ArrayRef> + 'a> {
83 Box::new(
84 self.as_ref().slots()[ChunkedSlots::CHUNKS_OFFSET..]
85 .iter()
86 .map(|slot| slot.as_ref().vortex_expect("validated chunk slot")),
87 )
88 }
89
90 fn chunks(&self) -> Vec<ArrayRef> {
91 self.iter_chunks().cloned().collect()
92 }
93
94 fn non_empty_chunks<'a>(&'a self) -> Box<dyn Iterator<Item = &'a ArrayRef> + 'a> {
95 Box::new(self.iter_chunks().filter(|chunk| !chunk.is_empty()))
96 }
97
98 fn chunk_offset_values(&self) -> &[usize] {
100 &self.chunk_offsets
101 }
102
103 fn find_chunk_idx(&self, index: usize) -> VortexResult<(usize, usize)> {
104 assert!(
105 index <= self.as_ref().len(),
106 "Index out of bounds of the array"
107 );
108 let chunk_offset_values = self.chunk_offset_values();
109 let index_chunk = chunk_offset_values
110 .search_sorted(&index, SearchSortedSide::Right)?
111 .to_ends_index(self.nchunks() + 1)
112 .saturating_sub(1);
113 let chunk_start = chunk_offset_values[index_chunk];
114 let index_in_chunk = index - chunk_start;
115 Ok((index_chunk, index_in_chunk))
116 }
117
118 fn array_iterator(&self) -> impl ArrayIterator + '_ {
119 ArrayIteratorAdapter::new(
120 self.as_ref().dtype().clone(),
121 self.iter_chunks().map(|chunk| Ok(chunk.clone())),
122 )
123 }
124
125 fn array_stream(&self) -> impl ArrayStream + '_ {
126 ArrayStreamAdapter::new(
127 self.as_ref().dtype().clone(),
128 stream::iter(self.iter_chunks().map(|chunk| Ok(chunk.clone()))),
129 )
130 }
131}
132impl<T: TypedArrayRef<Chunked>> ChunkedArrayExt for T {}
133
134impl ChunkedData {
135 pub(super) fn new(chunk_offsets: Vec<usize>) -> Self {
136 Self {
137 chunk_offsets,
138 next_builder_slot: ChunkedSlots::CHUNKS_OFFSET,
139 }
140 }
141
142 pub(super) fn make_chunk_offsets_array(chunk_offsets: &[usize]) -> ArrayRef {
143 let mut chunk_offsets_buf = BufferMut::<u64>::with_capacity(chunk_offsets.len());
144 for &offset in chunk_offsets {
145 let offset = u64::try_from(offset)
146 .vortex_expect("chunk offset must fit in u64 for serialization");
147 unsafe { chunk_offsets_buf.push_unchecked(offset) }
148 }
149 PrimitiveArray::new(chunk_offsets_buf.freeze(), Validity::NonNullable).into_array()
150 }
151
152 pub fn validate(chunks: &[ArrayRef], dtype: &DType) -> VortexResult<()> {
156 for chunk in chunks {
157 if chunk.dtype() != dtype {
158 vortex_bail!(MismatchedTypes: dtype, chunk.dtype());
159 }
160 }
161
162 Ok(())
163 }
164}
165
166impl Array<Chunked> {
167 fn parts_from_chunks<const VALIDATE: bool>(
168 chunks: impl IntoIterator<Item = ArrayRef>,
169 dtype: DType,
170 ) -> VortexResult<ArrayParts<Chunked>> {
171 let chunks = chunks.into_iter();
172 let (lower, _) = chunks.size_hint();
173 let mut slots = ArraySlots::with_capacity(ChunkedSlots::CHUNKS_OFFSET + lower);
174 slots.push(None);
175 let mut chunk_offsets = Vec::with_capacity(lower + 1);
176 chunk_offsets.push(0);
177 let mut len = 0usize;
178
179 for chunk in chunks {
180 if VALIDATE && chunk.dtype() != &dtype {
181 vortex_bail!(MismatchedTypes: &dtype, chunk.dtype());
182 }
183 len += chunk.len();
184 chunk_offsets.push(len);
185 slots.push(Some(chunk));
186 }
187
188 slots[ChunkedSlots::CHUNK_OFFSETS] =
189 Some(ChunkedData::make_chunk_offsets_array(&chunk_offsets));
190 Ok(ArrayParts::new(Chunked, dtype, len, ChunkedData::new(chunk_offsets)).with_slots(slots))
191 }
192
193 pub(super) fn with_next_builder_slot(mut self, next_builder_slot: usize) -> Self {
194 if let Some(data) = self.data_mut() {
195 data.next_builder_slot = next_builder_slot;
196 return self;
197 }
198 let stats = self.statistics().to_owned();
201 let mut data = self.data().clone();
202 data.next_builder_slot = next_builder_slot;
203 unsafe {
205 Array::from_parts_unchecked(
206 ArrayParts::new(Chunked, self.dtype().clone(), self.len(), data)
207 .with_slots(self.slots().iter().cloned().collect::<ArraySlots>()),
208 )
209 }
210 .with_stats_set(stats)
211 }
212
213 pub fn try_new(chunks: impl IntoIterator<Item = ArrayRef>, dtype: DType) -> VortexResult<Self> {
215 Array::try_from_parts(Self::parts_from_chunks::<true>(chunks, dtype)?)
216 }
217
218 pub fn rechunk(
219 &self,
220 target_bytesize: u64,
221 target_rowsize: usize,
222 ctx: &mut ExecutionCtx,
223 ) -> VortexResult<Self> {
224 let mut new_chunks = Vec::new();
225 let mut chunks_to_combine = Vec::new();
226 let mut new_chunk_n_bytes = 0;
227 let mut new_chunk_n_elements = 0;
228 for chunk in self.iter_chunks() {
229 let n_bytes = chunk.nbytes();
230 let n_elements = chunk.len();
231
232 if (new_chunk_n_bytes + n_bytes > target_bytesize
233 || new_chunk_n_elements + n_elements > target_rowsize)
234 && !chunks_to_combine.is_empty()
235 {
236 let canonical = unsafe {
237 Array::<Chunked>::new_unchecked(chunks_to_combine, self.dtype().clone())
238 }
239 .into_array()
240 .execute::<Canonical>(ctx)?
241 .into_array();
242 new_chunks.push(canonical);
243
244 new_chunk_n_bytes = 0;
245 new_chunk_n_elements = 0;
246 chunks_to_combine = Vec::new();
247 }
248
249 if n_bytes > target_bytesize || n_elements > target_rowsize {
250 new_chunks.push(chunk.clone());
251 } else {
252 new_chunk_n_bytes += n_bytes;
253 new_chunk_n_elements += n_elements;
254 chunks_to_combine.push(chunk.clone());
255 }
256 }
257
258 if !chunks_to_combine.is_empty() {
259 let canonical =
260 unsafe { Array::<Chunked>::new_unchecked(chunks_to_combine, self.dtype().clone()) }
261 .into_array()
262 .execute::<Canonical>(ctx)?
263 .into_array();
264 new_chunks.push(canonical);
265 }
266
267 unsafe { Ok(Self::new_unchecked(new_chunks, self.dtype().clone())) }
268 }
269
270 pub unsafe fn new_unchecked(chunks: impl IntoIterator<Item = ArrayRef>, dtype: DType) -> Self {
276 let parts = Self::parts_from_chunks::<false>(chunks, dtype)
277 .vortex_expect("unchecked chunked construction cannot fail");
278 unsafe { Array::from_parts_unchecked(parts) }
279 }
280}
281
282impl FromIterator<ArrayRef> for Array<Chunked> {
283 fn from_iter<T: IntoIterator<Item = ArrayRef>>(iter: T) -> Self {
284 let chunks: Vec<ArrayRef> = iter.into_iter().collect();
285 let dtype = chunks
286 .first()
287 .map(|c| c.dtype().clone())
288 .vortex_expect("Cannot infer DType from an empty iterator");
289 Array::<Chunked>::try_new(chunks, dtype)
290 .vortex_expect("Failed to create chunked array from iterator")
291 }
292}
293
294#[cfg(test)]
295mod test {
296 use vortex_buffer::buffer;
297 use vortex_error::VortexResult;
298
299 use crate::IntoArray;
300 use crate::VortexSessionExecute;
301 use crate::array_session;
302 use crate::arrays::ChunkedArray;
303 use crate::arrays::PrimitiveArray;
304 use crate::arrays::chunked::ChunkedArrayExt;
305 use crate::assert_arrays_eq;
306 use crate::dtype::DType;
307 use crate::dtype::Nullability;
308 use crate::dtype::PType;
309 use crate::validity::Validity;
310
311 #[test]
312 fn test_rechunk_one_chunk() {
313 let mut ctx = array_session().create_execution_ctx();
314 let chunked = ChunkedArray::try_new(
315 vec![buffer![0u64].into_array()],
316 DType::Primitive(PType::U64, Nullability::NonNullable),
317 )
318 .unwrap();
319
320 let rechunked = chunked.rechunk(1 << 16, 1 << 16, &mut ctx).unwrap();
321
322 assert_arrays_eq!(chunked, rechunked, &mut ctx);
323 }
324
325 #[test]
326 fn test_rechunk_two_chunks() {
327 let mut ctx = array_session().create_execution_ctx();
328 let chunked = ChunkedArray::try_new(
329 vec![buffer![0u64].into_array(), buffer![5u64].into_array()],
330 DType::Primitive(PType::U64, Nullability::NonNullable),
331 )
332 .unwrap();
333
334 let rechunked = chunked.rechunk(1 << 16, 1 << 16, &mut ctx).unwrap();
335
336 assert_eq!(rechunked.nchunks(), 1);
337 assert_arrays_eq!(chunked, rechunked, &mut ctx);
338 }
339
340 #[test]
341 fn test_rechunk_tiny_target_chunks() {
342 let mut ctx = array_session().create_execution_ctx();
343 let chunked = ChunkedArray::try_new(
344 vec![
345 buffer![0u64, 1, 2, 3].into_array(),
346 buffer![4u64, 5].into_array(),
347 ],
348 DType::Primitive(PType::U64, Nullability::NonNullable),
349 )
350 .unwrap();
351
352 let rechunked = chunked.rechunk(1 << 16, 5, &mut ctx).unwrap();
353
354 assert_eq!(rechunked.nchunks(), 2);
355 assert!(rechunked.iter_chunks().all(|c| c.len() < 5));
356 assert_arrays_eq!(chunked, rechunked, &mut ctx);
357 }
358
359 #[test]
360 fn test_rechunk_with_too_big_chunk() {
361 let mut ctx = array_session().create_execution_ctx();
362 let chunked = ChunkedArray::try_new(
363 vec![
364 buffer![0u64, 1, 2].into_array(),
365 buffer![42_u64; 6].into_array(),
366 buffer![4u64, 5].into_array(),
367 buffer![6u64, 7].into_array(),
368 buffer![8u64, 9].into_array(),
369 ],
370 DType::Primitive(PType::U64, Nullability::NonNullable),
371 )
372 .unwrap();
373
374 let rechunked = chunked.rechunk(1 << 16, 5, &mut ctx).unwrap();
375 assert_eq!(rechunked.nchunks(), 4);
378 assert_arrays_eq!(chunked, rechunked, &mut ctx);
379 }
380
381 #[test]
382 fn test_empty_chunks_all_valid() -> VortexResult<()> {
383 let mut ctx = array_session().create_execution_ctx();
384 let chunks = vec![
386 PrimitiveArray::new(buffer![1u64, 2, 3], Validity::AllValid).into_array(),
387 PrimitiveArray::new(buffer![0u64; 0], Validity::AllValid).into_array(), PrimitiveArray::new(buffer![4u64, 5], Validity::AllValid).into_array(),
389 PrimitiveArray::new(buffer![0u64; 0], Validity::AllValid).into_array(), ];
391
392 let chunked =
393 ChunkedArray::try_new(chunks, DType::Primitive(PType::U64, Nullability::Nullable))?;
394
395 assert!(chunked.all_valid(&mut ctx)?);
397 assert!(!chunked.into_array().all_invalid(&mut ctx)?);
398
399 Ok(())
400 }
401
402 #[test]
403 fn test_empty_chunks_all_invalid() -> VortexResult<()> {
404 let mut ctx = array_session().create_execution_ctx();
405 let chunks = vec![
407 PrimitiveArray::new(buffer![1u64, 2], Validity::AllInvalid).into_array(),
408 PrimitiveArray::new(buffer![0u64; 0], Validity::AllInvalid).into_array(), PrimitiveArray::new(buffer![3u64, 4, 5], Validity::AllInvalid).into_array(),
410 PrimitiveArray::new(buffer![0u64; 0], Validity::AllInvalid).into_array(), ];
412
413 let chunked =
414 ChunkedArray::try_new(chunks, DType::Primitive(PType::U64, Nullability::Nullable))?;
415
416 assert!(!chunked.all_valid(&mut ctx)?);
418 assert!(chunked.into_array().all_invalid(&mut ctx)?);
419
420 Ok(())
421 }
422
423 #[test]
424 fn test_empty_chunks_mixed_validity() -> VortexResult<()> {
425 let mut ctx = array_session().create_execution_ctx();
426 let chunks = vec![
428 PrimitiveArray::new(buffer![1u64, 2], Validity::AllValid).into_array(),
429 PrimitiveArray::new(buffer![0u64; 0], Validity::AllValid).into_array(), PrimitiveArray::new(buffer![3u64, 4], Validity::AllInvalid).into_array(),
431 PrimitiveArray::new(buffer![0u64; 0], Validity::AllInvalid).into_array(), ];
433
434 let chunked =
435 ChunkedArray::try_new(chunks, DType::Primitive(PType::U64, Nullability::Nullable))?;
436
437 assert!(!chunked.all_valid(&mut ctx)?);
439 assert!(!chunked.into_array().all_invalid(&mut ctx)?);
440
441 Ok(())
442 }
443}