vortex_datetime_parts/
array.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4use std::fmt::Debug;
5use std::fmt::Display;
6use std::fmt::Formatter;
7use std::hash::Hasher;
8
9use prost::Message;
10use vortex_array::Array;
11use vortex_array::ArrayEq;
12use vortex_array::ArrayHash;
13use vortex_array::ArrayId;
14use vortex_array::ArrayParts;
15use vortex_array::ArrayRef;
16use vortex_array::ArrayView;
17use vortex_array::ExecutionCtx;
18use vortex_array::ExecutionResult;
19use vortex_array::IntoArray;
20use vortex_array::Precision;
21use vortex_array::TypedArrayRef;
22use vortex_array::arrays::TemporalArray;
23use vortex_array::buffer::BufferHandle;
24use vortex_array::dtype::DType;
25use vortex_array::dtype::Nullability;
26use vortex_array::dtype::PType;
27use vortex_array::serde::ArrayChildren;
28use vortex_array::vtable::VTable;
29use vortex_array::vtable::ValidityChild;
30use vortex_array::vtable::ValidityVTableFromChild;
31use vortex_error::VortexExpect;
32use vortex_error::VortexResult;
33use vortex_error::vortex_bail;
34use vortex_error::vortex_ensure;
35use vortex_error::vortex_err;
36use vortex_error::vortex_panic;
37use vortex_session::VortexSession;
38use vortex_session::registry::CachedId;
39
40use crate::TemporalParts;
41use crate::canonical::decode_to_temporal;
42use crate::compute::kernel::PARENT_KERNELS;
43use crate::compute::rules::PARENT_RULES;
44use crate::split_temporal;
45
46/// A [`DateTimeParts`]-encoded Vortex array.
47pub type DateTimePartsArray = Array<DateTimeParts>;
48
49impl ArrayHash for DateTimePartsData {
50    fn array_hash<H: Hasher>(&self, _state: &mut H, _precision: Precision) {}
51}
52
53impl ArrayEq for DateTimePartsData {
54    fn array_eq(&self, _other: &Self, _precision: Precision) -> bool {
55        true
56    }
57}
58
59#[derive(Clone, prost::Message)]
60#[repr(C)]
61pub struct DateTimePartsMetadata {
62    // Validity lives in the days array
63    // TODO(ngates): we should actually model this with a Tuple array when we have one.
64    #[prost(enumeration = "PType", tag = "1")]
65    pub days_ptype: i32,
66    #[prost(enumeration = "PType", tag = "2")]
67    pub seconds_ptype: i32,
68    #[prost(enumeration = "PType", tag = "3")]
69    pub subseconds_ptype: i32,
70}
71
72impl DateTimePartsMetadata {
73    pub fn get_days_ptype(&self) -> VortexResult<PType> {
74        PType::try_from(self.days_ptype)
75            .map_err(|_| vortex_err!("Invalid PType {}", self.days_ptype))
76    }
77
78    pub fn get_seconds_ptype(&self) -> VortexResult<PType> {
79        PType::try_from(self.seconds_ptype)
80            .map_err(|_| vortex_err!("Invalid PType {}", self.seconds_ptype))
81    }
82
83    pub fn get_subseconds_ptype(&self) -> VortexResult<PType> {
84        PType::try_from(self.subseconds_ptype)
85            .map_err(|_| vortex_err!("Invalid PType {}", self.subseconds_ptype))
86    }
87}
88
89impl VTable for DateTimeParts {
90    type ArrayData = DateTimePartsData;
91
92    type OperationsVTable = Self;
93    type ValidityVTable = ValidityVTableFromChild;
94
95    fn id(&self) -> ArrayId {
96        static ID: CachedId = CachedId::new("vortex.datetimeparts");
97        *ID
98    }
99
100    fn validate(
101        &self,
102        _data: &Self::ArrayData,
103        dtype: &DType,
104        len: usize,
105        slots: &[Option<ArrayRef>],
106    ) -> VortexResult<()> {
107        let days = slots[DAYS_SLOT]
108            .as_ref()
109            .vortex_expect("DateTimePartsArray days slot");
110        let seconds = slots[SECONDS_SLOT]
111            .as_ref()
112            .vortex_expect("DateTimePartsArray seconds slot");
113        let subseconds = slots[SUBSECONDS_SLOT]
114            .as_ref()
115            .vortex_expect("DateTimePartsArray subseconds slot");
116        DateTimePartsData::validate(dtype, days, seconds, subseconds, len)
117    }
118
119    fn nbuffers(_array: ArrayView<'_, Self>) -> usize {
120        0
121    }
122
123    fn buffer(_array: ArrayView<'_, Self>, idx: usize) -> BufferHandle {
124        vortex_panic!("DateTimePartsArray buffer index {idx} out of bounds")
125    }
126
127    fn buffer_name(_array: ArrayView<'_, Self>, idx: usize) -> Option<String> {
128        vortex_panic!("DateTimePartsArray buffer_name index {idx} out of bounds")
129    }
130
131    fn serialize(
132        array: ArrayView<'_, Self>,
133        _session: &VortexSession,
134    ) -> VortexResult<Option<Vec<u8>>> {
135        Ok(Some(
136            DateTimePartsMetadata {
137                days_ptype: PType::try_from(array.days().dtype())? as i32,
138                seconds_ptype: PType::try_from(array.seconds().dtype())? as i32,
139                subseconds_ptype: PType::try_from(array.subseconds().dtype())? as i32,
140            }
141            .encode_to_vec(),
142        ))
143    }
144
145    fn deserialize(
146        &self,
147        dtype: &DType,
148        len: usize,
149        metadata: &[u8],
150        _buffers: &[BufferHandle],
151        children: &dyn ArrayChildren,
152        _session: &VortexSession,
153    ) -> VortexResult<ArrayParts<Self>> {
154        let metadata = DateTimePartsMetadata::decode(metadata)?;
155        if children.len() != 3 {
156            vortex_bail!(
157                "Expected 3 children for datetime-parts encoding, found {}",
158                children.len()
159            )
160        }
161
162        let days = children.get(
163            0,
164            &DType::Primitive(metadata.get_days_ptype()?, dtype.nullability()),
165            len,
166        )?;
167        let seconds = children.get(
168            1,
169            &DType::Primitive(metadata.get_seconds_ptype()?, Nullability::NonNullable),
170            len,
171        )?;
172        let subseconds = children.get(
173            2,
174            &DType::Primitive(metadata.get_subseconds_ptype()?, Nullability::NonNullable),
175            len,
176        )?;
177
178        let slots = vec![Some(days), Some(seconds), Some(subseconds)];
179        let data = DateTimePartsData {};
180        Ok(ArrayParts::new(self.clone(), dtype.clone(), len, data).with_slots(slots))
181    }
182
183    fn slot_name(_array: ArrayView<'_, Self>, idx: usize) -> String {
184        SLOT_NAMES[idx].to_string()
185    }
186
187    fn execute(array: Array<Self>, ctx: &mut ExecutionCtx) -> VortexResult<ExecutionResult> {
188        Ok(ExecutionResult::done(
189            decode_to_temporal(&array, ctx)?.into_array(),
190        ))
191    }
192
193    fn reduce_parent(
194        array: ArrayView<'_, Self>,
195        parent: &ArrayRef,
196        child_idx: usize,
197    ) -> VortexResult<Option<ArrayRef>> {
198        PARENT_RULES.evaluate(array, parent, child_idx)
199    }
200
201    fn execute_parent(
202        array: ArrayView<'_, Self>,
203        parent: &ArrayRef,
204        child_idx: usize,
205        ctx: &mut ExecutionCtx,
206    ) -> VortexResult<Option<ArrayRef>> {
207        PARENT_KERNELS.execute(array, parent, child_idx, ctx)
208    }
209}
210
211/// The days component of the datetime, stored as an integer array.
212pub(super) const DAYS_SLOT: usize = 0;
213/// The seconds component of the datetime (within the day).
214pub(super) const SECONDS_SLOT: usize = 1;
215/// The sub-second component of the datetime.
216pub(super) const SUBSECONDS_SLOT: usize = 2;
217pub(super) const NUM_SLOTS: usize = 3;
218pub(super) const SLOT_NAMES: [&str; NUM_SLOTS] = ["days", "seconds", "subseconds"];
219
220#[derive(Clone, Debug)]
221pub struct DateTimePartsData {}
222
223impl Display for DateTimePartsData {
224    fn fmt(&self, _f: &mut Formatter<'_>) -> std::fmt::Result {
225        Ok(())
226    }
227}
228
229pub trait DateTimePartsArrayExt: TypedArrayRef<DateTimeParts> {
230    fn days(&self) -> &ArrayRef {
231        self.as_ref().slots()[DAYS_SLOT]
232            .as_ref()
233            .vortex_expect("DateTimePartsArray days slot")
234    }
235
236    fn seconds(&self) -> &ArrayRef {
237        self.as_ref().slots()[SECONDS_SLOT]
238            .as_ref()
239            .vortex_expect("DateTimePartsArray seconds slot")
240    }
241
242    fn subseconds(&self) -> &ArrayRef {
243        self.as_ref().slots()[SUBSECONDS_SLOT]
244            .as_ref()
245            .vortex_expect("DateTimePartsArray subseconds slot")
246    }
247}
248
249impl<T: TypedArrayRef<DateTimeParts>> DateTimePartsArrayExt for T {}
250
251#[derive(Clone, Debug)]
252pub struct DateTimeParts;
253
254impl DateTimeParts {
255    /// Construct a new [`DateTimePartsArray`] from its components.
256    pub fn try_new(
257        dtype: DType,
258        days: ArrayRef,
259        seconds: ArrayRef,
260        subseconds: ArrayRef,
261    ) -> VortexResult<DateTimePartsArray> {
262        let len = days.len();
263        DateTimePartsData::validate(&dtype, &days, &seconds, &subseconds, len)?;
264        let slots = vec![Some(days), Some(seconds), Some(subseconds)];
265        let data = DateTimePartsData {};
266        Ok(unsafe {
267            Array::from_parts_unchecked(
268                ArrayParts::new(DateTimeParts, dtype, len, data).with_slots(slots),
269            )
270        })
271    }
272
273    /// Construct a [`DateTimePartsArray`] from a [`TemporalArray`].
274    pub fn try_from_temporal(temporal: TemporalArray) -> VortexResult<DateTimePartsArray> {
275        let dtype = temporal.dtype().clone();
276        let TemporalParts {
277            days,
278            seconds,
279            subseconds,
280        } = split_temporal(temporal)?;
281        Self::try_new(dtype, days, seconds, subseconds)
282    }
283}
284
285impl DateTimePartsData {
286    pub fn validate(
287        dtype: &DType,
288        days: &ArrayRef,
289        seconds: &ArrayRef,
290        subseconds: &ArrayRef,
291        len: usize,
292    ) -> VortexResult<()> {
293        vortex_ensure!(days.len() == len, "expected len {len}, got {}", days.len());
294
295        if !days.dtype().is_int() || (dtype.is_nullable() != days.dtype().is_nullable()) {
296            vortex_bail!(
297                "Expected integer with nullability {}, got {}",
298                dtype.is_nullable(),
299                days.dtype()
300            );
301        }
302        if !seconds.dtype().is_int() || seconds.dtype().is_nullable() {
303            vortex_bail!(MismatchedTypes: "non-nullable integer", seconds.dtype());
304        }
305        if !subseconds.dtype().is_int() || subseconds.dtype().is_nullable() {
306            vortex_bail!(MismatchedTypes: "non-nullable integer", subseconds.dtype());
307        }
308
309        if len != seconds.len() || len != subseconds.len() {
310            vortex_bail!(
311                "Mismatched lengths {} {} {}",
312                days.len(),
313                seconds.len(),
314                subseconds.len()
315            );
316        }
317
318        Ok(())
319    }
320}
321
322impl ValidityChild<DateTimeParts> for DateTimeParts {
323    fn validity_child(array: ArrayView<'_, DateTimeParts>) -> ArrayRef {
324        array.days().clone()
325    }
326}
vortex_datetime_parts/array.rs

vortex_datetime_parts/
array.rs
