vortex_array/stats/
array.rs

1// SPDX-License-Identifier: Apache-2.0
2// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4//! Stats as they are stored on arrays.
5
6use std::sync::Arc;
7
8use parking_lot::RwLock;
9use vortex_array::ExecutionCtx;
10use vortex_error::VortexError;
11use vortex_error::VortexResult;
12use vortex_error::vortex_panic;
13
14use super::MutTypedStatsSetRef;
15use super::StatsSet;
16use super::StatsSetIntoIter;
17use super::TypedStatsSetRef;
18use crate::ArrayRef;
19use crate::aggregate_fn::fns::is_constant::is_constant;
20use crate::aggregate_fn::fns::is_sorted::is_sorted;
21use crate::aggregate_fn::fns::is_sorted::is_strict_sorted;
22use crate::aggregate_fn::fns::min_max::MinMaxResult;
23use crate::aggregate_fn::fns::min_max::min_max;
24use crate::aggregate_fn::fns::nan_count::nan_count;
25use crate::aggregate_fn::fns::sum::sum;
26use crate::aggregate_fn::fns::uncompressed_size_in_bytes::uncompressed_size_in_bytes;
27use crate::expr::stats::Precision;
28use crate::expr::stats::Stat;
29use crate::expr::stats::StatsProvider;
30use crate::scalar::Scalar;
31use crate::scalar::ScalarValue;
32
33/// A shared [`StatsSet`] stored in an array. Can be shared by copies of the array and can also be mutated in place.
34// TODO(adamg): This is a very bad name.
35#[derive(Clone, Default, Debug)]
36pub struct ArrayStats {
37    inner: Arc<RwLock<StatsSet>>,
38}
39
40/// Reference to an array's [`StatsSet`]. Can be used to get and mutate the underlying stats.
41///
42/// Constructed by calling [`ArrayStats::to_ref`].
43pub struct StatsSetRef<'a> {
44    // We need to reference back to the array
45    dyn_array_ref: &'a ArrayRef,
46    array_stats: &'a ArrayStats,
47}
48
49impl ArrayStats {
50    pub fn to_ref<'a>(&'a self, array: &'a ArrayRef) -> StatsSetRef<'a> {
51        StatsSetRef {
52            dyn_array_ref: array,
53            array_stats: self,
54        }
55    }
56
57    pub fn set(&self, stat: Stat, value: Precision<ScalarValue>) {
58        self.inner.write().set(stat, value);
59    }
60
61    pub fn clear(&self, stat: Stat) {
62        self.inner.write().clear(stat);
63    }
64
65    pub fn retain(&self, stats: &[Stat]) {
66        self.inner.write().retain_only(stats);
67    }
68}
69
70impl From<StatsSet> for ArrayStats {
71    fn from(value: StatsSet) -> Self {
72        Self {
73            inner: Arc::new(RwLock::new(value)),
74        }
75    }
76}
77
78impl From<ArrayStats> for StatsSet {
79    fn from(value: ArrayStats) -> Self {
80        value.inner.read().clone()
81    }
82}
83
84impl StatsSetRef<'_> {
85    pub(crate) fn replace(&self, stats: StatsSet) {
86        *self.array_stats.inner.write() = stats;
87    }
88
89    pub fn set_iter(&self, iter: StatsSetIntoIter) {
90        let mut guard = self.array_stats.inner.write();
91        for (stat, value) in iter {
92            guard.set(stat, value);
93        }
94    }
95
96    pub fn inherit_from(&self, stats: StatsSetRef<'_>) {
97        // Only inherit if the underlying stats are different
98        if !Arc::ptr_eq(&self.array_stats.inner, &stats.array_stats.inner) {
99            stats.with_iter(|iter| self.inherit(iter));
100        }
101    }
102
103    pub fn inherit<'a>(&self, iter: impl Iterator<Item = &'a (Stat, Precision<ScalarValue>)>) {
104        let mut guard = self.array_stats.inner.write();
105        for (stat, value) in iter {
106            if !value.is_exact() {
107                if !guard.get(*stat).is_some_and(|v| v.is_exact()) {
108                    guard.set(*stat, value.clone());
109                }
110            } else {
111                guard.set(*stat, value.clone());
112            }
113        }
114    }
115
116    pub fn with_typed_stats_set<U, F: FnOnce(TypedStatsSetRef) -> U>(&self, apply: F) -> U {
117        apply(
118            self.array_stats
119                .inner
120                .read()
121                .as_typed_ref(self.dyn_array_ref.dtype()),
122        )
123    }
124
125    pub fn with_mut_typed_stats_set<U, F: FnOnce(MutTypedStatsSetRef) -> U>(&self, apply: F) -> U {
126        apply(
127            self.array_stats
128                .inner
129                .write()
130                .as_mut_typed_ref(self.dyn_array_ref.dtype()),
131        )
132    }
133
134    pub fn to_owned(&self) -> StatsSet {
135        self.array_stats.inner.read().clone()
136    }
137
138    /// Returns a clone of the underlying [`ArrayStats`].
139    ///
140    /// Since [`ArrayStats`] uses `Arc` internally, this is a cheap reference-count increment.
141    pub fn to_array_stats(&self) -> ArrayStats {
142        self.array_stats.clone()
143    }
144
145    pub fn with_iter<
146        F: for<'a> FnOnce(&mut dyn Iterator<Item = &'a (Stat, Precision<ScalarValue>)>) -> R,
147        R,
148    >(
149        &self,
150        f: F,
151    ) -> R {
152        let lock = self.array_stats.inner.read();
153        f(&mut lock.iter())
154    }
155
156    pub fn compute_stat(&self, stat: Stat, ctx: &mut ExecutionCtx) -> VortexResult<Option<Scalar>> {
157        // If it's already computed and exact, we can return it.
158        if let Some(Precision::Exact(s)) = self.get(stat) {
159            return Ok(Some(s));
160        }
161
162        Ok(match stat {
163            Stat::Min => min_max(self.dyn_array_ref, ctx)?.map(|MinMaxResult { min, max: _ }| min),
164            Stat::Max => min_max(self.dyn_array_ref, ctx)?.map(|MinMaxResult { min: _, max }| max),
165            Stat::Sum => {
166                Stat::Sum
167                    .dtype(self.dyn_array_ref.dtype())
168                    .is_some()
169                    .then(|| {
170                        // Sum is supported for this dtype.
171                        sum(self.dyn_array_ref, ctx)
172                    })
173                    .transpose()?
174            }
175            Stat::NullCount => self.dyn_array_ref.invalid_count(ctx).ok().map(Into::into),
176            Stat::IsConstant => {
177                if self.dyn_array_ref.is_empty() {
178                    None
179                } else {
180                    Some(is_constant(self.dyn_array_ref, ctx)?.into())
181                }
182            }
183            Stat::IsSorted => Some(is_sorted(self.dyn_array_ref, ctx)?.into()),
184            Stat::IsStrictSorted => Some(is_strict_sorted(self.dyn_array_ref, ctx)?.into()),
185            Stat::UncompressedSizeInBytes => Stat::UncompressedSizeInBytes
186                .dtype(self.dyn_array_ref.dtype())
187                .is_some()
188                .then(|| uncompressed_size_in_bytes(self.dyn_array_ref, ctx))
189                .transpose()?
190                .map(|s| s.into()),
191            Stat::NaNCount => {
192                Stat::NaNCount
193                    .dtype(self.dyn_array_ref.dtype())
194                    .is_some()
195                    .then(|| {
196                        // NaNCount is supported for this dtype.
197                        nan_count(self.dyn_array_ref, ctx)
198                    })
199                    .transpose()?
200                    .map(|s| s.into())
201            }
202        })
203    }
204
205    pub fn compute_all(&self, stats: &[Stat], ctx: &mut ExecutionCtx) -> VortexResult<StatsSet> {
206        let mut stats_set = StatsSet::default();
207        for &stat in stats {
208            if let Some(s) = self.compute_stat(stat, ctx)?
209                && let Some(value) = s.into_value()
210            {
211                stats_set.set(stat, Precision::exact(value));
212            }
213        }
214        Ok(stats_set)
215    }
216}
217
218impl StatsSetRef<'_> {
219    pub fn compute_as<U: for<'a> TryFrom<&'a Scalar, Error = VortexError>>(
220        &self,
221        stat: Stat,
222        ctx: &mut ExecutionCtx,
223    ) -> Option<U> {
224        self.compute_stat(stat, ctx)
225            .inspect_err(|e| tracing::warn!("Failed to compute stat {stat}: {e}"))
226            .ok()
227            .flatten()
228            .map(|s| U::try_from(&s))
229            .transpose()
230            .unwrap_or_else(|err| {
231                vortex_panic!(
232                    err,
233                    "Failed to compute stat {} as {}",
234                    stat,
235                    std::any::type_name::<U>()
236                )
237            })
238    }
239
240    pub fn set(&self, stat: Stat, value: Precision<ScalarValue>) {
241        self.array_stats.set(stat, value);
242    }
243
244    pub fn clear(&self, stat: Stat) {
245        self.array_stats.clear(stat);
246    }
247
248    pub fn compute_min<U: for<'a> TryFrom<&'a Scalar, Error = VortexError>>(
249        &self,
250        ctx: &mut ExecutionCtx,
251    ) -> Option<U> {
252        self.compute_as(Stat::Min, ctx)
253    }
254
255    pub fn compute_max<U: for<'a> TryFrom<&'a Scalar, Error = VortexError>>(
256        &self,
257        ctx: &mut ExecutionCtx,
258    ) -> Option<U> {
259        self.compute_as(Stat::Max, ctx)
260    }
261
262    pub fn compute_is_sorted(&self, ctx: &mut ExecutionCtx) -> Option<bool> {
263        self.compute_as(Stat::IsSorted, ctx)
264    }
265
266    pub fn compute_is_strict_sorted(&self, ctx: &mut ExecutionCtx) -> Option<bool> {
267        self.compute_as(Stat::IsStrictSorted, ctx)
268    }
269
270    pub fn compute_is_constant(&self, ctx: &mut ExecutionCtx) -> Option<bool> {
271        self.compute_as(Stat::IsConstant, ctx)
272    }
273
274    pub fn compute_null_count(&self, ctx: &mut ExecutionCtx) -> Option<usize> {
275        self.compute_as(Stat::NullCount, ctx)
276    }
277
278    pub fn compute_uncompressed_size_in_bytes(&self, ctx: &mut ExecutionCtx) -> Option<usize> {
279        self.compute_as(Stat::UncompressedSizeInBytes, ctx)
280    }
281}
282
283impl StatsProvider for StatsSetRef<'_> {
284    fn get(&self, stat: Stat) -> Option<Precision<Scalar>> {
285        self.array_stats
286            .inner
287            .read()
288            .as_typed_ref(self.dyn_array_ref.dtype())
289            .get(stat)
290    }
291
292    fn len(&self) -> usize {
293        self.array_stats.inner.read().len()
294    }
295}
vortex_array/stats/array.rs

vortex_array/stats/
array.rs
