16598973893

Committed 29 Jul 2025 02:25PM UTC coverage: 82.692% (-0.01%) from 82.703%

Build # 16598973893

Build Type

push

github

Committed by

web-flow

Commit Message

Clean up stats propagation for slicing (#3356)

Reduces the amount we copy some stats (by removing into_iter that forces
a full stats copy)

---------

Signed-off-by: Nicholas Gates <nick@nickgates.com>
Signed-off-by: Robert Kruszewski <github@robertk.io>
Signed-off-by: Will Manning <will@willmanning.io>
Co-authored-by: Robert Kruszewski <github@robertk.io>
Co-authored-by: Will Manning <will@willmanning.io>

Run Details

130 of 157 new or added lines in 15 files covered. (82.8%)

30 existing lines in 13 files now uncovered.

45215 of 54679 relevant lines covered (82.69%)

184610.34 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

73.26

/vortex-array/src/stats/array.rs

// SPDX-License-Identifier: Apache-2.0
// SPDX-FileCopyrightText: Copyright the Vortex contributors

//! Stats as they are stored on arrays.

use std::sync::Arc;

use parking_lot::RwLock;
use vortex_error::{VortexError, VortexResult, vortex_panic};
use vortex_scalar::ScalarValue;

use super::{
    Precision, Stat, StatType, StatsProvider, StatsProviderExt, StatsSet, StatsSetIntoIter,
};
use crate::Array;
use crate::compute::{
    MinMaxResult, is_constant, is_sorted, is_strict_sorted, min_max, nan_count, sum,
};

/// A shared [`StatsSet`] stored in an array. Can be shared by copies of the array and can also be mutated in place.
// TODO(adamg): This is a very bad name.
#[derive(Clone, Default, Debug)]
pub struct ArrayStats {
    inner: Arc<RwLock<StatsSet>>,
}

/// Reference to an array's [`StatsSet`]. Can be used to get and mutate the underlying stats.
///
/// Constructed by calling [`ArrayStats::to_ref`].
pub struct StatsSetRef<'a> {
    // We need to reference back to the array
    dyn_array_ref: &'a dyn Array,
    array_stats: &'a ArrayStats,
}

impl ArrayStats {
    pub fn to_ref<'a>(&'a self, array: &'a dyn Array) -> StatsSetRef<'a> {
        StatsSetRef {
            dyn_array_ref: array,
            array_stats: self,
        }
    }

    pub fn set(&self, stat: Stat, value: Precision<ScalarValue>) {
        self.inner.write().set(stat, value);
    }

    pub fn clear(&self, stat: Stat) {
        self.inner.write().clear(stat);
    }

    pub fn retain(&self, stats: &[Stat]) {
        self.inner.write().retain_only(stats);
    }
}

impl From<StatsSet> for ArrayStats {
    fn from(value: StatsSet) -> Self {
        Self {
            inner: Arc::new(RwLock::new(value)),
        }
    }
}

impl From<ArrayStats> for StatsSet {
    fn from(value: ArrayStats) -> Self {
        value.inner.read().clone()
    }
}

impl StatsProvider for ArrayStats {
    fn get(&self, stat: Stat) -> Option<Precision<ScalarValue>> {
        let guard = self.inner.read();
        guard.get(stat)
    }

    fn len(&self) -> usize {
        let guard = self.inner.read();
        guard.len()
    }
}

impl StatsSetRef<'_> {
    pub fn set_iter(&self, iter: StatsSetIntoIter) {
        let mut guard = self.array_stats.inner.write();
        for (stat, value) in iter {
            guard.set(stat, value);
        }
    }

    pub fn inherit_from(&self, stats: StatsSetRef<'_>) {
        stats.with_iter(|iter| self.inherit(iter));
    }

    pub fn inherit<'a>(&self, iter: impl Iterator<Item = &'a (Stat, Precision<ScalarValue>)>) {
        // TODO(ngates): depending on statistic, this should choose the more precise one
        let mut guard = self.array_stats.inner.write();
        for (stat, value) in iter {
            guard.set(*stat, value.clone());
        }
    }

    pub fn replace(&self, stats: StatsSet) {
        *self.array_stats.inner.write() = stats;
    }

    pub fn to_owned(&self) -> StatsSet {
        self.array_stats.inner.read().clone()
    }

    pub fn with_iter<
        F: for<'a> FnOnce(&mut dyn Iterator<Item = &'a (Stat, Precision<ScalarValue>)>) -> R,
        R,
    >(
        &self,
        f: F,
    ) -> R {
        let lock = self.array_stats.inner.read();
        f(&mut lock.iter())
    }

    pub fn compute_stat(&self, stat: Stat) -> VortexResult<Option<ScalarValue>> {
        // If it's already computed and exact, we can return it.
        if let Some(Precision::Exact(stat)) = self.get(stat) {
            return Ok(Some(stat));
        }

        Ok(match stat {
            Stat::Min => {
                min_max(self.dyn_array_ref)?.map(|MinMaxResult { min, max: _ }| min.into_value())
            }
            Stat::Max => {
                min_max(self.dyn_array_ref)?.map(|MinMaxResult { min: _, max }| max.into_value())
            }
            Stat::Sum => {
                Stat::Sum
                    .dtype(self.dyn_array_ref.dtype())
                    .is_some()
                    .then(|| {
                        // Sum is supported for this dtype.
                        sum(self.dyn_array_ref)
                    })
                    .transpose()?
                    .map(|s| s.into_value())
            }
            Stat::NullCount => Some(self.dyn_array_ref.invalid_count()?.into()),
            Stat::IsConstant => {
                if self.dyn_array_ref.is_empty() {
                    None
                } else {
                    is_constant(self.dyn_array_ref)?.map(ScalarValue::from)
                }
            }
            Stat::IsSorted => Some(is_sorted(self.dyn_array_ref)?.into()),
            Stat::IsStrictSorted => Some(is_strict_sorted(self.dyn_array_ref)?.into()),
            Stat::UncompressedSizeInBytes => {
                let nbytes: ScalarValue =
                    self.dyn_array_ref.to_canonical()?.as_ref().nbytes().into();
                self.set(stat, Precision::exact(nbytes.clone()));
                Some(nbytes)
            }
            Stat::NaNCount => {
                Stat::NaNCount
                    .dtype(self.dyn_array_ref.dtype())
                    .is_some()
                    .then(|| {
                        // NaNCount is supported for this dtype.
                        nan_count(self.dyn_array_ref)
                    })
                    .transpose()?
                    .map(|s| s.into())
            }
        })
    }

    pub fn compute_all(&self, stats: &[Stat]) -> VortexResult<StatsSet> {
        let mut stats_set = StatsSet::default();
        for &stat in stats {
            if let Some(s) = self.compute_stat(stat)? {
                stats_set.set(stat, Precision::exact(s))
            }
        }
        Ok(stats_set)
    }
}

impl StatsSetRef<'_> {
    pub fn get_as<U: for<'a> TryFrom<&'a ScalarValue, Error = VortexError>>(
        &self,
        stat: Stat,
    ) -> Option<Precision<U>> {
        StatsProviderExt::get_as::<U>(self, stat)
    }

    pub fn get_as_bound<S, U>(&self) -> Option<S::Bound>
    where
        S: StatType<U>,
        U: for<'a> TryFrom<&'a ScalarValue, Error = VortexError>,
    {
        StatsProviderExt::get_as_bound::<S, U>(self)
    }

    pub fn compute_as<U: for<'a> TryFrom<&'a ScalarValue, Error = VortexError>>(
        &self,
        stat: Stat,
    ) -> Option<U> {
        self.compute_stat(stat)
            .inspect_err(|e| log::warn!("Failed to compute stat {stat}: {e}"))
            .ok()
            .flatten()
            .map(|s| U::try_from(&s))
            .transpose()
            .unwrap_or_else(|err| {
                vortex_panic!(
                    err,
                    "Failed to compute stat {} as {}",
                    stat,
                    std::any::type_name::<U>()
                )
            })
    }

    pub fn set(&self, stat: Stat, value: Precision<ScalarValue>) {
        self.array_stats.set(stat, value);
    }

    pub fn clear(&self, stat: Stat) {
        self.array_stats.clear(stat);
    }

    pub fn retain(&self, stats: &[Stat]) {
        self.array_stats.retain(stats);
    }

    pub fn compute_min<U: for<'a> TryFrom<&'a ScalarValue, Error = VortexError>>(
        &self,
    ) -> Option<U> {
        self.compute_as(Stat::Min)
    }

    pub fn compute_max<U: for<'a> TryFrom<&'a ScalarValue, Error = VortexError>>(
        &self,
    ) -> Option<U> {
        self.compute_as(Stat::Max)
    }

    pub fn compute_is_sorted(&self) -> Option<bool> {
        self.compute_as(Stat::IsSorted)
    }

    pub fn compute_is_strict_sorted(&self) -> Option<bool> {
        self.compute_as(Stat::IsStrictSorted)
    }

    pub fn compute_is_constant(&self) -> Option<bool> {
        self.compute_as(Stat::IsConstant)
    }

    pub fn compute_null_count(&self) -> Option<usize> {
        self.compute_as(Stat::NullCount)
    }

    pub fn compute_uncompressed_size_in_bytes(&self) -> Option<usize> {
        self.compute_as(Stat::UncompressedSizeInBytes)
    }
}

impl StatsProvider for StatsSetRef<'_> {
    fn get(&self, stat: Stat) -> Option<Precision<ScalarValue>> {
        self.array_stats.get(stat)
    }

    fn len(&self) -> usize {
        self.array_stats.len()
    }
}

1	// SPDX-License-Identifier: Apache-2.0
2	// SPDX-FileCopyrightText: Copyright the Vortex contributors
3
4	//! Stats as they are stored on arrays.
5
6	use std::sync::Arc;
7
8	use parking_lot::RwLock;
9	use vortex_error::{VortexError, VortexResult, vortex_panic};
10	use vortex_scalar::ScalarValue;
11
12	use super::{
13	Precision, Stat, StatType, StatsProvider, StatsProviderExt, StatsSet, StatsSetIntoIter,
14	};
15	use crate::Array;
16	use crate::compute::{
17	MinMaxResult, is_constant, is_sorted, is_strict_sorted, min_max, nan_count, sum,
18	};
19
20	/// A shared [`StatsSet`] stored in an array. Can be shared by copies of the array and can also be mutated in place.
21	// TODO(adamg): This is a very bad name.
22	#[derive(Clone, Default, Debug)]
23	pub struct ArrayStats {
24	inner: Arc<RwLock<StatsSet>>,
25	}
26
27	/// Reference to an array's [`StatsSet`]. Can be used to get and mutate the underlying stats.
28	///
29	/// Constructed by calling [`ArrayStats::to_ref`].
30	pub struct StatsSetRef<'a> {
31	// We need to reference back to the array
32	dyn_array_ref: &'a dyn Array,
33	array_stats: &'a ArrayStats,
34	}
35
36	impl ArrayStats {
37	pub fn to_ref<'a>(&'a self, array: &'a dyn Array) -> StatsSetRef<'a> {	2,244,351✔
38	StatsSetRef {	2,244,351✔
39	dyn_array_ref: array,	2,244,351✔
40	array_stats: self,	2,244,351✔
41	}	2,244,351✔
42	}	2,244,351✔
43
44	pub fn set(&self, stat: Stat, value: Precision<ScalarValue>) {	343,711✔
45	self.inner.write().set(stat, value);	343,711✔
46	}	343,711✔
47
48	pub fn clear(&self, stat: Stat) {	×
49	self.inner.write().clear(stat);	×
50	}	×
51
52	pub fn retain(&self, stats: &[Stat]) {	×
53	self.inner.write().retain_only(stats);	×
54	}	×
55	}
56
57	impl From<StatsSet> for ArrayStats {
UNCOV 58	fn from(value: StatsSet) -> Self {	×
UNCOV 59	Self {	×
UNCOV 60	inner: Arc::new(RwLock::new(value)),	×
UNCOV 61	}	×
UNCOV 62	}	×
63	}
64
65	impl From<ArrayStats> for StatsSet {
66	fn from(value: ArrayStats) -> Self {	×
67	value.inner.read().clone()	×
68	}	×
69	}
70
71	impl StatsProvider for ArrayStats {
72	fn get(&self, stat: Stat) -> Option<Precision<ScalarValue>> {	864,679✔
73	let guard = self.inner.read();	864,679✔
74	guard.get(stat)	864,679✔
75	}	864,679✔
76
77	fn len(&self) -> usize {	×
78	let guard = self.inner.read();	×
79	guard.len()	×
80	}	×
81	}
82
83	impl StatsSetRef<'_> {
84	pub fn set_iter(&self, iter: StatsSetIntoIter) {	38✔
85	let mut guard = self.array_stats.inner.write();	38✔
86	for (stat, value) in iter {	304✔
87	guard.set(stat, value);	266✔
88	}	266✔
89	}	38✔
90
NEW 91	pub fn inherit_from(&self, stats: StatsSetRef<'_>) {	×
NEW 92	stats.with_iter(\|iter\| self.inherit(iter));	×
NEW 93	}	×
94
95	pub fn inherit<'a>(&self, iter: impl Iterator<Item = &'a (Stat, Precision<ScalarValue>)>) {	56,314✔
96	// TODO(ngates): depending on statistic, this should choose the more precise one
97	let mut guard = self.array_stats.inner.write();	56,314✔
98	for (stat, value) in iter {	58,345✔
99	guard.set(*stat, value.clone());	2,031✔
100	}	2,031✔
101	}	56,314✔
102
103	pub fn replace(&self, stats: StatsSet) {	442,512✔
104	*self.array_stats.inner.write() = stats;	442,512✔
105	}	442,512✔
106
107	pub fn to_owned(&self) -> StatsSet {	556,573✔
108	self.array_stats.inner.read().clone()	556,573✔
109	}	556,573✔
110
111	pub fn with_iter<	56,314✔
112	F: for<'a> FnOnce(&mut dyn Iterator<Item = &'a (Stat, Precision<ScalarValue>)>) -> R,	56,314✔
113	R,	56,314✔
114	>(	56,314✔
115	&self,	56,314✔
116	f: F,	56,314✔
117	) -> R {	56,314✔
118	let lock = self.array_stats.inner.read();	56,314✔
119	f(&mut lock.iter())	56,314✔
120	}	56,314✔
121
122	pub fn compute_stat(&self, stat: Stat) -> VortexResult<Option<ScalarValue>> {	256,830✔
123	// If it's already computed and exact, we can return it.
124	if let Some(Precision::Exact(stat)) = self.get(stat) {	256,830✔
125	return Ok(Some(stat));	145,073✔
126	}	111,757✔
127
128	Ok(match stat {	111,757✔
129	Stat::Min => {
130	min_max(self.dyn_array_ref)?.map(\|MinMaxResult { min, max: _ }\| min.into_value())	48,251✔
131	}
132	Stat::Max => {
133	min_max(self.dyn_array_ref)?.map(\|MinMaxResult { min: _, max }\| max.into_value())	9,312✔
134	}
135	Stat::Sum => {
136	Stat::Sum	14,613✔
137	.dtype(self.dyn_array_ref.dtype())	14,613✔
138	.is_some()	14,613✔
139	.then(\|\| {	14,613✔
140	// Sum is supported for this dtype.
141	sum(self.dyn_array_ref)	8,043✔
142	})	8,043✔
143	.transpose()?	14,613✔
144	.map(\|s\| s.into_value())	14,613✔
145	}
146	Stat::NullCount => Some(self.dyn_array_ref.invalid_count()?.into()),	4,984✔
147	Stat::IsConstant => {
148	if self.dyn_array_ref.is_empty() {	6,073✔
149	None	38✔
150	} else {
151	is_constant(self.dyn_array_ref)?.map(ScalarValue::from)	6,035✔
152	}
153	}
154	Stat::IsSorted => Some(is_sorted(self.dyn_array_ref)?.into()),	5,732✔
155	Stat::IsStrictSorted => Some(is_strict_sorted(self.dyn_array_ref)?.into()),	5,745✔
156	Stat::UncompressedSizeInBytes => {
157	let nbytes: ScalarValue =	5,732✔
158	self.dyn_array_ref.to_canonical()?.as_ref().nbytes().into();	5,732✔
159	self.set(stat, Precision::exact(nbytes.clone()));	5,732✔
160	Some(nbytes)	5,732✔
161	}
162	Stat::NaNCount => {
163	Stat::NaNCount	11,315✔
164	.dtype(self.dyn_array_ref.dtype())	11,315✔
165	.is_some()	11,315✔
166	.then(\|\| {	11,315✔
167	// NaNCount is supported for this dtype.
168	nan_count(self.dyn_array_ref)	114✔
169	})	114✔
170	.transpose()?	11,315✔
171	.map(\|s\| s.into())	11,315✔
172	}
173	})
174	}	256,830✔
175
176	pub fn compute_all(&self, stats: &[Stat]) -> VortexResult<StatsSet> {	11,282✔
177	let mut stats_set = StatsSet::default();	11,282✔
178	for &stat in stats {	90,606✔
179	if let Some(s) = self.compute_stat(stat)? {	79,324✔
180	stats_set.set(stat, Precision::exact(s))	56,179✔
181	}	23,145✔
182	}
183	Ok(stats_set)	11,282✔
184	}	11,282✔
185	}
186
187	impl StatsSetRef<'_> {
188	pub fn get_as<U: for<'a> TryFrom<&'a ScalarValue, Error = VortexError>>(	343,747✔
189	&self,	343,747✔
190	stat: Stat,	343,747✔
191	) -> Option<Precision<U>> {	343,747✔
192	StatsProviderExt::get_as::<U>(self, stat)	343,747✔
193	}	343,747✔
194
195	pub fn get_as_bound<S, U>(&self) -> Option<S::Bound>	×
196	where	×
197	S: StatType<U>,	×
198	U: for<'a> TryFrom<&'a ScalarValue, Error = VortexError>,	×
199	{
200	StatsProviderExt::get_as_bound::<S, U>(self)	×
201	}	×
202
203	pub fn compute_as<U: for<'a> TryFrom<&'a ScalarValue, Error = VortexError>>(	96,945✔
204	&self,	96,945✔
205	stat: Stat,	96,945✔
206	) -> Option<U> {	96,945✔
207	self.compute_stat(stat)	96,945✔
208	.inspect_err(\|e\| log::warn!("Failed to compute stat {stat}: {e}"))	96,945✔
209	.ok()	96,945✔
210	.flatten()	96,945✔
211	.map(\|s\| U::try_from(&s))	96,945✔
212	.transpose()	96,945✔
213	.unwrap_or_else(\|err\| {	96,945✔
214	vortex_panic!(	×
215	err,	×
216	"Failed to compute stat {} as {}",	×
217	stat,
218	std::any::type_name::<U>()	×
219	)
220	})
221	}	96,945✔
222
223	pub fn set(&self, stat: Stat, value: Precision<ScalarValue>) {	343,711✔
224	self.array_stats.set(stat, value);	343,711✔
225	}	343,711✔
226
227	pub fn clear(&self, stat: Stat) {	×
NEW 228	self.array_stats.clear(stat);	×
229	}	×
230
231	pub fn retain(&self, stats: &[Stat]) {	×
NEW 232	self.array_stats.retain(stats);	×
233	}	×
234
235	pub fn compute_min<U: for<'a> TryFrom<&'a ScalarValue, Error = VortexError>>(	15,537✔
236	&self,	15,537✔
237	) -> Option<U> {	15,537✔
238	self.compute_as(Stat::Min)	15,537✔
239	}	15,537✔
240
241	pub fn compute_max<U: for<'a> TryFrom<&'a ScalarValue, Error = VortexError>>(	18✔
242	&self,	18✔
243	) -> Option<U> {	18✔
244	self.compute_as(Stat::Max)	18✔
245	}	18✔
246
247	pub fn compute_is_sorted(&self) -> Option<bool> {	×
248	self.compute_as(Stat::IsSorted)	×
249	}	×
250
251	pub fn compute_is_strict_sorted(&self) -> Option<bool> {	3,516✔
252	self.compute_as(Stat::IsStrictSorted)	3,516✔
253	}	3,516✔
254
255	pub fn compute_is_constant(&self) -> Option<bool> {	379✔
256	self.compute_as(Stat::IsConstant)	379✔
257	}	379✔
258
259	pub fn compute_null_count(&self) -> Option<usize> {	10,512✔
260	self.compute_as(Stat::NullCount)	10,512✔
261	}	10,512✔
262
263	pub fn compute_uncompressed_size_in_bytes(&self) -> Option<usize> {	×
264	self.compute_as(Stat::UncompressedSizeInBytes)	×
265	}	×
266	}
267
268	impl StatsProvider for StatsSetRef<'_> {
269	fn get(&self, stat: Stat) -> Option<Precision<ScalarValue>> {	864,679✔
270	self.array_stats.get(stat)	864,679✔
271	}	864,679✔
272
273	fn len(&self) -> usize {	×
NEW 274	self.array_stats.len()	×
275	}	×
276	}

vortex-data / vortex / 16598973893

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous