Stats result (#52)

spiraldb · Mar 4, 2024 · 8e9845b · 8e9845b
1 parent db64f5c
commit 8e9845b
Show file tree

Hide file tree

Showing 19 changed files with 65 additions and 53 deletions.
diff --git a/deps/fastlanez b/deps/fastlanez
diff --git a/vortex-alp/src/stats.rs b/vortex-alp/src/stats.rs
@@ -1,11 +1,12 @@
 use std::collections::HashMap;
+use vortex::error::VortexResult;
 
 use crate::ALPArray;
 use vortex::stats::{Stat, StatsCompute, StatsSet};
 
 impl StatsCompute for ALPArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
         // TODO(ngates): implement based on the encoded array
-        StatsSet::from(HashMap::new())
+        Ok(StatsSet::from(HashMap::new()))
     }
 }
diff --git a/vortex-dict/src/stats.rs b/vortex-dict/src/stats.rs
@@ -1,9 +1,10 @@
+use vortex::error::VortexResult;
 use vortex::stats::{Stat, StatsCompute, StatsSet};
 
 use crate::dict::DictArray;
 
 impl StatsCompute for DictArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
         let mut stats = StatsSet::new();
 
         if let Some(rc) = self.codes().stats().get_or_compute(&Stat::RunCount) {
@@ -46,6 +47,6 @@ impl StatsCompute for DictArray {
             }
         }
 
-        stats
+        Ok(stats)
     }
 }
diff --git a/vortex-fastlanes/src/bitpacking/mod.rs b/vortex-fastlanes/src/bitpacking/mod.rs
@@ -1,5 +1,4 @@
 use std::any::Any;
-use std::collections::HashMap;
 use std::sync::{Arc, RwLock};
 
 use vortex::array::{
@@ -161,9 +160,8 @@ impl ArrayDisplay for BitPackedArray {
 }
 
 impl StatsCompute for BitPackedArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
-        // TODO(ngates): implement based on the encoded array
-        StatsSet::from(HashMap::new())
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
+        Ok(StatsSet::default())
     }
 }
 

diff --git a/vortex-fastlanes/src/for/mod.rs b/vortex-fastlanes/src/for/mod.rs
@@ -1,5 +1,4 @@
 use std::any::Any;
-use std::collections::HashMap;
 use std::sync::{Arc, RwLock};
 
 use vortex::array::{Array, ArrayRef, ArrowIterator, Encoding, EncodingId, EncodingRef};
@@ -123,9 +122,8 @@ impl ArrayDisplay for FoRArray {
 }
 
 impl StatsCompute for FoRArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
-        // TODO(ngates): implement based on the encoded array
-        StatsSet::from(HashMap::new())
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
+        Ok(StatsSet::default())
     }
 }
 

diff --git a/vortex-ffor/src/stats.rs b/vortex-ffor/src/stats.rs
@@ -1,11 +1,10 @@
-use std::collections::HashMap;
+use vortex::error::VortexResult;
 
 use crate::FFORArray;
 use vortex::stats::{Stat, StatsCompute, StatsSet};
 
 impl StatsCompute for FFORArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
-        // TODO(ngates): implement based on the encoded array
-        StatsSet::from(HashMap::new())
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
+        Ok(StatsSet::default())
     }
 }
diff --git a/vortex-ree/src/stats.rs b/vortex-ree/src/stats.rs
@@ -1,8 +1,9 @@
 use crate::REEArray;
+use vortex::error::VortexResult;
 use vortex::stats::{Stat, StatsCompute, StatsSet};
 
 impl StatsCompute for REEArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
         todo!()
     }
 }
diff --git a/vortex-roaring/src/boolean/stats.rs b/vortex-roaring/src/boolean/stats.rs
@@ -1,10 +1,11 @@
 use vortex::array::Array;
+use vortex::error::VortexResult;
 use vortex::stats::{Stat, StatsCompute, StatsSet};
 
 use crate::boolean::RoaringBoolArray;
 
 impl StatsCompute for RoaringBoolArray {
-    fn compute(&self, stat: &Stat) -> StatsSet {
+    fn compute(&self, stat: &Stat) -> VortexResult<StatsSet> {
         let cardinality = self.bitmap().cardinality() as usize;
         if let Some(value) = match stat {
             Stat::IsConstant => Some((cardinality == self.len() || cardinality == 0).into()),
@@ -26,9 +27,9 @@ impl StatsCompute for RoaringBoolArray {
             Stat::NullCount => Some(0.into()),
             _ => None,
         } {
-            StatsSet::of(stat.clone(), value)
+            Ok(StatsSet::of(stat.clone(), value))
         } else {
-            StatsSet::default()
+            Ok(StatsSet::default())
         }
     }
 }

diff --git a/vortex-roaring/src/integer/stats.rs b/vortex-roaring/src/integer/stats.rs
@@ -1,9 +1,10 @@
+use vortex::error::VortexResult;
 use vortex::stats::{Stat, StatsCompute, StatsSet};
 
 use crate::RoaringIntArray;
 
 impl StatsCompute for RoaringIntArray {
-    fn compute(&self, stat: &Stat) -> StatsSet {
+    fn compute(&self, stat: &Stat) -> VortexResult<StatsSet> {
         if let Some(value) = match stat {
             Stat::IsConstant => Some((self.bitmap.cardinality() <= 1).into()),
             Stat::IsSorted => Some(true.into()),
@@ -13,9 +14,9 @@ impl StatsCompute for RoaringIntArray {
             Stat::NullCount => Some(0.into()),
             _ => None,
         } {
-            StatsSet::of(stat.clone(), value)
+            Ok(StatsSet::of(stat.clone(), value))
         } else {
-            StatsSet::default()
+            Ok(StatsSet::default())
         }
     }
 }
diff --git a/vortex-zigzag/src/stats.rs b/vortex-zigzag/src/stats.rs
@@ -1,12 +1,13 @@
 use std::collections::HashMap;
+use vortex::error::VortexResult;
 
 use vortex::stats::{Stat, StatsCompute, StatsSet};
 
 use crate::zigzag::ZigZagArray;
 
 impl StatsCompute for ZigZagArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
         // TODO(ngates): implement based on the encoded array
-        StatsSet::from(HashMap::new())
+        Ok(StatsSet::from(HashMap::new()))
     }
 }
diff --git a/vortex/src/array/bool/stats.rs b/vortex/src/array/bool/stats.rs
@@ -2,15 +2,16 @@ use std::collections::HashMap;
 
 use crate::array::bool::BoolArray;
 use crate::array::Array;
+use crate::error::VortexResult;
 use crate::stats::{Stat, StatsCompute, StatsSet};
 
 impl StatsCompute for BoolArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
         if self.len() == 0 {
-            return StatsSet::from(HashMap::from([
+            return Ok(StatsSet::from(HashMap::from([
                 (Stat::TrueCount, 0.into()),
                 (Stat::RunCount, 0.into()),
-            ]));
+            ])));
         }
 
         let mut prev_bit = self.buffer().value(0);
@@ -28,7 +29,7 @@ impl StatsCompute for BoolArray {
         }
         run_count += 1;
 
-        StatsSet::from(HashMap::from([
+        Ok(StatsSet::from(HashMap::from([
             (Stat::Min, (true_count == self.len()).into()),
             (Stat::Max, (true_count > 0).into()),
             (
@@ -37,6 +38,6 @@ impl StatsCompute for BoolArray {
             ),
             (Stat::RunCount, run_count.into()),
             (Stat::TrueCount, true_count.into()),
-        ]))
+        ])))
     }
 }
diff --git a/vortex/src/array/chunked/stats.rs b/vortex/src/array/chunked/stats.rs
@@ -1,9 +1,11 @@
 use crate::array::chunked::ChunkedArray;
+use crate::error::VortexResult;
 use crate::stats::{Stat, StatsCompute, StatsSet};
 
 impl StatsCompute for ChunkedArray {
-    fn compute(&self, stat: &Stat) -> StatsSet {
-        self.chunks()
+    fn compute(&self, stat: &Stat) -> VortexResult<StatsSet> {
+        Ok(self
+            .chunks()
             .iter()
             .map(|c| {
                 let s = c.stats();
@@ -14,6 +16,6 @@ impl StatsCompute for ChunkedArray {
             .fold(StatsSet::new(), |mut acc, x| {
                 acc.merge(&x);
                 acc
-            })
+            }))
     }
 }
diff --git a/vortex/src/array/constant/stats.rs b/vortex/src/array/constant/stats.rs
@@ -3,11 +3,12 @@ use std::collections::HashMap;
 use crate::array::constant::ConstantArray;
 use crate::array::Array;
 use crate::dtype::{DType, Nullability};
+use crate::error::VortexResult;
 use crate::scalar::{BoolScalar, PScalar, Scalar};
 use crate::stats::{Stat, StatsCompute, StatsSet};
 
 impl StatsCompute for ConstantArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
         let mut m = HashMap::from([
             (Stat::Max, dyn_clone::clone_box(self.scalar())),
             (Stat::Min, dyn_clone::clone_box(self.scalar())),
@@ -31,6 +32,7 @@ impl StatsCompute for ConstantArray {
                 .boxed(),
             );
         }
-        StatsSet::from(m)
+
+        Ok(StatsSet::from(m))
     }
 }
diff --git a/vortex/src/array/primitive/stats.rs b/vortex/src/array/primitive/stats.rs
@@ -6,13 +6,14 @@ use half::f16;
 use num_traits::{NumCast, PrimInt};
 
 use crate::array::primitive::PrimitiveArray;
+use crate::error::VortexResult;
 use crate::ptype::match_each_native_ptype;
 use crate::scalar::ListScalarVec;
 use crate::scalar::Scalar;
 use crate::stats::{Stat, StatsCompute, StatsSet};
 
 impl StatsCompute for PrimitiveArray {
-    fn compute(&self, stat: &Stat) -> StatsSet {
+    fn compute(&self, stat: &Stat) -> VortexResult<StatsSet> {
         match_each_native_ptype!(self.ptype(), |$P| {
             WrappedPrimitive::<$P>::new(self).compute(stat)
         })
@@ -30,7 +31,7 @@ impl<'a, P> WrappedPrimitive<'a, P> {
 macro_rules! integer_stats {
     ($T:ty) => {
         impl StatsCompute for WrappedPrimitive<'_, $T> {
-            fn compute(&self, _stat: &Stat) -> StatsSet {
+            fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
                 integer_stats::<$T>(self.0)
             }
         }
@@ -49,7 +50,7 @@ integer_stats!(u64);
 macro_rules! float_stats {
     ($T:ty) => {
         impl StatsCompute for WrappedPrimitive<'_, $T> {
-            fn compute(&self, _stat: &Stat) -> StatsSet {
+            fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
                 float_stats::<$T>(self.0)
             }
         }
@@ -60,7 +61,9 @@ float_stats!(f16);
 float_stats!(f32);
 float_stats!(f64);
 
-fn integer_stats<T: ArrowNativeType + NumCast + PrimInt>(array: &PrimitiveArray) -> StatsSet
+fn integer_stats<T: ArrowNativeType + NumCast + PrimInt>(
+    array: &PrimitiveArray,
+) -> VortexResult<StatsSet>
 where
     Box<dyn Scalar>: From<T>,
 {
@@ -97,18 +100,18 @@ where
     }
     run_count += 1;
 
-    StatsSet::from(HashMap::from([
+    Ok(StatsSet::from(HashMap::from([
         (Stat::Min, min.into()),
         (Stat::Max, max.into()),
         (Stat::IsConstant, (min == max).into()),
         (Stat::BitWidthFreq, ListScalarVec(bit_widths).into()),
         (Stat::IsSorted, is_sorted.into()),
         (Stat::IsStrictSorted, (is_sorted && is_strict_sorted).into()),
         (Stat::RunCount, run_count.into()),
-    ]))
+    ])))
 }
 
-fn float_stats<T: ArrowNativeType + NumCast>(array: &PrimitiveArray) -> StatsSet
+fn float_stats<T: ArrowNativeType + NumCast>(array: &PrimitiveArray) -> VortexResult<StatsSet>
 where
     Box<dyn Scalar>: From<T>,
 {
@@ -136,13 +139,13 @@ where
     }
     run_count += 1;
 
-    StatsSet::from(HashMap::from([
+    Ok(StatsSet::from(HashMap::from([
         (Stat::Min, min.into()),
         (Stat::Max, max.into()),
         (Stat::IsConstant, (min == max).into()),
         (Stat::IsSorted, is_sorted.into()),
         (Stat::RunCount, run_count.into()),
-    ]))
+    ])))
 }
 
 #[cfg(test)]

diff --git a/vortex/src/array/sparse/stats.rs b/vortex/src/array/sparse/stats.rs
@@ -1,8 +1,9 @@
 use crate::array::sparse::SparseArray;
+use crate::error::VortexResult;
 use crate::stats::{Stat, StatsCompute, StatsSet};
 
 impl StatsCompute for SparseArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
         todo!()
     }
 }
diff --git a/vortex/src/array/struct_/stats.rs b/vortex/src/array/struct_/stats.rs
@@ -1,8 +1,9 @@
 use crate::array::struct_::StructArray;
+use crate::error::VortexResult;
 use crate::stats::{Stat, StatsCompute, StatsSet};
 
 impl StatsCompute for StructArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
         todo!()
     }
 }
diff --git a/vortex/src/array/typed/stats.rs b/vortex/src/array/typed/stats.rs
@@ -1,8 +1,9 @@
 use crate::array::typed::TypedArray;
+use crate::error::VortexResult;
 use crate::stats::{Stat, StatsCompute, StatsSet};
 
 impl StatsCompute for TypedArray {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
         todo!()
     }
 }
diff --git a/vortex/src/array/varbin/stats.rs b/vortex/src/array/varbin/stats.rs
@@ -16,7 +16,7 @@ impl<T> StatsCompute for T
 where
     T: BinaryArray + Array,
 {
-    fn compute(&self, _stat: &Stat) -> StatsSet {
+    fn compute(&self, _stat: &Stat) -> VortexResult<StatsSet> {
         let mut min = vec![0xFF];
         let mut max = vec![0x00];
         let mut is_constant = true;
@@ -41,7 +41,7 @@ where
             runs += 1;
         }
 
-        StatsSet::from(HashMap::from([
+        Ok(StatsSet::from(HashMap::from([
             (
                 Stat::Min,
                 if matches!(self.dtype(), DType::Utf8(_)) {
@@ -61,7 +61,7 @@ where
             (Stat::RunCount, runs.into()),
             (Stat::IsSorted, is_sorted.into()),
             (Stat::IsConstant, is_constant.into()),
-        ]))
+        ])))
     }
 }