Is constant to run stats compute

spiraldb · Nov 27, 2024 · 0cae9bc · 0cae9bc
1 parent 630835b
commit 0cae9bc
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 1 deletion.
diff --git a/vortex-array/src/array/chunked/stats.rs b/vortex-array/src/array/chunked/stats.rs
@@ -6,6 +6,14 @@ use crate::stats::{ArrayStatistics, Stat, StatisticsVTable, StatsSet};
 
 impl StatisticsVTable<ChunkedArray> for ChunkedEncoding {
     fn compute_statistics(&self, array: &ChunkedArray, stat: Stat) -> VortexResult<StatsSet> {
+        if stat == Stat::IsConstant {
+            // Short-circuiting implementation of is_constant
+            return Ok(StatsSet::from_iter([(
+                Stat::IsConstant,
+                is_constant(array).into(),
+            )]));
+        }
+
         // for UncompressedSizeInBytes, we end up with sum of chunk uncompressed sizes
         // this ignores the `chunk_offsets` array child, so it won't exactly match self.nbytes()
         Ok(array
@@ -22,3 +30,7 @@ impl StatisticsVTable<ChunkedArray> for ChunkedEncoding {
             .unwrap_or_default())
     }
 }
+
+fn is_constant(array: &ChunkedArray) -> bool {
+    array.chunks().all(|c| c.is_constant())
+}
diff --git a/vortex-array/src/data/mod.rs b/vortex-array/src/data/mod.rs
@@ -184,7 +184,7 @@ impl ArrayData {
     /// Return whether the array is constant.
     pub fn is_constant(&self) -> bool {
         self.statistics()
-            .get_as::<bool>(Stat::IsConstant)
+            .compute_as::<bool>(Stat::IsConstant)
             .unwrap_or(false)
     }
 

diff --git a/vortex-file/src/read/filtering.rs b/vortex-file/src/read/filtering.rs
@@ -74,6 +74,7 @@ impl VortexExpr for RowFilter {
             .next()
             .vortex_expect("must have at least one predicate")
             .evaluate(batch)?;
+
         for expr in filter_iter {
             if mask.statistics().compute_true_count().unwrap_or_default() == 0 {
                 return Ok(ConstantArray::new(false, batch.len()).into_array());