reserve pushable containers in parquet extend_from_decoder

src/io/parquet/read/deserialize/binary/utils.rs

@@ -25,6 +25,9 @@ impl<O: Offset> Offsets<O> {
 }
 
 impl<O: Offset> Pushable<O> for Offsets<O> {
+    fn reserve(&mut self, additional: usize) {
+        self.0.reserve(additional)
+    }
     #[inline]
     fn len(&self) -> usize {
         self.0.len() - 1
@@ -88,6 +91,9 @@ impl<O: Offset> Binary<O> {
 }
 
 impl<'a, O: Offset> Pushable<&'a [u8]> for Binary<O> {
+    fn reserve(&mut self, _additional: usize) {
+        // a no-op as it is not really clear how to resize a binary
+    }
     #[inline]
     fn len(&self) -> usize {
         self.len()

src/io/parquet/read/deserialize/fixed_size_binary/utils.rs

@@ -30,6 +30,9 @@ impl FixedSizeBinary {
 }
 
 impl<'a> Pushable<&'a [u8]> for FixedSizeBinary {
+    fn reserve(&mut self, additional: usize) {
+        self.values.reserve(additional * self.size);
+    }
     #[inline]
     fn push(&mut self, value: &[u8]) {
         debug_assert_eq!(value.len(), self.size);

src/io/parquet/read/deserialize/utils.rs

@@ -30,14 +30,17 @@ pub fn not_implemented(page: &DataPage) -> Error {
 
 /// A private trait representing structs that can receive elements.
 pub(super) trait Pushable<T>: Sized {
-    //fn reserve(&mut self, additional: usize);
+    fn reserve(&mut self, additional: usize);
     fn push(&mut self, value: T);
     fn len(&self) -> usize;
     fn push_null(&mut self);
     fn extend_constant(&mut self, additional: usize, value: T);
 }
 
 impl Pushable<bool> for MutableBitmap {
+    fn reserve(&mut self, additional: usize) {
+        MutableBitmap::reserve(self, additional)
+    }
     #[inline]
     fn len(&self) -> usize {
         self.len()
@@ -60,6 +63,9 @@ impl Pushable<bool> for MutableBitmap {
 }
 
 impl<A: Copy + Default> Pushable<A> for Vec<A> {
+    fn reserve(&mut self, additional: usize) {
+        Vec::reserve(self, additional)
+    }
     #[inline]
     fn len(&self) -> usize {
         self.len()
@@ -304,6 +310,7 @@ pub(super) fn extend_from_decoder<'a, T: Default, P: Pushable<T>, I: Iterator<It
                 // consume `length` items
                 let iter = BitmapIter::new(values, offset, length);
                 let iter = Zip::new(iter, &mut values_iter);
+                pushable.reserve(length);
 
                 for item in iter {
                     if let Some(item) = item {
@@ -319,6 +326,7 @@ pub(super) fn extend_from_decoder<'a, T: Default, P: Pushable<T>, I: Iterator<It
             FilteredHybridEncoded::Repeated { is_set, length } => {
                 validity.extend_constant(length, is_set);
                 if is_set {
+                    pushable.reserve(length);
                     (0..length).for_each(|_| pushable.push(values_iter.next().unwrap()));
                 } else {
                     pushable.extend_constant(length, T::default());