add bytes_estimate for binary push in parquet deserialize

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

@@ -480,9 +480,13 @@ impl<'a, O: Offset> utils::Decoder<'a> for BinaryDecoder<O> {
 
 pub(super) fn finish<O: Offset, A: TraitBinaryArray<O>>(
     data_type: &DataType,
-    values: Binary<O>,
-    validity: MutableBitmap,
+    mut values: Binary<O>,
+    mut validity: MutableBitmap,
 ) -> Result<A> {
+    values.offsets.0.shrink_to_fit();
+    values.values.shrink_to_fit();
+    validity.shrink_to_fit();
+
     A::try_new(
         data_type.clone(),
         values.offsets.0.into(),

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

@@ -56,13 +56,21 @@ impl<O: Offset> Binary<O> {
         offsets.push(O::default());
         Self {
             offsets: Offsets(offsets),
-            values: Vec::with_capacity(capacity * 24),
+            values: Vec::with_capacity(capacity.min(100) * 24),
             last_offset: O::default(),
         }
     }
 
     #[inline]
     pub fn push(&mut self, v: &[u8]) {
+        if self.offsets.0.len() == 101 && self.offsets.0.capacity() > 101 {
+            let bytes_per_row = self.values.len() / 100 + 1;
+            let bytes_estimate = bytes_per_row * self.offsets.0.capacity();
+            if bytes_estimate > self.values.capacity() {
+                self.values.reserve(bytes_estimate - self.values.capacity());
+            }
+        }
+
         self.values.extend(v);
         self.last_offset += O::from_usize(v.len()).unwrap();
         self.offsets.push(self.last_offset)