From 3c61372d7ac76bba149316b1fbad6e981752e502 Mon Sep 17 00:00:00 2001
From: Jk Xu <54522439+Dousir9@users.noreply.github.com>
Date: Wed, 18 Oct 2023 15:26:47 +0800
Subject: [PATCH 1/2] fix bitmap new_trued (#1580)

---
 src/bitmap/immutable.rs      |  2 +-
 tests/it/bitmap/immutable.rs | 19 +++++++++++++++++++
 2 files changed, 20 insertions(+), 1 deletion(-)

diff --git a/src/bitmap/immutable.rs b/src/bitmap/immutable.rs
index 57502e911e..6883d3312f 100644
--- a/src/bitmap/immutable.rs
+++ b/src/bitmap/immutable.rs
@@ -307,7 +307,7 @@ impl Bitmap {
         // just set each byte to u8::MAX
         // we will not access data with index >= length
         let bytes = vec![0b11111111u8; length.saturating_add(7) / 8];
-        unsafe { Bitmap::from_inner_unchecked(Arc::new(bytes.into()), 0, length, length) }
+        unsafe { Bitmap::from_inner_unchecked(Arc::new(bytes.into()), 0, length, 0) }
     }
 
     /// Counts the nulls (unset bits) starting from `offset` bits and for `length` bits.
diff --git a/tests/it/bitmap/immutable.rs b/tests/it/bitmap/immutable.rs
index 5e6157413e..cc003009e0 100644
--- a/tests/it/bitmap/immutable.rs
+++ b/tests/it/bitmap/immutable.rs
@@ -32,6 +32,25 @@ fn as_slice_offset_middle() {
     assert_eq!(length, 5);
 }
 
+#[test]
+fn new_constant() {
+    let b = Bitmap::new_constant(true, 9);
+    let (slice, offset, length) = b.as_slice();
+    assert_eq!(slice[0], 0b11111111);
+    assert!((slice[1] & 0b00000001) > 0);
+    assert_eq!(offset, 0);
+    assert_eq!(length, 9);
+    assert_eq!(b.unset_bits(), 0);
+
+    let b = Bitmap::new_constant(false, 9);
+    let (slice, offset, length) = b.as_slice();
+    assert_eq!(slice[0], 0b00000000);
+    assert!((slice[1] & 0b00000001) == 0);
+    assert_eq!(offset, 0);
+    assert_eq!(length, 9);
+    assert_eq!(b.unset_bits(), 9);
+}
+
 #[test]
 fn debug() {
     let b = Bitmap::from([true, true, false, true, true, true, true, true, true]);

From 9a26422d00b83c65245f75e02eb436dedd91b5b8 Mon Sep 17 00:00:00 2001
From: sundyli <543950155@qq.com>
Date: Thu, 19 Oct 2023 17:19:15 -0700
Subject: [PATCH 2/2] chore: add max bytes_estimate to reserve the capacity of
 binary (#1581)

---
 src/io/parquet/read/deserialize/binary/utils.rs | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/src/io/parquet/read/deserialize/binary/utils.rs b/src/io/parquet/read/deserialize/binary/utils.rs
index ec514766fa..a48063c56e 100644
--- a/src/io/parquet/read/deserialize/binary/utils.rs
+++ b/src/io/parquet/read/deserialize/binary/utils.rs
@@ -48,7 +48,8 @@ impl<O: Offset> Binary<O> {
         if self.offsets.len_proxy() == 100 && self.offsets.capacity() > 100 {
             let bytes_per_row = self.values.len() / 100 + 1;
             let bytes_estimate = bytes_per_row * self.offsets.capacity();
-            if bytes_estimate > self.values.capacity() {
+
+            if bytes_estimate > self.values.capacity() && bytes_estimate < 10 * 1024 * 1024 {
                 self.values.reserve(bytes_estimate - self.values.capacity());
             }
         }