From 720d858217f28485324a567e0946bcecdc90a583 Mon Sep 17 00:00:00 2001
From: Ritchie Vink <ritchie46@gmail.com>
Date: Wed, 8 Jun 2022 08:43:03 +0200
Subject: [PATCH 1/3] prevent unneeded offset check

---
 src/array/list/mutable.rs  | 48 ++++++++++++++++++++++++--------------
 src/io/avro/read/nested.rs | 32 +++++++++++++++----------
 2 files changed, 50 insertions(+), 30 deletions(-)

diff --git a/src/array/list/mutable.rs b/src/array/list/mutable.rs
index 6e5ffa2eb19..26bd29476a0 100644
--- a/src/array/list/mutable.rs
+++ b/src/array/list/mutable.rs
@@ -50,12 +50,16 @@ impl<O: Offset, M: MutableArray + Default> Default for MutableListArray<O, M> {
 
 impl<O: Offset, M: MutableArray> From<MutableListArray<O, M>> for ListArray<O> {
     fn from(mut other: MutableListArray<O, M>) -> Self {
-        ListArray::new(
-            other.data_type,
-            other.offsets.into(),
-            other.values.as_arc(),
-            other.validity.map(|x| x.into()),
-        )
+        // Safety:
+        // MutableListArray has monotonically increasing offsets
+        unsafe {
+            ListArray::new_unchecked(
+                other.data_type,
+                other.offsets.into(),
+                other.values.as_arc(),
+                other.validity.map(|x| x.into()),
+            )
+        }
     }
 }
 
@@ -209,21 +213,29 @@ impl<O: Offset, M: MutableArray + Default + 'static> MutableArray for MutableLis
     }
 
     fn as_box(&mut self) -> Box<dyn Array> {
-        Box::new(ListArray::new(
-            self.data_type.clone(),
-            std::mem::take(&mut self.offsets).into(),
-            self.values.as_arc(),
-            std::mem::take(&mut self.validity).map(|x| x.into()),
-        ))
+        // Safety:
+        // MutableListArray has monotonically increasing offsets
+        unsafe {
+            Box::new(ListArray::new_unchecked(
+                self.data_type.clone(),
+                std::mem::take(&mut self.offsets).into(),
+                self.values.as_arc(),
+                std::mem::take(&mut self.validity).map(|x| x.into()),
+            ))
+        }
     }
 
     fn as_arc(&mut self) -> Arc<dyn Array> {
-        Arc::new(ListArray::new(
-            self.data_type.clone(),
-            std::mem::take(&mut self.offsets).into(),
-            self.values.as_arc(),
-            std::mem::take(&mut self.validity).map(|x| x.into()),
-        ))
+        // Safety:
+        // MutableListArray has monotonically increasing offsets
+        unsafe {
+            Arc::new(ListArray::new_unchecked(
+                self.data_type.clone(),
+                std::mem::take(&mut self.offsets).into(),
+                self.values.as_arc(),
+                std::mem::take(&mut self.validity).map(|x| x.into()),
+            ))
+        }
     }
 
     fn data_type(&self) -> &DataType {
diff --git a/src/io/avro/read/nested.rs b/src/io/avro/read/nested.rs
index daa07fd911c..935e9ee2008 100644
--- a/src/io/avro/read/nested.rs
+++ b/src/io/avro/read/nested.rs
@@ -80,21 +80,29 @@ impl<O: Offset> MutableArray for DynMutableListArray<O> {
     }
 
     fn as_box(&mut self) -> Box<dyn Array> {
-        Box::new(ListArray::new(
-            self.data_type.clone(),
-            std::mem::take(&mut self.offsets).into(),
-            self.values.as_arc(),
-            std::mem::take(&mut self.validity).map(|x| x.into()),
-        ))
+        // Safety:
+        // MutableListArray has monotonically increasing offsets
+        unsafe {
+            Box::new(ListArray::new_unchecked(
+                self.data_type.clone(),
+                std::mem::take(&mut self.offsets).into(),
+                self.values.as_arc(),
+                std::mem::take(&mut self.validity).map(|x| x.into()),
+            ))
+        }
     }
 
     fn as_arc(&mut self) -> Arc<dyn Array> {
-        Arc::new(ListArray::new(
-            self.data_type.clone(),
-            std::mem::take(&mut self.offsets).into(),
-            self.values.as_arc(),
-            std::mem::take(&mut self.validity).map(|x| x.into()),
-        ))
+        // Safety:
+        // MutableListArray has monotonically increasing offsets
+        unsafe {
+            Arc::new(ListArray::new_unchecked(
+                self.data_type.clone(),
+                std::mem::take(&mut self.offsets).into(),
+                self.values.as_arc(),
+                std::mem::take(&mut self.validity).map(|x| x.into()),
+            ))
+        }
     }
 
     fn data_type(&self) -> &DataType {

From 6aca7c6c9046e9267308bde563ba10207d758303 Mon Sep 17 00:00:00 2001
From: Ritchie Vink <ritchie46@gmail.com>
Date: Wed, 8 Jun 2022 08:47:12 +0200
Subject: [PATCH 2/3] utf8 unneeded check

---
 src/io/json_integration/read/array.rs | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/src/io/json_integration/read/array.rs b/src/io/json_integration/read/array.rs
index 6210a02ca91..2f2c9cac307 100644
--- a/src/io/json_integration/read/array.rs
+++ b/src/io/json_integration/read/array.rs
@@ -185,7 +185,14 @@ fn to_utf8<O: Offset>(json_col: &ArrowJsonColumn, data_type: DataType) -> Arc<dy
         .iter()
         .flat_map(|value| value.as_str().unwrap().as_bytes().to_vec())
         .collect();
-    Arc::new(Utf8Array::new(data_type, offsets, values, validity))
+    // Safety:
+    // - we just created monotonically increasing offsets
+    // - we converted valid `&str` to bytes, so not needed to check utf8 again
+    unsafe {
+        Arc::new(Utf8Array::new_unchecked(
+            data_type, offsets, values, validity,
+        ))
+    }
 }
 
 fn to_list<O: Offset>(

From 49e5c82417002f0593d16a5e995ee5b590579925 Mon Sep 17 00:00:00 2001
From: Ritchie Vink <ritchie46@gmail.com>
Date: Wed, 8 Jun 2022 13:20:37 +0200
Subject: [PATCH 3/3] revert io to pass ci

---
 src/io/avro/read/nested.rs            | 32 ++++++++++-----------------
 src/io/json_integration/read/array.rs |  9 +-------
 2 files changed, 13 insertions(+), 28 deletions(-)

diff --git a/src/io/avro/read/nested.rs b/src/io/avro/read/nested.rs
index 935e9ee2008..daa07fd911c 100644
--- a/src/io/avro/read/nested.rs
+++ b/src/io/avro/read/nested.rs
@@ -80,29 +80,21 @@ impl<O: Offset> MutableArray for DynMutableListArray<O> {
     }
 
     fn as_box(&mut self) -> Box<dyn Array> {
-        // Safety:
-        // MutableListArray has monotonically increasing offsets
-        unsafe {
-            Box::new(ListArray::new_unchecked(
-                self.data_type.clone(),
-                std::mem::take(&mut self.offsets).into(),
-                self.values.as_arc(),
-                std::mem::take(&mut self.validity).map(|x| x.into()),
-            ))
-        }
+        Box::new(ListArray::new(
+            self.data_type.clone(),
+            std::mem::take(&mut self.offsets).into(),
+            self.values.as_arc(),
+            std::mem::take(&mut self.validity).map(|x| x.into()),
+        ))
     }
 
     fn as_arc(&mut self) -> Arc<dyn Array> {
-        // Safety:
-        // MutableListArray has monotonically increasing offsets
-        unsafe {
-            Arc::new(ListArray::new_unchecked(
-                self.data_type.clone(),
-                std::mem::take(&mut self.offsets).into(),
-                self.values.as_arc(),
-                std::mem::take(&mut self.validity).map(|x| x.into()),
-            ))
-        }
+        Arc::new(ListArray::new(
+            self.data_type.clone(),
+            std::mem::take(&mut self.offsets).into(),
+            self.values.as_arc(),
+            std::mem::take(&mut self.validity).map(|x| x.into()),
+        ))
     }
 
     fn data_type(&self) -> &DataType {
diff --git a/src/io/json_integration/read/array.rs b/src/io/json_integration/read/array.rs
index 2f2c9cac307..6210a02ca91 100644
--- a/src/io/json_integration/read/array.rs
+++ b/src/io/json_integration/read/array.rs
@@ -185,14 +185,7 @@ fn to_utf8<O: Offset>(json_col: &ArrowJsonColumn, data_type: DataType) -> Arc<dy
         .iter()
         .flat_map(|value| value.as_str().unwrap().as_bytes().to_vec())
         .collect();
-    // Safety:
-    // - we just created monotonically increasing offsets
-    // - we converted valid `&str` to bytes, so not needed to check utf8 again
-    unsafe {
-        Arc::new(Utf8Array::new_unchecked(
-            data_type, offsets, values, validity,
-        ))
-    }
+    Arc::new(Utf8Array::new(data_type, offsets, values, validity))
 }
 
 fn to_list<O: Offset>(