From f7b343e787aeb747c96c5c1aaf141260636c3337 Mon Sep 17 00:00:00 2001
From: NilsBarlaug <NilsBarlaug@users.noreply.github.com>
Date: Sat, 9 Oct 2021 05:29:32 +0200
Subject: [PATCH] Do not check offsets or utf8 validity in ffi (#510)

---
 src/array/binary/ffi.rs |  2 +-
 src/array/binary/mod.rs | 37 +++++++++++++++++++++++++++++++++++--
 src/array/utf8/ffi.rs   |  2 +-
 3 files changed, 37 insertions(+), 4 deletions(-)

diff --git a/src/array/binary/ffi.rs b/src/array/binary/ffi.rs
index 3030dedca31..4ca8d981ebd 100644
--- a/src/array/binary/ffi.rs
+++ b/src/array/binary/ffi.rs
@@ -44,7 +44,7 @@ impl<O: Offset, A: ffi::ArrowArrayRef> FromFfi<A> for BinaryArray<O> {
             validity = validity.map(|x| x.slice(offset, length))
         }
 
-        Ok(Self::from_data(
+        Ok(Self::from_data_unchecked(
             Self::default_data_type(),
             offsets,
             values,
diff --git a/src/array/binary/mod.rs b/src/array/binary/mod.rs
index 11b76c37367..036e3f74122 100644
--- a/src/array/binary/mod.rs
+++ b/src/array/binary/mod.rs
@@ -1,8 +1,8 @@
 use crate::{bitmap::Bitmap, buffer::Buffer, datatypes::DataType};
 
 use super::{
-    display_fmt, display_helper, specification::check_offsets, specification::Offset, Array,
-    GenericBinaryArray,
+    display_fmt, display_helper, specification::check_offsets,
+    specification::check_offsets_minimal, specification::Offset, Array, GenericBinaryArray,
 };
 
 mod ffi;
@@ -84,6 +84,39 @@ impl<O: Offset> BinaryArray<O> {
         }
     }
 
+    /// The same as [`BinaryArray::from_data`] but does not check for offsets.
+    /// # Safety
+    /// * `offsets` MUST be monotonically increasing
+    /// # Panics
+    /// This function panics iff:
+    /// * The `data_type`'s physical type is not consistent with the offset `O`.
+    /// * The last element of `offsets` is different from `values.len()`.
+    /// * The validity is not `None` and its length is different from `offsets.len() - 1`.
+    pub fn from_data_unchecked(
+        data_type: DataType,
+        offsets: Buffer<O>,
+        values: Buffer<u8>,
+        validity: Option<Bitmap>,
+    ) -> Self {
+        check_offsets_minimal(&offsets, values.len());
+
+        if let Some(validity) = &validity {
+            assert_eq!(offsets.len() - 1, validity.len());
+        }
+
+        if data_type.to_physical_type() != Self::default_data_type().to_physical_type() {
+            panic!("BinaryArray can only be initialized with DataType::Binary or DataType::LargeBinary")
+        }
+
+        Self {
+            data_type,
+            offsets,
+            values,
+            validity,
+            offset: 0,
+        }
+    }
+
     /// Creates a new [`BinaryArray`] by slicing this [`BinaryArray`].
     /// # Implementation
     /// This function is `O(1)`: all data will be shared between both arrays.
diff --git a/src/array/utf8/ffi.rs b/src/array/utf8/ffi.rs
index 98d47efa9dc..c9600d7d9d9 100644
--- a/src/array/utf8/ffi.rs
+++ b/src/array/utf8/ffi.rs
@@ -33,6 +33,6 @@ impl<O: Offset, A: ffi::ArrowArrayRef> FromFfi<A> for Utf8Array<O> {
             validity = validity.map(|x| x.slice(offset, length))
         }
         let data_type = Self::default_data_type();
-        Ok(Self::from_data(data_type, offsets, values, validity))
+        Ok(Self::from_data_unchecked(data_type, offsets, values, validity))
     }
 }