From 5cc49c82a4a7eff19c3734a3ba3e2aaed83e3bc2 Mon Sep 17 00:00:00 2001
From: Ritchie Vink <ritchie46@gmail.com>
Date: Wed, 15 Feb 2023 21:13:56 +0100
Subject: [PATCH] Added cast for FixedSizeBinary to (Large)Binary (#1403)

---
 src/compute/cast/binary_to.rs | 28 +++++++++++++++++++++++++++-
 src/compute/cast/mod.rs       | 17 ++++++++++++++++-
 tests/it/compute/cast.rs      | 32 ++++++++++++++++++++++++++++++++
 3 files changed, 75 insertions(+), 2 deletions(-)

diff --git a/src/compute/cast/binary_to.rs b/src/compute/cast/binary_to.rs
index 98cf4105b4b..d84c7dd1bd4 100644
--- a/src/compute/cast/binary_to.rs
+++ b/src/compute/cast/binary_to.rs
@@ -1,5 +1,5 @@
 use crate::error::Result;
-use crate::offset::Offset;
+use crate::offset::{Offset, Offsets};
 use crate::{array::*, datatypes::DataType, types::NativeType};
 
 use super::CastOptions;
@@ -118,3 +118,29 @@ pub(super) fn binary_to_dictionary_dyn<O: Offset, K: DictionaryKey>(
     let values = from.as_any().downcast_ref().unwrap();
     binary_to_dictionary::<O, K>(values).map(|x| Box::new(x) as Box<dyn Array>)
 }
+
+fn fixed_size_to_offsets<O: Offset>(values_len: usize, fixed_size: usize) -> Offsets<O> {
+    let offsets = (0..(values_len + 1))
+        .step_by(fixed_size)
+        .map(|v| O::from_usize(v).unwrap())
+        .collect();
+    // Safety
+    // * every element is `>= 0`
+    // * element at position `i` is >= than element at position `i-1`.
+    unsafe { Offsets::new_unchecked(offsets) }
+}
+
+/// Conversion of `FixedSizeBinary` to `Binary`.
+pub fn fixed_size_binary_binary<O: Offset>(
+    from: &FixedSizeBinaryArray,
+    to_data_type: DataType,
+) -> BinaryArray<O> {
+    let values = from.values().clone();
+    let offsets = fixed_size_to_offsets(values.len(), from.size());
+    BinaryArray::<O>::new(
+        to_data_type,
+        offsets.into(),
+        values,
+        from.validity().cloned(),
+    )
+}
diff --git a/src/compute/cast/mod.rs b/src/compute/cast/mod.rs
index 24f5689f38f..e42f769e7e5 100644
--- a/src/compute/cast/mod.rs
+++ b/src/compute/cast/mod.rs
@@ -132,7 +132,7 @@ pub fn can_cast_types(from_type: &DataType, to_type: &DataType) -> bool {
             is_numeric(to_type) || matches!(to_type, LargeBinary | Utf8 | LargeUtf8)
         }
         (LargeBinary, to_type) => is_numeric(to_type) || matches!(to_type, Binary | LargeUtf8),
-
+        (FixedSizeBinary(_), to_type) => matches!(to_type, Binary | LargeBinary),
         (Timestamp(_, _), Utf8) => true,
         (Timestamp(_, _), LargeUtf8) => true,
         (_, Utf8) => is_numeric(from_type) || from_type == &Binary,
@@ -686,6 +686,21 @@ pub fn cast(array: &dyn Array, to_type: &DataType, options: CastOptions) -> Resu
                 "Casting from {from_type:?} to {to_type:?} not supported",
             ))),
         },
+        (FixedSizeBinary(_), _) => match to_type {
+            Binary => Ok(fixed_size_binary_binary::<i32>(
+                array.as_any().downcast_ref().unwrap(),
+                to_type.clone(),
+            )
+            .boxed()),
+            LargeBinary => Ok(fixed_size_binary_binary::<i64>(
+                array.as_any().downcast_ref().unwrap(),
+                to_type.clone(),
+            )
+            .boxed()),
+            _ => Err(Error::NotYetImplemented(format!(
+                "Casting from {from_type:?} to {to_type:?} not supported",
+            ))),
+        },
 
         (_, Binary) => match from_type {
             UInt8 => primitive_to_binary_dyn::<u8, i32>(array),
diff --git a/tests/it/compute/cast.rs b/tests/it/compute/cast.rs
index 0631ba20baf..01cb31d2f24 100644
--- a/tests/it/compute/cast.rs
+++ b/tests/it/compute/cast.rs
@@ -227,6 +227,38 @@ fn binary_to_i32_partial() {
     assert_eq!(c, &expected);
 }
 
+#[test]
+fn fixed_size_binary_to_binary() {
+    let slice = [[0, 1], [2, 3]];
+    let array = FixedSizeBinaryArray::from_slice(slice);
+
+    // large-binary
+    let b = cast(
+        &array,
+        &DataType::LargeBinary,
+        CastOptions {
+            ..Default::default()
+        },
+    )
+    .unwrap();
+    let c = b.as_any().downcast_ref::<BinaryArray<i64>>().unwrap();
+    let expected = BinaryArray::<i64>::from_slice(slice);
+    assert_eq!(c, &expected);
+
+    // binary
+    let b = cast(
+        &array,
+        &DataType::Binary,
+        CastOptions {
+            ..Default::default()
+        },
+    )
+    .unwrap();
+    let c = b.as_any().downcast_ref::<BinaryArray<i32>>().unwrap();
+    let expected = BinaryArray::<i32>::from_slice(slice);
+    assert_eq!(c, &expected);
+}
+
 #[test]
 fn utf8_to_i32() {
     let array = Utf8Array::<i32>::from_slice(["5", "6", "seven", "8", "9.1"]);