Changed methods to slice arrays (jorgecarleitao#1396)

ritchie46 · Mar 29, 2023 · 6137e16 · 6137e16
1 parent ee1df36
commit 6137e16
Show file tree

Hide file tree

Showing 71 changed files with 480 additions and 437 deletions.
diff --git a/benches/bitmap_ops.rs b/benches/bitmap_ops.rs
@@ -26,7 +26,8 @@ fn add_benchmark(c: &mut Criterion) {
             &format!("bitmap count zeros 85% slice 2^{log2_size}"),
             |b| {
                 b.iter(|| {
-                    let r = bitmap.clone().slice(offset, len);
+                    let mut r = bitmap.clone();
+                    r.slice(offset, len);
                     assert!(r.unset_bits() > 0);
                 })
             },
@@ -39,13 +40,15 @@ fn add_benchmark(c: &mut Criterion) {
             &format!("bitmap count zeros 51% slice 2^{log2_size}"),
             |b| {
                 b.iter(|| {
-                    let r = bitmap.clone().slice(offset, len);
+                    let mut r = bitmap.clone();
+                    r.slice(offset, len);
                     assert!(r.unset_bits() > 0);
                 })
             },
         );
 
-        let bitmap1 = bitmap.clone().slice(1, size - 1);
+        let mut bitmap1 = bitmap.clone();
+        bitmap1.slice(1, size - 1);
         c.bench_function(&format!("bitmap not 2^{log2_size}"), |b| {
             b.iter(|| {
                 let r = !&bitmap1;

diff --git a/guide/src/low_level.md b/guide/src/low_level.md
@@ -34,7 +34,7 @@ let x = vec![1u32, 2, 3];
 let x: Buffer<u32> = x.into();
 assert_eq!(x.as_slice(), &[1u32, 2, 3]);
 
-let x = x.slice(1, 2); // O(1)
+let x = x.sliced(1, 2); // O(1)
 assert_eq!(x.as_slice(), &[2, 3]);
 # }
 ```

diff --git a/src/array/binary/mod.rs b/src/array/binary/mod.rs
@@ -175,41 +175,34 @@ impl<O: Offset> BinaryArray<O> {
         self.validity.as_ref()
     }
 
-    /// Creates a new [`BinaryArray`] by slicing this [`BinaryArray`].
+    /// Slices this [`BinaryArray`].
     /// # Implementation
-    /// This function is `O(1)`: all data will be shared between both arrays.
+    /// This function is `O(1)`.
     /// # Panics
     /// iff `offset + length > self.len()`.
-    #[must_use]
-    pub fn slice(&self, offset: usize, length: usize) -> Self {
+    pub fn slice(&mut self, offset: usize, length: usize) {
         assert!(
             offset + length <= self.len(),
             "the offset of the new Buffer cannot exceed the existing length"
         );
         unsafe { self.slice_unchecked(offset, length) }
     }
 
-    /// Creates a new [`BinaryArray`] by slicing this [`BinaryArray`].
+    /// Slices this [`BinaryArray`].
     /// # Implementation
-    /// This function is `O(1)`: all data will be shared between both arrays.
+    /// This function is `O(1)`.
     /// # Safety
     /// The caller must ensure that `offset + length <= self.len()`.
-    #[must_use]
-    pub unsafe fn slice_unchecked(&self, offset: usize, length: usize) -> Self {
-        let validity = self
-            .validity
-            .clone()
-            .map(|bitmap| bitmap.slice_unchecked(offset, length))
-            .and_then(|bitmap| (bitmap.unset_bits() > 0).then(|| bitmap));
-        let offsets = self.offsets.clone().slice_unchecked(offset, length + 1);
-        Self {
-            data_type: self.data_type.clone(),
-            offsets,
-            values: self.values.clone(),
-            validity,
-        }
+    pub unsafe fn slice_unchecked(&mut self, offset: usize, length: usize) {
+        self.validity.as_mut().and_then(|bitmap| {
+            bitmap.slice_unchecked(offset, length);
+            (bitmap.unset_bits() > 0).then(|| bitmap)
+        });
+        self.offsets.slice_unchecked(offset, length + 1);
     }
 
+    impl_sliced!();
+
     /// Boxes self into a [`Box<dyn Array>`].
     pub fn boxed(self) -> Box<dyn Array> {
         Box::new(self)
@@ -440,12 +433,14 @@ impl<O: Offset> Array for BinaryArray<O> {
         self.validity.as_ref()
     }
 
-    fn slice(&self, offset: usize, length: usize) -> Box<dyn Array> {
-        Box::new(self.slice(offset, length))
+    fn slice(&mut self, offset: usize, length: usize) {
+        self.slice(offset, length)
     }
-    unsafe fn slice_unchecked(&self, offset: usize, length: usize) -> Box<dyn Array> {
-        Box::new(self.slice_unchecked(offset, length))
+
+    unsafe fn slice_unchecked(&mut self, offset: usize, length: usize) {
+        self.slice_unchecked(offset, length)
     }
+
     fn with_validity(&self, validity: Option<Bitmap>) -> Box<dyn Array> {
         Box::new(self.clone().with_validity(validity))
     }

diff --git a/src/array/boolean/mod.rs b/src/array/boolean/mod.rs
@@ -143,41 +143,36 @@ impl BooleanArray {
         self.values.get_bit_unchecked(i)
     }
 
-    /// Returns a slice of this [`BooleanArray`].
+    /// Slices this [`BooleanArray`].
     /// # Implementation
     /// This operation is `O(1)` as it amounts to increase up to two ref counts.
     /// # Panic
     /// This function panics iff `offset + length > self.len()`.
     #[inline]
-    #[must_use]
-    pub fn slice(&self, offset: usize, length: usize) -> Self {
+    pub fn slice(&mut self, offset: usize, length: usize) {
         assert!(
             offset + length <= self.len(),
             "the offset of the new Buffer cannot exceed the existing length"
         );
         unsafe { self.slice_unchecked(offset, length) }
     }
 
-    /// Returns a slice of this [`BooleanArray`].
+    /// Slices this [`BooleanArray`].
     /// # Implementation
     /// This operation is `O(1)` as it amounts to increase two ref counts.
     /// # Safety
     /// The caller must ensure that `offset + length <= self.len()`.
     #[inline]
-    #[must_use]
-    pub unsafe fn slice_unchecked(&self, offset: usize, length: usize) -> Self {
-        let validity = self
-            .validity
-            .clone()
-            .map(|bitmap| bitmap.slice_unchecked(offset, length))
-            .and_then(|bitmap| (bitmap.unset_bits() > 0).then(|| bitmap));
-        Self {
-            data_type: self.data_type.clone(),
-            values: self.values.clone().slice_unchecked(offset, length),
-            validity,
-        }
+    pub unsafe fn slice_unchecked(&mut self, offset: usize, length: usize) {
+        self.validity.as_mut().and_then(|bitmap| {
+            bitmap.slice_unchecked(offset, length);
+            (bitmap.unset_bits() > 0).then(|| bitmap)
+        });
+        self.values.slice_unchecked(offset, length);
     }
 
+    impl_sliced!();
+
     /// Returns this [`BooleanArray`] with a new validity.
     /// # Panic
     /// This function panics iff `validity.len() != self.len()`.
@@ -400,13 +395,15 @@ impl Array for BooleanArray {
     }
 
     #[inline]
-    fn slice(&self, offset: usize, length: usize) -> Box<dyn Array> {
-        Box::new(self.slice(offset, length))
+    fn slice(&mut self, offset: usize, length: usize) {
+        self.slice(offset, length)
     }
+
     #[inline]
-    unsafe fn slice_unchecked(&self, offset: usize, length: usize) -> Box<dyn Array> {
-        Box::new(self.slice_unchecked(offset, length))
+    unsafe fn slice_unchecked(&mut self, offset: usize, length: usize) {
+        self.slice_unchecked(offset, length)
     }
+
     fn with_validity(&self, validity: Option<Bitmap>) -> Box<dyn Array> {
         Box::new(self.clone().with_validity(validity))
     }

diff --git a/src/array/dictionary/mod.rs b/src/array/dictionary/mod.rs
@@ -291,28 +291,22 @@ impl<K: DictionaryKey> DictionaryArray<K> {
         DataType::Dictionary(K::KEY_TYPE, Box::new(values_datatype), false)
     }
 
-    /// Creates a new [`DictionaryArray`] by slicing the existing [`DictionaryArray`].
+    /// Slices this [`DictionaryArray`].
     /// # Panics
     /// iff `offset + length > self.len()`.
-    pub fn slice(&self, offset: usize, length: usize) -> Self {
-        Self {
-            data_type: self.data_type.clone(),
-            keys: self.keys.clone().slice(offset, length),
-            values: self.values.clone(),
-        }
+    pub fn slice(&mut self, offset: usize, length: usize) {
+        self.keys.slice(offset, length);
     }
 
-    /// Creates a new [`DictionaryArray`] by slicing the existing [`DictionaryArray`].
+    /// Slices this [`DictionaryArray`].
     /// # Safety
     /// Safe iff `offset + length <= self.len()`.
-    pub unsafe fn slice_unchecked(&self, offset: usize, length: usize) -> Self {
-        Self {
-            data_type: self.data_type.clone(),
-            keys: self.keys.clone().slice_unchecked(offset, length),
-            values: self.values.clone(),
-        }
+    pub unsafe fn slice_unchecked(&mut self, offset: usize, length: usize) {
+        self.keys.slice_unchecked(offset, length);
     }
 
+    impl_sliced!();
+
     /// Returns this [`DictionaryArray`] with a new validity.
     /// # Panic
     /// This function panics iff `validity.len() != self.len()`.
@@ -437,12 +431,12 @@ impl<K: DictionaryKey> Array for DictionaryArray<K> {
         self.keys.validity()
     }
 
-    fn slice(&self, offset: usize, length: usize) -> Box<dyn Array> {
-        Box::new(self.slice(offset, length))
+    fn slice(&mut self, offset: usize, length: usize) {
+        self.slice(offset, length)
     }
 
-    unsafe fn slice_unchecked(&self, offset: usize, length: usize) -> Box<dyn Array> {
-        Box::new(self.slice_unchecked(offset, length))
+    unsafe fn slice_unchecked(&mut self, offset: usize, length: usize) {
+        self.slice_unchecked(offset, length)
     }
 
     fn with_validity(&self, validity: Option<Bitmap>) -> Box<dyn Array> {

diff --git a/src/array/equal/struct_.rs b/src/array/equal/struct_.rs
@@ -13,7 +13,7 @@ pub(super) fn equal(lhs: &StructArray, rhs: &StructArray) -> bool {
                     l_validity.iter().zip(r_validity.iter()).enumerate().all(
                         |(i, (lhs_is_valid, rhs_is_valid))| {
                             if lhs_is_valid && rhs_is_valid {
-                                lhs.slice(i, 1) == rhs.slice(i, 1)
+                                lhs.sliced(i, 1) == rhs.sliced(i, 1)
                             } else {
                                 lhs_is_valid == rhs_is_valid
                             }
@@ -27,7 +27,7 @@ pub(super) fn equal(lhs: &StructArray, rhs: &StructArray) -> bool {
                     .all(|(lhs, rhs)| {
                         l_validity.iter().enumerate().all(|(i, lhs_is_valid)| {
                             if lhs_is_valid {
-                                lhs.slice(i, 1) == rhs.slice(i, 1)
+                                lhs.sliced(i, 1) == rhs.sliced(i, 1)
                             } else {
                                 // rhs is always valid => different
                                 false
@@ -42,7 +42,7 @@ pub(super) fn equal(lhs: &StructArray, rhs: &StructArray) -> bool {
                     .all(|(lhs, rhs)| {
                         r_validity.iter().enumerate().all(|(i, rhs_is_valid)| {
                             if rhs_is_valid {
-                                lhs.slice(i, 1) == rhs.slice(i, 1)
+                                lhs.sliced(i, 1) == rhs.sliced(i, 1)
                             } else {
                                 // lhs is always valid => different
                                 false

diff --git a/src/array/fixed_size_binary/mod.rs b/src/array/fixed_size_binary/mod.rs
@@ -97,44 +97,35 @@ impl FixedSizeBinaryArray {
 
 // must use
 impl FixedSizeBinaryArray {
-    /// Returns a slice of this [`FixedSizeBinaryArray`].
+    /// Slices this [`FixedSizeBinaryArray`].
     /// # Implementation
-    /// This operation is `O(1)` as it amounts to increase 3 ref counts.
+    /// This operation is `O(1)`.
     /// # Panics
     /// panics iff `offset + length > self.len()`
-    #[must_use]
-    pub fn slice(&self, offset: usize, length: usize) -> Self {
+    pub fn slice(&mut self, offset: usize, length: usize) {
         assert!(
             offset + length <= self.len(),
             "the offset of the new Buffer cannot exceed the existing length"
         );
         unsafe { self.slice_unchecked(offset, length) }
     }
 
-    /// Returns a slice of this [`FixedSizeBinaryArray`].
+    /// Slices this [`FixedSizeBinaryArray`].
     /// # Implementation
-    /// This operation is `O(1)` as it amounts to increase 3 ref counts.
+    /// This operation is `O(1)`.
     /// # Safety
     /// The caller must ensure that `offset + length <= self.len()`.
-    #[must_use]
-    pub unsafe fn slice_unchecked(&self, offset: usize, length: usize) -> Self {
-        let validity = self
-            .validity
-            .clone()
-            .map(|bitmap| bitmap.slice_unchecked(offset, length))
-            .and_then(|bitmap| (bitmap.unset_bits() > 0).then(|| bitmap));
-        let values = self
-            .values
-            .clone()
+    pub unsafe fn slice_unchecked(&mut self, offset: usize, length: usize) {
+        self.validity.as_mut().and_then(|bitmap| {
+            bitmap.slice_unchecked(offset, length);
+            (bitmap.unset_bits() > 0).then(|| bitmap)
+        });
+        self.values
             .slice_unchecked(offset * self.size, length * self.size);
-        Self {
-            data_type: self.data_type.clone(),
-            size: self.size,
-            values,
-            validity,
-        }
     }
 
+    impl_sliced!();
+
     /// Returns this [`FixedSizeBinaryArray`] with a new validity.
     /// # Panic
     /// This function panics iff `validity.len() != self.len()`.
@@ -267,12 +258,12 @@ impl Array for FixedSizeBinaryArray {
         self.validity.as_ref()
     }
 
-    fn slice(&self, offset: usize, length: usize) -> Box<dyn Array> {
-        Box::new(self.slice(offset, length))
+    fn slice(&mut self, offset: usize, length: usize) {
+        self.slice(offset, length)
     }
 
-    unsafe fn slice_unchecked(&self, offset: usize, length: usize) -> Box<dyn Array> {
-        Box::new(self.slice_unchecked(offset, length))
+    unsafe fn slice_unchecked(&mut self, offset: usize, length: usize) {
+        self.slice_unchecked(offset, length)
     }
 
     fn with_validity(&self, validity: Option<Bitmap>) -> Box<dyn Array> {