Added TryExtendFromSelf

src/array/binary/mod.rs

@@ -114,7 +114,7 @@ impl<O: Offset> BinaryArray<O> {
     /// Creates a new [`BinaryArray`] from a slice of optional `&[u8]`.
     // Note: this can't be `impl From` because Rust does not allow double `AsRef` on it.
     pub fn from<T: AsRef<[u8]>, P: AsRef<[Option<T>]>>(slice: P) -> Self {
-        Self::from_trusted_len_iter(slice.as_ref().iter().map(|x| x.as_ref()))
+        MutableBinaryArray::<O>::from(slice).into()
     }
 
     /// Returns an iterator of `Option<&[u8]>` over every element of this array.

src/array/binary/mutable.rs

@@ -1,21 +1,24 @@
 use std::{iter::FromIterator, sync::Arc};
 
 use crate::{
-    array::{Array, MutableArray, Offset, TryExtend, TryPush},
-    bitmap::{Bitmap, MutableBitmap},
+    array::{Array, MutableArray, Offset, TryExtend, TryExtendFromSelf, TryPush},
+    bitmap::{
+        utils::{BitmapIter, ZipValidity},
+        Bitmap, MutableBitmap,
+    },
     datatypes::DataType,
     error::{Error, Result},
     trusted_len::TrustedLen,
 };
 
-use super::{BinaryArray, MutableBinaryValuesArray};
+use super::{BinaryArray, MutableBinaryValuesArray, MutableBinaryValuesIter};
 use crate::array::physical_binary::*;
 
 /// The Arrow's equivalent to `Vec<Option<Vec<u8>>>`.
 /// Converting a [`MutableBinaryArray`] into a [`BinaryArray`] is `O(1)`.
 /// # Implementation
 /// This struct does not allocate a validity until one is required (i.e. push a null to it).
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 pub struct MutableBinaryArray<O: Offset> {
     values: MutableBinaryValuesArray<O>,
     validity: Option<MutableBitmap>,
@@ -96,6 +99,12 @@ impl<O: Offset> MutableBinaryArray<O> {
         Self { values, validity }
     }
 
+    /// Creates a new [`MutableBinaryArray`] from a slice of optional `&[u8]`.
+    // Note: this can't be `impl From` because Rust does not allow double `AsRef` on it.
+    pub fn from<T: AsRef<[u8]>, P: AsRef<[Option<T>]>>(slice: P) -> Self {
+        Self::from_trusted_len_iter(slice.as_ref().iter().map(|x| x.as_ref()))
+    }
+
     fn default_data_type() -> DataType {
         BinaryArray::<O>::default_data_type()
     }
@@ -193,6 +202,16 @@ impl<O: Offset> MutableBinaryArray<O> {
     pub fn offsets(&self) -> &Vec<O> {
         self.values.offsets()
     }
+
+    /// Returns an iterator of `Option<&[u8]>`
+    pub fn iter(&self) -> ZipValidity<&[u8], MutableBinaryValuesIter<O>, BitmapIter> {
+        ZipValidity::new(self.values_iter(), self.validity.as_ref().map(|x| x.iter()))
+    }
+
+    /// Returns an iterator over the values of this array
+    pub fn values_iter(&self) -> MutableBinaryValuesIter<O> {
+        self.values.iter()
+    }
 }
 
 impl<O: Offset> MutableArray for MutableBinaryArray<O> {
@@ -478,3 +497,17 @@ impl<O: Offset, T: AsRef<[u8]>> TryPush<Option<T>> for MutableBinaryArray<O> {
         Ok(())
     }
 }
+
+impl<O: Offset> PartialEq for MutableBinaryArray<O> {
+    fn eq(&self, other: &Self) -> bool {
+        self.iter().eq(other.iter())
+    }
+}
+
+impl<O: Offset> TryExtendFromSelf for MutableBinaryArray<O> {
+    fn try_extend_from_self(&mut self, other: &Self) -> Result<()> {
+        extend_validity(self.len(), &mut self.validity, &other.validity);
+
+        self.values.try_extend_from_self(&other.values)
+    }
+}

src/array/binary/mutable_values.rs

@@ -3,7 +3,8 @@ use std::{iter::FromIterator, sync::Arc};
 use crate::{
     array::{
         specification::{check_offsets_minimal, try_check_offsets},
-        Array, ArrayAccessor, ArrayValuesIter, MutableArray, Offset, TryExtend, TryPush,
+        Array, ArrayAccessor, ArrayValuesIter, MutableArray, Offset, TryExtend, TryExtendFromSelf,
+        TryPush,
     },
     bitmap::MutableBitmap,
     datatypes::DataType,
@@ -408,3 +409,10 @@ unsafe impl<'a, O: Offset> ArrayAccessor<'a> for MutableBinaryValuesArray<O> {
         self.len()
     }
 }
+
+impl<O: Offset> TryExtendFromSelf for MutableBinaryValuesArray<O> {
+    fn try_extend_from_self(&mut self, other: &Self) -> Result<()> {
+        self.values.extend_from_slice(&other.values);
+        try_extend_offsets(&mut self.offsets, &other.offsets)
+    }
+}

src/array/boolean/mutable.rs

@@ -2,7 +2,10 @@ use std::iter::FromIterator;
 use std::sync::Arc;
 
 use crate::{
-    array::{Array, MutableArray, TryExtend, TryPush},
+    array::{
+        physical_binary::extend_validity, Array, MutableArray, TryExtend, TryExtendFromSelf,
+        TryPush,
+    },
     bitmap::MutableBitmap,
     datatypes::{DataType, PhysicalType},
     error::Result,
@@ -15,7 +18,7 @@ use super::BooleanArray;
 /// Converting a [`MutableBooleanArray`] into a [`BooleanArray`] is `O(1)`.
 /// # Implementation
 /// This struct does not allocate a validity until one is required (i.e. push a null to it).
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 pub struct MutableBooleanArray {
     data_type: DataType,
     values: MutableBitmap,
@@ -533,3 +536,13 @@ impl PartialEq for MutableBooleanArray {
         self.iter().eq(other.iter())
     }
 }
+
+impl TryExtendFromSelf for MutableBooleanArray {
+    fn try_extend_from_self(&mut self, other: &Self) -> Result<()> {
+        extend_validity(self.len(), &mut self.validity, &other.validity);
+
+        let slice = other.values.as_slice();
+        self.values.extend_from_slice(slice, 0, other.values.len());
+        Ok(())
+    }
+}

src/array/fixed_size_binary/mod.rs

@@ -1,9 +1,4 @@
-use crate::{
-    bitmap::{Bitmap, MutableBitmap},
-    buffer::Buffer,
-    datatypes::DataType,
-    error::Error,
-};
+use crate::{bitmap::Bitmap, buffer::Buffer, datatypes::DataType, error::Error};
 
 use super::Array;
 
@@ -321,18 +316,7 @@ impl FixedSizeBinaryArray {
     /// Creates a new [`FixedSizeBinaryArray`] from a slice of optional `[u8]`.
     // Note: this can't be `impl From` because Rust does not allow double `AsRef` on it.
     pub fn from<const N: usize, P: AsRef<[Option<[u8; N]>]>>(slice: P) -> Self {
-        let values = slice
-            .as_ref()
-            .iter()
-            .copied()
-            .flat_map(|x| x.unwrap_or([0; N]))
-            .collect::<Vec<_>>();
-        let validity = slice
-            .as_ref()
-            .iter()
-            .map(|x| x.is_some())
-            .collect::<MutableBitmap>();
-        Self::new(DataType::FixedSizeBinary(N), values.into(), validity.into())
+        MutableFixedSizeBinaryArray::from(slice).into()
     }
 }
 

src/array/fixed_size_binary/mutable.rs

@@ -1,7 +1,7 @@
 use std::sync::Arc;
 
 use crate::{
-    array::{Array, MutableArray},
+    array::{physical_binary::extend_validity, Array, MutableArray, TryExtendFromSelf},
     bitmap::MutableBitmap,
     datatypes::DataType,
     error::{Error, Result},
@@ -13,7 +13,7 @@ use super::{FixedSizeBinaryArray, FixedSizeBinaryValues};
 /// Converting a [`MutableFixedSizeBinaryArray`] into a [`FixedSizeBinaryArray`] is `O(1)`.
 /// # Implementation
 /// This struct does not allocate a validity until one is required (i.e. push a null to it).
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 pub struct MutableFixedSizeBinaryArray {
     data_type: DataType,
     size: usize,
@@ -73,6 +73,23 @@ impl MutableFixedSizeBinaryArray {
         )
     }
 
+    /// Creates a new [`MutableFixedSizeBinaryArray`] from a slice of optional `[u8]`.
+    // Note: this can't be `impl From` because Rust does not allow double `AsRef` on it.
+    pub fn from<const N: usize, P: AsRef<[Option<[u8; N]>]>>(slice: P) -> Self {
+        let values = slice
+            .as_ref()
+            .iter()
+            .copied()
+            .flat_map(|x| x.unwrap_or([0; N]))
+            .collect::<Vec<_>>();
+        let validity = slice
+            .as_ref()
+            .iter()
+            .map(|x| x.is_some())
+            .collect::<MutableBitmap>();
+        Self::from_data(DataType::FixedSizeBinary(N), values, validity.into())
+    }
+
     /// tries to push a new entry to [`MutableFixedSizeBinaryArray`].
     /// # Error
     /// Errors iff the size of `value` is not equal to its own size.
@@ -278,3 +295,13 @@ impl PartialEq for MutableFixedSizeBinaryArray {
         self.iter().eq(other.iter())
     }
 }
+
+impl TryExtendFromSelf for MutableFixedSizeBinaryArray {
+    fn try_extend_from_self(&mut self, other: &Self) -> Result<()> {
+        extend_validity(self.len(), &mut self.validity, &other.validity);
+
+        let slice = other.values.as_slice();
+        self.values.extend_from_slice(slice);
+        Ok(())
+    }
+}

src/array/fixed_size_list/mutable.rs

@@ -1,7 +1,10 @@
 use std::sync::Arc;
 
 use crate::{
-    array::{Array, MutableArray, TryExtend, TryPush},
+    array::{
+        physical_binary::extend_validity, Array, MutableArray, TryExtend, TryExtendFromSelf,
+        TryPush,
+    },
     bitmap::MutableBitmap,
     datatypes::{DataType, Field},
     error::{Error, Result},
@@ -10,7 +13,7 @@ use crate::{
 use super::FixedSizeListArray;
 
 /// The mutable version of [`FixedSizeListArray`].
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 pub struct MutableFixedSizeListArray<M: MutableArray> {
     data_type: DataType,
     size: usize,
@@ -59,6 +62,11 @@ impl<M: MutableArray> MutableFixedSizeListArray<M> {
         }
     }
 
+    /// The length of this array
+    pub fn len(&self) -> usize {
+        self.values.len() / self.size
+    }
+
     /// The inner values
     pub fn values(&self) -> &M {
         &self.values
@@ -205,3 +213,14 @@ where
         Ok(())
     }
 }
+
+impl<M> TryExtendFromSelf for MutableFixedSizeListArray<M>
+where
+    M: MutableArray + TryExtendFromSelf,
+{
+    fn try_extend_from_self(&mut self, other: &Self) -> Result<()> {
+        extend_validity(self.len(), &mut self.validity, &other.validity);
+
+        self.values.try_extend_from_self(&other.values)
+    }
+}

src/array/list/iterator.rs

@@ -46,3 +46,27 @@ impl<'a, O: Offset> ListArray<O> {
         ListValuesIter::new(self)
     }
 }
+
+struct Iter<T, I: Iterator<Item = Option<T>>> {
+    current: i32,
+    offsets: std::vec::IntoIter<i32>,
+    values: I,
+}
+
+impl<T, I: Iterator<Item = Option<T>> + Clone> Iterator for Iter<T, I> {
+    type Item = Option<std::iter::Take<std::iter::Skip<I>>>;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        let next = self.offsets.next();
+        next.map(|next| {
+            let length = next - self.current;
+            let iter = self
+                .values
+                .clone()
+                .skip(self.current as usize)
+                .take(length as usize);
+            self.current = next;
+            Some(iter)
+        })
+    }
+}

src/array/list/mutable.rs

@@ -1,7 +1,10 @@
 use std::sync::Arc;
 
 use crate::{
-    array::{Array, MutableArray, Offset, TryExtend, TryPush},
+    array::{
+        physical_binary::{extend_validity, try_extend_offsets},
+        Array, MutableArray, Offset, TryExtend, TryExtendFromSelf, TryPush,
+    },
     bitmap::MutableBitmap,
     datatypes::{DataType, Field},
     error::{Error, Result},
@@ -10,7 +13,7 @@ use crate::{
 use super::ListArray;
 
 /// The mutable version of [`ListArray`].
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 pub struct MutableListArray<O: Offset, M: MutableArray> {
     data_type: DataType,
     offsets: Vec<O>,
@@ -70,6 +73,8 @@ where
     I: IntoIterator<Item = Option<T>>,
 {
     fn try_extend<II: IntoIterator<Item = Option<I>>>(&mut self, iter: II) -> Result<()> {
+        let iter = iter.into_iter();
+        self.reserve(iter.size_hint().0);
         for items in iter {
             self.try_push(items)?;
         }
@@ -96,6 +101,20 @@ where
     }
 }
 
+impl<O, M> TryExtendFromSelf for MutableListArray<O, M>
+where
+    O: Offset,
+    M: MutableArray + TryExtendFromSelf,
+{
+    fn try_extend_from_self(&mut self, other: &Self) -> Result<()> {
+        extend_validity(self.len(), &mut self.validity, &other.validity);
+
+        self.values.try_extend_from_self(&other.values)?;
+
+        try_extend_offsets(&mut self.offsets, &other.offsets)
+    }
+}
+
 impl<O: Offset, M: MutableArray> MutableListArray<O, M> {
     /// Creates a new [`MutableListArray`] from a [`MutableArray`] and capacity.
     pub fn new_from(values: M, data_type: DataType, capacity: usize) -> Self {
@@ -152,6 +171,12 @@ impl<O: Offset, M: MutableArray> MutableListArray<O, M> {
         }
     }
 
+    /// Returns the length of this array
+    #[inline]
+    pub fn len(&self) -> usize {
+        self.offsets.len() - 1
+    }
+
     /// The values
     pub fn mut_values(&mut self) -> &mut M {
         &mut self.values

src/array/mod.rs

@@ -417,6 +417,13 @@ pub trait TryPush<A> {
     fn try_push(&mut self, item: A) -> Result<()>;
 }
 
+/// A trait describing the ability of a struct to extend from a reference of itself.
+/// Specialization of [`TryExtend`].
+pub trait TryExtendFromSelf {
+    /// Tries to extend itself with elements from `other`, failing only on overflow.
+    fn try_extend_from_self(&mut self, other: &Self) -> Result<()>;
+}
+
 /// Trait that [`BinaryArray`] and [`Utf8Array`] implement for the purposes of DRY.
 /// # Safety
 /// The implementer must ensure that