Added GrowableFixedSizeList and improved MutableFixedSizeListArray

src/array/fixed_size_list/mutable.rs

@@ -1,13 +1,10 @@
 use std::sync::Arc;
 
 use crate::{
-    array::{
-        Array, MutableArray, MutableBinaryArray, MutablePrimitiveArray, MutableUtf8Array, Offset,
-    },
+    array::{Array, MutableArray, TryExtend, TryPush},
     bitmap::MutableBitmap,
     datatypes::DataType,
     error::{ArrowError, Result},
-    types::NativeType,
 };
 
 use super::FixedSizeListArray;
@@ -32,6 +29,7 @@ impl<M: MutableArray> From<MutableFixedSizeListArray<M>> for FixedSizeListArray
 }
 
 impl<M: MutableArray> MutableFixedSizeListArray<M> {
+    /// Creates a new [`MutableFixedSizeListArray`] from a [`MutableArray`] and size.
     pub fn new(values: M, size: usize) -> Self {
         let data_type = FixedSizeListArray::default_datatype(values.data_type().clone(), size);
         assert_eq!(values.len(), 0);
@@ -43,20 +41,38 @@ impl<M: MutableArray> MutableFixedSizeListArray<M> {
         }
     }
 
-    pub fn mut_values(&mut self) -> &mut M {
-        &mut self.values
-    }
-
+    /// The inner values
     pub fn values(&self) -> &M {
         &self.values
     }
 
     fn init_validity(&mut self) {
-        self.validity = Some(MutableBitmap::from_trusted_len_iter(
-            std::iter::repeat(true)
-                .take(self.values.len() - 1)
-                .chain(std::iter::once(false)),
-        ))
+        let len = self.values.len() / self.size;
+
+        let mut validity = MutableBitmap::new();
+        validity.extend_constant(len, true);
+        validity.set(len - 1, false);
+        self.validity = Some(validity)
+    }
+
+    #[inline]
+    fn try_push_valid(&mut self) -> Result<()> {
+        if self.values.len() % self.size != 0 {
+            return Err(ArrowError::KeyOverflowError);
+        };
+        if let Some(validity) = &mut self.validity {
+            validity.push(true)
+        }
+        Ok(())
+    }
+
+    #[inline]
+    fn push_null(&mut self) {
+        (0..self.size).for_each(|_| self.values.push_null());
+        match &mut self.validity {
+            Some(validity) => validity.push(false),
+            None => self.init_validity(),
+        }
     }
 }
 
@@ -97,6 +113,7 @@ impl<M: MutableArray + 'static> MutableArray for MutableFixedSizeListArray<M> {
         self
     }
 
+    #[inline]
     fn push_null(&mut self) {
         (0..self.size).for_each(|_| {
             self.values.push_null();
@@ -109,69 +126,33 @@ impl<M: MutableArray + 'static> MutableArray for MutableFixedSizeListArray<M> {
     }
 }
 
-impl<T: NativeType> MutableFixedSizeListArray<MutablePrimitiveArray<T>> {
-    pub fn try_from_iter<P: IntoIterator<Item = Option<T>>, I: IntoIterator<Item = Option<P>>>(
-        iter: I,
-        size: usize,
-        data_type: DataType,
-    ) -> Result<Self> {
-        let iterator = iter.into_iter();
-        let (lower, _) = iterator.size_hint();
-        let array = MutablePrimitiveArray::<T>::with_capacity_from(lower * size, data_type);
-        let mut array = MutableFixedSizeListArray::new(array, size);
-        for items in iterator {
-            if let Some(items) = items {
-                let values = array.mut_values();
-                let len = values.len();
-                values.extend(items);
-                if values.len() - len != size {
-                    return Err(ArrowError::InvalidArgumentError(
-                        "A FixedSizeList must have all its values with the same size".to_string(),
-                    ));
-                };
-            } else {
-                array.push_null();
-            }
+impl<M, I, T> TryExtend<Option<I>> for MutableFixedSizeListArray<M>
+where
+    M: MutableArray + TryExtend<Option<T>>,
+    I: IntoIterator<Item = Option<T>>,
+{
+    #[inline]
+    fn try_extend<II: IntoIterator<Item = Option<I>>>(&mut self, iter: II) -> Result<()> {
+        for items in iter {
+            self.try_push(items)?;
         }
-        Ok(array)
+        Ok(())
     }
 }
 
-macro_rules! impl_offsets {
-    ($mutable:ident, $type:ty) => {
-        impl<O: Offset> MutableFixedSizeListArray<$mutable<O>> {
-            pub fn try_from_iter<
-                T: AsRef<$type>,
-                P: IntoIterator<Item = Option<T>>,
-                I: IntoIterator<Item = Option<P>>,
-            >(
-                iter: I,
-                size: usize,
-            ) -> Result<Self> {
-                let iterator = iter.into_iter();
-                let (lower, _) = iterator.size_hint();
-                let array = $mutable::<O>::with_capacity(lower * size);
-                let mut array = MutableFixedSizeListArray::new(array, size);
-                for items in iterator {
-                    if let Some(items) = items {
-                        let values = array.mut_values();
-                        let len = values.len();
-                        values.extend(items);
-                        if values.len() - len != size {
-                            return Err(ArrowError::InvalidArgumentError(
-                                "A FixedSizeList must have all its values with the same size"
-                                    .to_string(),
-                            ));
-                        };
-                    } else {
-                        array.push_null();
-                    }
-                }
-                Ok(array)
-            }
+impl<M, I, T> TryPush<Option<I>> for MutableFixedSizeListArray<M>
+where
+    M: MutableArray + TryExtend<Option<T>>,
+    I: IntoIterator<Item = Option<T>>,
+{
+    #[inline]
+    fn try_push(&mut self, item: Option<I>) -> Result<()> {
+        if let Some(items) = item {
+            self.values.try_extend(items)?;
+            self.try_push_valid()?;
+        } else {
+            self.push_null();
         }
-    };
+        Ok(())
+    }
 }
-
-impl_offsets!(MutableUtf8Array, str);
-impl_offsets!(MutableBinaryArray, [u8]);

src/array/growable/fixed_size_list.rs

@@ -0,0 +1,108 @@
+use std::sync::Arc;
+
+use crate::{
+    array::{Array, FixedSizeListArray},
+    bitmap::MutableBitmap,
+    datatypes::DataType,
+};
+
+use super::{
+    make_growable,
+    utils::{build_extend_null_bits, ExtendNullBits},
+    Growable,
+};
+
+/// Concrete [`Growable`] for the [`FixedSizeListArray`].
+pub struct GrowableFixedSizeList<'a> {
+    arrays: Vec<&'a FixedSizeListArray>,
+    validity: MutableBitmap,
+    values: Box<dyn Growable<'a> + 'a>,
+    extend_null_bits: Vec<ExtendNullBits<'a>>,
+    size: usize,
+}
+
+impl<'a> GrowableFixedSizeList<'a> {
+    /// Creates a new [`GrowableList`] bound to `arrays` with a pre-allocated `capacity`.
+    /// # Panics
+    /// If `arrays` is empty.
+    pub fn new(
+        arrays: Vec<&'a FixedSizeListArray>,
+        mut use_validity: bool,
+        capacity: usize,
+    ) -> Self {
+        assert!(!arrays.is_empty());
+
+        // if any of the arrays has nulls, insertions from any array requires setting bits
+        // as there is at least one array with nulls.
+        if !use_validity & arrays.iter().any(|array| array.null_count() > 0) {
+            use_validity = true;
+        };
+
+        let size =
+            if let DataType::FixedSizeList(_, size) = &arrays[0].data_type().to_logical_type() {
+                *size as usize
+            } else {
+                unreachable!("`GrowableFixedSizeList` expects `DataType::FixedSizeList`")
+            };
+
+        let extend_null_bits = arrays
+            .iter()
+            .map(|array| build_extend_null_bits(*array, use_validity))
+            .collect();
+
+        let inner = arrays
+            .iter()
+            .map(|array| array.values().as_ref())
+            .collect::<Vec<_>>();
+        let values = make_growable(&inner, use_validity, 0);
+
+        Self {
+            arrays,
+            values,
+            validity: MutableBitmap::with_capacity(capacity),
+            extend_null_bits,
+            size,
+        }
+    }
+
+    fn to(&mut self) -> FixedSizeListArray {
+        let validity = std::mem::take(&mut self.validity);
+        let values = self.values.as_arc();
+
+        FixedSizeListArray::from_data(self.arrays[0].data_type().clone(), values, validity.into())
+    }
+}
+
+impl<'a> Growable<'a> for GrowableFixedSizeList<'a> {
+    fn extend(&mut self, index: usize, start: usize, len: usize) {
+        (self.extend_null_bits[index])(&mut self.validity, start, len);
+        self.values
+            .extend(index, start * self.size, len * self.size);
+    }
+
+    fn extend_validity(&mut self, additional: usize) {
+        self.values.extend_validity(additional * self.size);
+        self.validity.extend_constant(additional, false);
+    }
+
+    fn as_arc(&mut self) -> Arc<dyn Array> {
+        Arc::new(self.to())
+    }
+
+    fn as_box(&mut self) -> Box<dyn Array> {
+        Box::new(self.to())
+    }
+}
+
+impl<'a> From<GrowableFixedSizeList<'a>> for FixedSizeListArray {
+    fn from(val: GrowableFixedSizeList<'a>) -> Self {
+        let mut values = val.values;
+        let values = values.as_arc();
+
+        Self::from_data(
+            val.arrays[0].data_type().clone(),
+            values,
+            val.validity.into(),
+        )
+    }
+}

src/array/growable/mod.rs

@@ -18,6 +18,8 @@ mod list;
 pub use list::GrowableList;
 mod structure;
 pub use structure::GrowableStruct;
+mod fixed_size_list;
+pub use fixed_size_list::GrowableFixedSizeList;
 mod utf8;
 pub use utf8::GrowableUtf8;
 mod dictionary;
@@ -201,7 +203,17 @@ pub fn make_growable<'a>(
                 capacity,
             ))
         }
-        FixedSizeList => todo!(),
+        FixedSizeList => {
+            let arrays = arrays
+                .iter()
+                .map(|array| array.as_any().downcast_ref().unwrap())
+                .collect::<Vec<_>>();
+            Box::new(fixed_size_list::GrowableFixedSizeList::new(
+                arrays,
+                use_validity,
+                capacity,
+            ))
+        }
         Union => todo!(),
         Dictionary(key_type) => {
             with_match_physical_dictionary_key_type!(key_type, |$T| {

tests/it/array/equal/fixed_size_list.rs

@@ -1,5 +1,5 @@
 use arrow2::{
-    array::{FixedSizeListArray, MutableFixedSizeListArray, MutablePrimitiveArray},
+    array::{FixedSizeListArray, MutableFixedSizeListArray, MutablePrimitiveArray, TryExtend},
     datatypes::DataType,
 };
 
@@ -16,9 +16,9 @@ fn create_fixed_size_list_array<U: AsRef<[i32]>, T: AsRef<[Option<U>]>>(
         })
     });
 
-    MutableFixedSizeListArray::<MutablePrimitiveArray<i32>>::try_from_iter(data, 3, DataType::Int32)
-        .unwrap()
-        .into()
+    let mut list = MutableFixedSizeListArray::new(MutablePrimitiveArray::<i32>::new(), 3);
+    list.try_extend(data).unwrap();
+    list.into()
 }
 
 #[test]

tests/it/array/fixed_size_list/mutable.rs

@@ -9,14 +9,9 @@ fn primitive() {
         Some(vec![Some(4), None, Some(6)]),
     ];
 
-    let list: FixedSizeListArray =
-        MutableFixedSizeListArray::<MutablePrimitiveArray<i32>>::try_from_iter(
-            data,
-            3,
-            DataType::Int32,
-        )
-        .unwrap()
-        .into();
+    let mut list = MutableFixedSizeListArray::new(MutablePrimitiveArray::<i32>::new(), 3);
+    list.try_extend(data).unwrap();
+    let list: FixedSizeListArray = list.into();
 
     let a = list.value(0);
     let a = a.as_any().downcast_ref::<Int32Array>().unwrap();