Support variant_to_arrow for utf8 (#8600)

# Which issue does this PR close?

- Closes #8567.

# Rationale for this change

Add support for Variant::Utf-8, LargeUtf8, Utf8View. This needs to add a
new builder VariantToStringArrowRowBuilder, because LargeUtf8, Utf8View
are not ArrowPritimitiveType's

# What changes are included in this PR?

- Added support for Variant::Utf-8, LargeUtf8, Utf8View by adding a new
enum and builder for utf8 and largeUtf8 and added utf8view to primitive
builder.
- Added a new variable `data_capacity` to
`make_string_variant_to_arrow_row_builder` to support string types.
- Updated the `make_string_variant_to_arrow_row_builder` in
`variant_get` to include the variable.

# Are these changes tested?

Added a variant_get test for utf8 type and created two separate tests
for largeUtf8 and Utf8view because these types can't be shredded.

# Are there any user-facing changes?

No

---------

Co-authored-by: Congxian Qiu <qcx978132955@gmail.com>
Co-authored-by: Ryan Johnson <scovich@users.noreply.github.com>
Co-authored-by: Andrew Lamb <andrew@nerdnetworks.org>

Author: 4 people

arrow-array/src/builder/generic_bytes_builder.rs

@@ -348,6 +348,50 @@ impl<O: OffsetSizeTrait> std::fmt::Write for GenericStringBuilder<O> {
     }
 }
 
+/// A byte size value representing the number of bytes to allocate per string in [`GenericStringBuilder`]
+///
+/// To create a [`GenericStringBuilder`] using `.with_capacity` we are required to provide: \
+/// - `item_capacity` - the row count \
+/// - `data_capacity` - total string byte count \
+///
+/// We will use the `AVERAGE_STRING_LENGTH` * row_count for `data_capacity`. \
+///
+/// These capacities are preallocation hints used to improve performance,
+/// but consuquences of passing a hint too large or too small should be negligible.
+const AVERAGE_STRING_LENGTH: usize = 16;
+/// Trait for string-like array builders
+///
+/// This trait provides unified interface for builders that append string-like data
+/// such as [`GenericStringBuilder<O>`] and [`crate::builder::StringViewBuilder`]
+pub trait StringLikeArrayBuilder: ArrayBuilder {
+    /// Returns a human-readable type name for the builder.
+    fn type_name() -> &'static str;
+
+    /// Creates a new builder with the given row capacity.
+    fn with_capacity(capacity: usize) -> Self;
+
+    /// Appends a non-null string value to the builder.
+    fn append_value(&mut self, value: &str);
+
+    /// Appends a null value to the builder.
+    fn append_null(&mut self);
+}
+
+impl<O: OffsetSizeTrait> StringLikeArrayBuilder for GenericStringBuilder<O> {
+    fn type_name() -> &'static str {
+        std::any::type_name::<Self>()
+    }
+    fn with_capacity(capacity: usize) -> Self {
+        Self::with_capacity(capacity, capacity * AVERAGE_STRING_LENGTH)
+    }
+    fn append_value(&mut self, value: &str) {
+        Self::append_value(self, value);
+    }
+    fn append_null(&mut self) {
+        Self::append_null(self);
+    }
+}
+
 ///  Array builder for [`GenericBinaryArray`][crate::GenericBinaryArray]
 ///
 /// Values can be appended using [`GenericByteBuilder::append_value`], and nulls with

arrow-array/src/builder/generic_bytes_view_builder.rs

@@ -25,7 +25,7 @@ use arrow_schema::ArrowError;
 use hashbrown::HashTable;
 use hashbrown::hash_table::Entry;
 
-use crate::builder::ArrayBuilder;
+use crate::builder::{ArrayBuilder, StringLikeArrayBuilder};
 use crate::types::bytes::ByteArrayNativeType;
 use crate::types::{BinaryViewType, ByteViewType, StringViewType};
 use crate::{Array, ArrayRef, GenericByteViewArray};
@@ -533,6 +533,21 @@ impl<T: ByteViewType + ?Sized, V: AsRef<T::Native>> Extend<Option<V>>
 /// ```
 pub type StringViewBuilder = GenericByteViewBuilder<StringViewType>;
 
+impl StringLikeArrayBuilder for StringViewBuilder {
+    fn type_name() -> &'static str {
+        std::any::type_name::<StringViewBuilder>()
+    }
+    fn with_capacity(capacity: usize) -> Self {
+        Self::with_capacity(capacity)
+    }
+    fn append_value(&mut self, value: &str) {
+        Self::append_value(self, value);
+    }
+    fn append_null(&mut self) {
+        Self::append_null(self);
+    }
+}
+
 ///  Array builder for [`BinaryViewArray`][crate::BinaryViewArray]
 ///
 /// Values can be appended using [`GenericByteViewBuilder::append_value`], and nulls with

parquet-variant-compute/src/variant_array.rs

@@ -967,6 +967,16 @@ fn typed_value_to_variant<'a>(
             let value = array.value(index);
             Ok(Variant::from(value))
         }
+        DataType::LargeUtf8 => {
+            let array = typed_value.as_string::<i64>();
+            let value = array.value(index);
+            Ok(Variant::from(value))
+        }
+        DataType::Utf8View => {
+            let array = typed_value.as_string_view();
+            let value = array.value(index);
+            Ok(Variant::from(value))
+        }
         DataType::Int8 => {
             primitive_conversion_single_value!(Int8Type, typed_value, index)
         }
@@ -1165,14 +1175,14 @@ fn canonicalize_and_verify_data_type(data_type: &DataType) -> Result<Cow<'_, Dat
         // Binary and string are allowed. Force Binary to BinaryView because that's what the parquet
         // reader returns and what the rest of the variant code expects.
         Binary => Cow::Owned(DataType::BinaryView),
-        BinaryView | Utf8 => borrow!(),
+        BinaryView | Utf8 | LargeUtf8 | Utf8View => borrow!(),
 
         // UUID maps to 16-byte fixed-size binary; no other width is allowed
         FixedSizeBinary(16) => borrow!(),
         FixedSizeBinary(_) | FixedSizeList(..) => fail!(),
 
         // We can _possibly_ allow (some of) these some day?
-        LargeBinary | LargeUtf8 | Utf8View | ListView(_) | LargeList(_) | LargeListView(_) => {
+        LargeBinary | ListView(_) | LargeList(_) | LargeListView(_) => {
             fail!()
         }
 

parquet-variant-compute/src/variant_get.rs

@@ -311,8 +311,8 @@ mod test {
     use arrow::array::{
         Array, ArrayRef, AsArray, BinaryViewArray, BooleanArray, Date32Array, Decimal32Array,
         Decimal64Array, Decimal128Array, Decimal256Array, Float32Array, Float64Array, Int8Array,
-        Int16Array, Int32Array, Int64Array, NullBuilder, StringArray, StructArray,
-        Time64MicrosecondArray,
+        Int16Array, Int32Array, Int64Array, LargeStringArray, NullBuilder, StringArray,
+        StringViewArray, StructArray, Time64MicrosecondArray,
     };
     use arrow::buffer::NullBuffer;
     use arrow::compute::CastOptions;
@@ -778,6 +778,27 @@ mod test {
         BooleanArray::from(vec![Some(true), Some(false), Some(true)])
     );
 
+    perfectly_shredded_to_arrow_primitive_test!(
+        get_variant_perfectly_shredded_utf8_as_utf8,
+        DataType::Utf8,
+        perfectly_shredded_utf8_variant_array,
+        StringArray::from(vec![Some("foo"), Some("bar"), Some("baz")])
+    );
+
+    perfectly_shredded_to_arrow_primitive_test!(
+        get_variant_perfectly_shredded_large_utf8_as_utf8,
+        DataType::Utf8,
+        perfectly_shredded_large_utf8_variant_array,
+        StringArray::from(vec![Some("foo"), Some("bar"), Some("baz")])
+    );
+
+    perfectly_shredded_to_arrow_primitive_test!(
+        get_variant_perfectly_shredded_utf8_view_as_utf8,
+        DataType::Utf8,
+        perfectly_shredded_utf8_view_variant_array,
+        StringArray::from(vec![Some("foo"), Some("bar"), Some("baz")])
+    );
+
     macro_rules! perfectly_shredded_variant_array_fn {
         ($func:ident, $typed_value_gen:expr) => {
             fn $func() -> ArrayRef {
@@ -801,6 +822,18 @@ mod test {
         };
     }
 
+    perfectly_shredded_variant_array_fn!(perfectly_shredded_utf8_variant_array, || {
+        StringArray::from(vec![Some("foo"), Some("bar"), Some("baz")])
+    });
+
+    perfectly_shredded_variant_array_fn!(perfectly_shredded_large_utf8_variant_array, || {
+        LargeStringArray::from(vec![Some("foo"), Some("bar"), Some("baz")])
+    });
+
+    perfectly_shredded_variant_array_fn!(perfectly_shredded_utf8_view_variant_array, || {
+        StringViewArray::from(vec![Some("foo"), Some("bar"), Some("baz")])
+    });
+
     perfectly_shredded_variant_array_fn!(perfectly_shredded_bool_variant_array, || {
         BooleanArray::from(vec![Some(true), Some(false), Some(true)])
     });

parquet-variant-compute/src/variant_to_arrow.rs

@@ -16,8 +16,9 @@
 // under the License.
 
 use arrow::array::{
-    ArrayRef, BinaryViewArray, BooleanBuilder, FixedSizeBinaryBuilder, NullArray,
-    NullBufferBuilder, PrimitiveBuilder,
+    ArrayRef, BinaryViewArray, BooleanBuilder, FixedSizeBinaryBuilder, LargeStringBuilder,
+    NullArray, NullBufferBuilder, PrimitiveBuilder, StringBuilder, StringLikeArrayBuilder,
+    StringViewBuilder,
 };
 use arrow::compute::{CastOptions, DecimalCast};
 use arrow::datatypes::{self, DataType, DecimalType};
@@ -62,6 +63,9 @@ pub(crate) enum PrimitiveVariantToArrowRowBuilder<'a> {
     Time(VariantToPrimitiveArrowRowBuilder<'a, datatypes::Time64MicrosecondType>),
     Date(VariantToPrimitiveArrowRowBuilder<'a, datatypes::Date32Type>),
     Uuid(VariantToUuidArrowRowBuilder<'a>),
+    String(VariantToStringArrowBuilder<'a, StringBuilder>),
+    LargeString(VariantToStringArrowBuilder<'a, LargeStringBuilder>),
+    StringView(VariantToStringArrowBuilder<'a, StringViewBuilder>),
 }
 
 /// Builder for converting variant values into strongly typed Arrow arrays.
@@ -104,6 +108,9 @@ impl<'a> PrimitiveVariantToArrowRowBuilder<'a> {
             Time(b) => b.append_null(),
             Date(b) => b.append_null(),
             Uuid(b) => b.append_null(),
+            String(b) => b.append_null(),
+            LargeString(b) => b.append_null(),
+            StringView(b) => b.append_null(),
         }
     }
 
@@ -134,6 +141,9 @@ impl<'a> PrimitiveVariantToArrowRowBuilder<'a> {
             Time(b) => b.append_value(value),
             Date(b) => b.append_value(value),
             Uuid(b) => b.append_value(value),
+            String(b) => b.append_value(value),
+            LargeString(b) => b.append_value(value),
+            StringView(b) => b.append_value(value),
         }
     }
 
@@ -164,6 +174,9 @@ impl<'a> PrimitiveVariantToArrowRowBuilder<'a> {
             Time(b) => b.finish(),
             Date(b) => b.finish(),
             Uuid(b) => b.finish(),
+            String(b) => b.finish(),
+            LargeString(b) => b.finish(),
+            StringView(b) => b.finish(),
         }
     }
 }
@@ -304,6 +317,11 @@ pub(crate) fn make_primitive_variant_to_arrow_row_builder<'a>(
                 "FixedSizeBinary({size}) is not a valid variant shredding type. Only FixedSizeBinary(16) for UUID is supported."
             )));
         }
+        DataType::Utf8 => String(VariantToStringArrowBuilder::new(cast_options, capacity)),
+        DataType::LargeUtf8 => {
+            LargeString(VariantToStringArrowBuilder::new(cast_options, capacity))
+        }
+        DataType::Utf8View => StringView(VariantToStringArrowBuilder::new(cast_options, capacity)),
         _ if data_type.is_primitive() => {
             return Err(ArrowError::NotYetImplemented(format!(
                 "Primitive data_type {data_type:?} not yet implemented"
@@ -451,6 +469,13 @@ macro_rules! define_variant_to_primitive_builder {
     }
 }
 
+define_variant_to_primitive_builder!(
+    struct VariantToStringArrowBuilder<'a, B: StringLikeArrayBuilder>
+    |capacity| -> B { B::with_capacity(capacity) },
+    |value| value.as_string(),
+    type_name: B::type_name()
+);
+
 define_variant_to_primitive_builder!(
     struct VariantToBooleanArrowRowBuilder<'a>
     |capacity| -> BooleanBuilder { BooleanBuilder::with_capacity(capacity) },