feat: Variant Support

crates/catalog/glue/src/schema.rs

@@ -25,7 +25,7 @@ pub(crate) const ICEBERG_FIELD_CURRENT: &str = "iceberg.field.current";
 use std::collections::HashMap;
 
 use aws_sdk_glue::types::Column;
-use iceberg::spec::{PrimitiveType, SchemaVisitor, TableMetadata, visit_schema};
+use iceberg::spec::{PrimitiveType, SchemaVisitor, TableMetadata, VariantType, visit_schema};
 use iceberg::{Error, ErrorKind, Result};
 
 use crate::error::from_aws_build_error;
@@ -182,6 +182,10 @@ impl SchemaVisitor for GlueSchemaBuilder {
 
         Ok(glue_type)
     }
+
+    fn variant(&mut self, _v: &VariantType) -> Result<Self::T> {
+        Ok("variant".to_string())
+    }
 }
 
 #[cfg(test)]

crates/catalog/hms/src/schema.rs

@@ -16,7 +16,7 @@
 // under the License.
 
 use hive_metastore::FieldSchema;
-use iceberg::spec::{PrimitiveType, Schema, SchemaVisitor, visit_schema};
+use iceberg::spec::{PrimitiveType, Schema, SchemaVisitor, VariantType, visit_schema};
 use iceberg::{Error, ErrorKind, Result};
 
 type HiveSchema = Vec<FieldSchema>;
@@ -139,6 +139,12 @@ impl SchemaVisitor for HiveSchemaBuilder {
 
         Ok(hive_type)
     }
+
+    fn variant(&mut self, _v: &VariantType) -> Result<Self::T> {
+        // Match iceberg-java's HiveSchemaUtil, which maps VARIANT to "unknown"
+        // (apache/iceberg#15964).
+        Ok("unknown".to_string())
+    }
 }
 
 #[cfg(test)]

crates/iceberg/public-api.txt

@@ -1679,6 +1679,7 @@ pub iceberg::spec::Type::List(iceberg::spec::ListType)
 pub iceberg::spec::Type::Map(iceberg::spec::MapType)
 pub iceberg::spec::Type::Primitive(iceberg::spec::PrimitiveType)
 pub iceberg::spec::Type::Struct(iceberg::spec::StructType)
+pub iceberg::spec::Type::Variant(iceberg::spec::VariantType)
 impl iceberg::spec::Type
 pub fn iceberg::spec::Type::as_primitive_type(&self) -> core::option::Option<&iceberg::spec::PrimitiveType>
 pub fn iceberg::spec::Type::decimal(precision: u32, scale: u32) -> iceberg::Result<Self>
@@ -1688,6 +1689,8 @@ pub fn iceberg::spec::Type::is_floating_type(&self) -> bool
 pub fn iceberg::spec::Type::is_nested(&self) -> bool
 pub fn iceberg::spec::Type::is_primitive(&self) -> bool
 pub fn iceberg::spec::Type::is_struct(&self) -> bool
+pub fn iceberg::spec::Type::is_variant(&self) -> bool
+pub fn iceberg::spec::Type::min_format_version(&self) -> iceberg::spec::FormatVersion
 pub fn iceberg::spec::Type::to_struct_type(self) -> core::option::Option<iceberg::spec::StructType>
 impl core::clone::Clone for iceberg::spec::Type
 pub fn iceberg::spec::Type::clone(&self) -> iceberg::spec::Type
@@ -1702,6 +1705,8 @@ impl core::convert::From<iceberg::spec::PrimitiveType> for iceberg::spec::Type
 pub fn iceberg::spec::Type::from(value: iceberg::spec::PrimitiveType) -> Self
 impl core::convert::From<iceberg::spec::StructType> for iceberg::spec::Type
 pub fn iceberg::spec::Type::from(value: iceberg::spec::StructType) -> Self
+impl core::convert::From<iceberg::spec::VariantType> for iceberg::spec::Type
+pub fn iceberg::spec::Type::from(_: iceberg::spec::VariantType) -> Self
 impl core::fmt::Debug for iceberg::spec::Type
 pub fn iceberg::spec::Type::fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result
 impl core::fmt::Display for iceberg::spec::Type
@@ -2295,6 +2300,7 @@ impl iceberg::spec::Schema
 pub fn iceberg::spec::Schema::accessor_by_field_id(&self, field_id: i32) -> core::option::Option<alloc::sync::Arc<iceberg::expr::accessor::StructAccessor>>
 pub fn iceberg::spec::Schema::as_struct(&self) -> &iceberg::spec::StructType
 pub fn iceberg::spec::Schema::builder() -> iceberg::spec::SchemaBuilder
+pub fn iceberg::spec::Schema::check_format_compatibility(&self, format_version: iceberg::spec::FormatVersion) -> iceberg::Result<()>
 pub fn iceberg::spec::Schema::field_by_alias(&self, alias: &str) -> core::option::Option<&iceberg::spec::NestedFieldRef>
 pub fn iceberg::spec::Schema::field_by_id(&self, field_id: i32) -> core::option::Option<&iceberg::spec::NestedFieldRef>
 pub fn iceberg::spec::Schema::field_by_name(&self, field_name: &str) -> core::option::Option<&iceberg::spec::NestedFieldRef>
@@ -2305,6 +2311,7 @@ pub fn iceberg::spec::Schema::field_id_to_name_map(&self) -> &std::collections::
 pub fn iceberg::spec::Schema::highest_field_id(&self) -> i32
 pub fn iceberg::spec::Schema::identifier_field_ids(&self) -> impl core::iter::traits::exact_size::ExactSizeIterator<Item = i32> + '_
 pub fn iceberg::spec::Schema::into_builder(self) -> iceberg::spec::SchemaBuilder
+pub fn iceberg::spec::Schema::min_format_version(&self) -> iceberg::spec::FormatVersion
 pub fn iceberg::spec::Schema::name_by_field_id(&self, field_id: i32) -> core::option::Option<&str>
 pub fn iceberg::spec::Schema::schema_id(&self) -> iceberg::spec::SchemaId
 impl core::clone::Clone for iceberg::spec::Schema
@@ -2802,6 +2809,26 @@ impl core::default::Default for iceberg::spec::UnboundPartitionSpecBuilder
 pub fn iceberg::spec::UnboundPartitionSpecBuilder::default() -> iceberg::spec::UnboundPartitionSpecBuilder
 impl core::fmt::Debug for iceberg::spec::UnboundPartitionSpecBuilder
 pub fn iceberg::spec::UnboundPartitionSpecBuilder::fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result
+pub struct iceberg::spec::VariantType
+impl core::clone::Clone for iceberg::spec::VariantType
+pub fn iceberg::spec::VariantType::clone(&self) -> iceberg::spec::VariantType
+impl core::cmp::Eq for iceberg::spec::VariantType
+impl core::cmp::PartialEq for iceberg::spec::VariantType
+pub fn iceberg::spec::VariantType::eq(&self, other: &iceberg::spec::VariantType) -> bool
+impl core::convert::From<iceberg::spec::VariantType> for iceberg::spec::Type
+pub fn iceberg::spec::Type::from(_: iceberg::spec::VariantType) -> Self
+impl core::fmt::Debug for iceberg::spec::VariantType
+pub fn iceberg::spec::VariantType::fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result
+impl core::fmt::Display for iceberg::spec::VariantType
+pub fn iceberg::spec::VariantType::fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result
+impl core::hash::Hash for iceberg::spec::VariantType
+pub fn iceberg::spec::VariantType::hash<__H: core::hash::Hasher>(&self, state: &mut __H)
+impl core::marker::Copy for iceberg::spec::VariantType
+impl core::marker::StructuralPartialEq for iceberg::spec::VariantType
+impl serde_core::ser::Serialize for iceberg::spec::VariantType
+pub fn iceberg::spec::VariantType::serialize<S>(&self, serializer: S) -> core::result::Result<<S as serde_core::ser::Serializer>::Ok, <S as serde_core::ser::Serializer>::Error> where S: serde_core::ser::Serializer
+impl<'de> serde_core::de::Deserialize<'de> for iceberg::spec::VariantType
+pub fn iceberg::spec::VariantType::deserialize<D>(deserializer: D) -> core::result::Result<Self, <D as serde_core::de::Deserializer>::Error> where D: serde_core::de::Deserializer<'de>
 pub struct iceberg::spec::ViewMetadata
 impl iceberg::spec::ViewMetadata
 pub fn iceberg::spec::ViewMetadata::current_schema(&self) -> &iceberg::spec::SchemaRef
@@ -2921,6 +2948,7 @@ pub const iceberg::spec::LIST_FIELD_NAME: &str
 pub const iceberg::spec::MAIN_BRANCH: &str
 pub const iceberg::spec::MAP_KEY_FIELD_NAME: &str
 pub const iceberg::spec::MAP_VALUE_FIELD_NAME: &str
+pub const iceberg::spec::MIN_FORMAT_VERSION_DEFAULT_VALUES: iceberg::spec::FormatVersion
 pub const iceberg::spec::MIN_FORMAT_VERSION_ROW_LINEAGE: iceberg::spec::FormatVersion
 pub const iceberg::spec::SCHEMA_NAME_DELIMITER: &str
 pub const iceberg::spec::UNASSIGNED_SEQUENCE_NUMBER: i64
@@ -2956,6 +2984,7 @@ pub fn iceberg::spec::SchemaVisitor::map(&mut self, map: &iceberg::spec::MapType
 pub fn iceberg::spec::SchemaVisitor::primitive(&mut self, p: &iceberg::spec::PrimitiveType) -> iceberg::Result<Self::T>
 pub fn iceberg::spec::SchemaVisitor::schema(&mut self, schema: &iceberg::spec::Schema, value: Self::T) -> iceberg::Result<Self::T>
 pub fn iceberg::spec::SchemaVisitor::struct(&mut self, struct: &iceberg::spec::StructType, results: alloc::vec::Vec<Self::T>) -> iceberg::Result<Self::T>
+pub fn iceberg::spec::SchemaVisitor::variant(&mut self, _v: &iceberg::spec::VariantType) -> iceberg::Result<Self::T>
 pub trait iceberg::spec::SchemaWithPartnerVisitor<P>
 pub type iceberg::spec::SchemaWithPartnerVisitor::T
 pub fn iceberg::spec::SchemaWithPartnerVisitor::after_list_element(&mut self, _field: &iceberg::spec::NestedFieldRef, _partner: &P) -> iceberg::Result<()>
@@ -2972,6 +3001,7 @@ pub fn iceberg::spec::SchemaWithPartnerVisitor::map(&mut self, map: &iceberg::sp
 pub fn iceberg::spec::SchemaWithPartnerVisitor::primitive(&mut self, p: &iceberg::spec::PrimitiveType, partner: &P) -> iceberg::Result<Self::T>
 pub fn iceberg::spec::SchemaWithPartnerVisitor::schema(&mut self, schema: &iceberg::spec::Schema, partner: &P, value: Self::T) -> iceberg::Result<Self::T>
 pub fn iceberg::spec::SchemaWithPartnerVisitor::struct(&mut self, struct: &iceberg::spec::StructType, partner: &P, results: alloc::vec::Vec<Self::T>) -> iceberg::Result<Self::T>
+pub fn iceberg::spec::SchemaWithPartnerVisitor::variant(&mut self, _v: &iceberg::spec::VariantType, _partner: &P) -> iceberg::Result<Self::T>
 pub fn iceberg::spec::deserialize_data_file_from_json(json: &str, partition_spec_id: i32, partition_type: &iceberg::spec::StructType, schema: &iceberg::spec::Schema) -> iceberg::Result<iceberg::spec::DataFile>
 pub fn iceberg::spec::prune_columns(schema: &iceberg::spec::Schema, selected: impl core::iter::traits::collect::IntoIterator<Item = i32>, select_full_types: bool) -> iceberg::Result<iceberg::spec::Type>
 pub fn iceberg::spec::read_data_files_from_avro<R: std::io::Read>(reader: &mut R, schema: &iceberg::spec::Schema, partition_spec_id: i32, partition_type: &iceberg::spec::StructType, version: iceberg::spec::FormatVersion) -> iceberg::Result<alloc::vec::Vec<iceberg::spec::DataFile>>

crates/iceberg/src/arrow/caching_delete_file_loader.rs

@@ -35,7 +35,7 @@ use crate::runtime::Runtime;
 use crate::scan::{ArrowRecordBatchStream, FileScanTaskDeleteFile};
 use crate::spec::{
     DataContentType, Datum, ListType, MapType, NestedField, NestedFieldRef, PartnerAccessor,
-    PrimitiveType, Schema, SchemaRef, SchemaWithPartnerVisitor, StructType, Type,
+    PrimitiveType, Schema, SchemaRef, SchemaWithPartnerVisitor, StructType, Type, VariantType,
     visit_schema_with_partner,
 };
 use crate::{Error, ErrorKind, Result};
@@ -544,6 +544,10 @@ impl SchemaWithPartnerVisitor<ArrayRef> for EqDelColumnProcessor<'_> {
     fn primitive(&mut self, _primitive: &PrimitiveType, _partner: &ArrayRef) -> Result<()> {
         Ok(())
     }
+
+    fn variant(&mut self, _v: &VariantType, _partner: &ArrayRef) -> Result<()> {
+        Ok(())
+    }
 }
 
 struct EqDelRecordBatchPartnerAccessor;

crates/iceberg/src/arrow/nan_val_cnt_visitor.rs

@@ -28,7 +28,7 @@ use crate::Result;
 use crate::arrow::{ArrowArrayAccessor, FieldMatchMode};
 use crate::spec::{
     ListType, MapType, NestedFieldRef, PrimitiveType, Schema, SchemaRef, SchemaWithPartnerVisitor,
-    StructType, visit_struct_with_partner,
+    StructType, VariantType, visit_struct_with_partner,
 };
 
 macro_rules! cast_and_update_cnt_map {
@@ -122,6 +122,10 @@ impl SchemaWithPartnerVisitor<ArrayRef> for NanValueCountVisitor {
         Ok(())
     }
 
+    fn variant(&mut self, _v: &VariantType, _col: &ArrayRef) -> Result<Self::T> {
+        Ok(())
+    }
+
     fn after_struct_field(&mut self, field: &NestedFieldRef, partner: &ArrayRef) -> Result<()> {
         let field_id = field.id;
         count_float_nans!(partner, self, field_id);