This is an automated email from the ASF dual-hosted git repository. mgrigorov pushed a commit to branch 402-disallow-two-same-named-schemas-in-a-union in repository https://gitbox.apache.org/repos/asf/avro-rs.git
commit b9b3ccd132e127ced29a1cda81465daf7e82dd0b Author: Martin Tzvetanov Grigorov <[email protected]> AuthorDate: Mon Jan 19 14:54:11 2026 +0200 fix: UnionSchema::new(...) should not accept duplicate named schemas Fixes #402 --- avro/src/error.rs | 3 +++ avro/src/schema/union.rs | 63 +++++++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 63 insertions(+), 3 deletions(-) diff --git a/avro/src/error.rs b/avro/src/error.rs index 48eb91b..91593d3 100644 --- a/avro/src/error.rs +++ b/avro/src/error.rs @@ -301,6 +301,9 @@ pub enum Details { #[error("Unions cannot contain duplicate types")] GetUnionDuplicate, + #[error("Unions cannot contain more than one named variant with the same name: {0}")] + GetUnionDuplicateNamedVariants(String), + #[error("One union type {0:?} must match the `default`'s value type {1:?}")] GetDefaultUnion(SchemaKind, ValueKind), diff --git a/avro/src/schema/union.rs b/avro/src/schema/union.rs index a707279..3846319 100644 --- a/avro/src/schema/union.rs +++ b/avro/src/schema/union.rs @@ -20,7 +20,7 @@ use crate::error::Details; use crate::schema::{Name, Namespace, ResolvedSchema, Schema, SchemaKind}; use crate::types; use std::borrow::Borrow; -use std::collections::{BTreeMap, HashMap}; +use std::collections::{BTreeMap, HashMap, HashSet}; use std::fmt::Debug; /// A description of a Union schema @@ -42,13 +42,19 @@ impl UnionSchema { /// Will return an error if `schemas` has duplicate unnamed schemas or if `schemas` /// contains a union. pub fn new(schemas: Vec<Schema>) -> AvroResult<Self> { + let mut named_schemas: HashSet<&Name> = HashSet::default(); let mut vindex = BTreeMap::new(); for (i, schema) in schemas.iter().enumerate() { if let Schema::Union(_) = schema { return Err(Details::GetNestedUnion.into()); - } - if !schema.is_named() && vindex.insert(SchemaKind::from(schema), i).is_some() { + } else if !schema.is_named() && vindex.insert(SchemaKind::from(schema), i).is_some() { return Err(Details::GetUnionDuplicate.into()); + } else if schema.is_named() { + let name = schema.name().unwrap(); + if !named_schemas.insert(name) { + return Err(Details::GetUnionDuplicateNamedVariants(name.to_string()).into()); + } + vindex.insert(SchemaKind::from(schema), i); } } Ok(UnionSchema { @@ -123,3 +129,54 @@ impl PartialEq for UnionSchema { self.schemas.eq(&other.schemas) } } + +#[cfg(test)] +mod tests { + use super::*; + use crate::error::{Details, Error}; + use crate::schema::RecordSchema; + use apache_avro_test_helper::TestResult; + + #[test] + fn avro_rs_402_new_union_schema() -> TestResult { + let schema1 = Schema::Int; + let schema2 = Schema::String; + let union_schema = UnionSchema::new(vec![schema1.clone(), schema2.clone()])?; + + assert_eq!(union_schema.variants(), &[schema1, schema2]); + + Ok(()) + } + + #[test] + fn avro_rs_402_new_union_schema_duplicate_names() -> TestResult { + let res = UnionSchema::new(vec![ + Schema::Record(RecordSchema { + name: Name::new("Something").unwrap(), + aliases: None, + doc: None, + fields: vec![], + lookup: Default::default(), + attributes: Default::default(), + }), + Schema::Record(RecordSchema { + name: Name::new("Something").unwrap(), + aliases: None, + doc: None, + fields: vec![], + lookup: Default::default(), + attributes: Default::default(), + }), + ]) + .map_err(Error::into_details); + + match res { + Err(Details::GetUnionDuplicateNamedVariants(name)) => { + assert_eq!(name, Name::new("Something")?.to_string()); + } + _ => panic!("Expected GetUnionDuplicateNamedVariants error"), + } + + Ok(()) + } +}
