feat(protobuf): enhance Protobuf data handling with LogicalMap support and enhance test coverage

[Gezi-lzq]

Protobuf to Avro Conversion Enhancements

• Added LogicalMapProtobufData: This new class ensures that Protobuf map fields are annotated with Iceberg's LogicalMap logical type, allowing downstream conversion to preserve MAP semantics instead of defaulting to ARRAY<record<key,value>>.
• Updated ProtoToAvroConverter: Now uses the new LogicalMapProtobufData singleton, applies logical type conversions (e.g., for timestamps), and consistently unwraps union schemas to their non-NULL element for conversion. Also improves handling of optional and oneof fields, ensuring correct Avro nullability and presence semantics. [1] [2]

Avro Schema Union Handling and Binder Precomputation

• Improved union schema resolution in RecordBinder: Now unions are only unwrapped if they contain exactly one non-NULL type, with clear error handling for unsupported multi-type unions. This prevents incorrect assumptions about union schemas and improves error messages. [1] [2]
• Refactored binder precomputation for nested types: The logic for creating binders for STRUCT, LIST, and MAP types is now modularized into dedicated methods, supporting both native Avro MAP and ARRAY-of-records representations, and correctly handling struct keys/values and unioned structs.

[Copilot]

Pull Request Overview

This PR enhances Protobuf-to-Avro data handling by introducing LogicalMap support for proper MAP type preservation and improving union schema handling. The changes include:

• A new LogicalMapProtobufData class that annotates Protobuf map fields with Iceberg's LogicalMap logical type
• Refactored ProtoToAvroConverter with improved union unwrapping and timestamp conversion
• Enhanced RecordBinder with better union validation and modularized binder precomputation
• Comprehensive test coverage additions for Protobuf conversion scenarios

Key Changes

• LogicalMap Support: Protobuf map fields now preserve MAP semantics instead of defaulting to ARRAY<record<key,value>>
• Union Handling: Improved validation to reject non-optional unions with multiple non-NULL types, with clear error messages
• Test Coverage: Added ~900 lines of new tests covering edge cases, optional fields, oneofs, and complex nested structures

Reviewed Changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 13 comments.

Show a summary per file

File	Description
LogicalMapProtobufData.java	New class extending ProtobufData to annotate map fields with LogicalMap logical type
ProtoToAvroConverter.java	Refactored conversion logic with consistent union unwrapping and improved timestamp handling
RecordBinder.java	Enhanced union resolution with validation and modularized binder creation for nested types
ProtobufRegistryConverter.java	Updated to use LogicalMapProtobufData singleton
CodecSetup.java	Added public accessor for LogicalMap instance
ProtobufRegistryConverterUnitTest.java	New comprehensive test suite (719 lines)
ProtobufRegistryConverterTest.java	Added map field test case
ProtoToAvroConverterTest.java	New unit tests for converter edge cases
AvroRecordBinderTypeTest.java	New comprehensive type conversion tests (1000 lines)
AvroRecordBinderTest.java	Refactored to focus on specific binding scenarios

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.