fix: ODCS to .proto, problem with array of objects

[Schokuroff]

• Tests pass
• ruff format
• README.md updated (if relevant)
•  CHANGELOG.md entry added

[Schokuroff]

schema:
- name: FsaRegister
  logicalType: object
  properties:
  - name: fsa_room
    logicalType: array
    items:
      name: FsaRoom
      logicalType: object
      properties:
      - name: rooms_count
        logicalType: integer
      - name: category_type
        logicalType: string
  - name: simple_array_int
    logicalType: array
    items:
      logicalType: integer
  - name: simple_int
    logicalType: integer
  - name: simple_obj
    logicalType: object
    properties:
    - name: rooms_count
      logicalType: integer
    - name: category_type
      logicalType: string

to

syntax = "proto3";

package example;

message FsaRegister {
message FsaRoom {
    int32 rooms_count = 1;
    string category_type = 2;
  }

message SimpleObj {
    int32 rooms_count = 1;
    string category_type = 2;
  }

  repeated FsaRoom fsa_room = 1;
  repeated int32 simple_array_int = 2;
  int32 simple_int = 3;
  SimpleObj simple_obj = 4;
}

[Schokuroff]

Sorry, I think I forgot to do the ruff and tests, I'll add them now.

[Schokuroff]

@jochenchrist
wow, looks like all checks have passed, can we merge it?

[Schokuroff]

I have done ruff check localy, but failed with "uv run pytest", oracle idk...

uv run pytest
========================================================= test session starts =========================================================
platform darwin -- Python 3.11.8, pytest-9.0.2, pluggy-1.6.0
rootdir: /Users/nshokurov/Documents/datacontract/datacontract-cli
configfile: pyproject.toml
plugins: anyio-4.12.1, xdist-3.8.0, typeguard-4.4.4
collected 261 items / 5 errors

=============================================================== ERRORS ================================================================
_____________________________________________ ERROR collecting tests/test_test_oracle.py ______________________________________________
.venv/lib/python3.11/site-packages/urllib3/connectionpool.py:787: in urlopen

[Schokuroff]

@jochenchrist sooo, lets merge?

[Schokuroff]

@jochenchrist hello! lets merge it?

[Copilot]

Pull request overview

This PR updates the Protobuf exporter to better handle ODCS schemas—specifically arrays of objects—by generating nested message definitions and updating the expected .proto output accordingly.

Changes:

• Extend the Protobuf export test fixture to include an array-of-objects field (users) and a non-required field (geo_description).
• Refactor Protobuf type/message naming and nested message generation logic to support arrays of objects.
• Add support for emitting optional fields for required: false properties.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 4 comments.

File	Description
tests/test_export_protobuf.py	Adds coverage for arrays of objects and required: false → optional output in generated Protobuf.
datacontract/export/protobuf_exporter.py	Refactors Protobuf generation to create nested messages (including for arrays of objects) and emit optional for non-required fields.

Comments suppressed due to low confidence (1)

datacontract/export/protobuf_exporter.py:22

• _get_config_value is typed as returning Optional[str], but cp.value can be a non-string (e.g., dict for enumValues, as used by _get_enum_values). This type mismatch can hide bugs in type checking; consider widening the return type (e.g., Optional[Any]) or making it generic.

def _get_config_value(prop: SchemaProperty, key: str) -> Optional[str]:
    """Get a custom property value from customProperties."""
    if prop.customProperties is None:
        return None
    for cp in prop.customProperties:
        if cp.property == key:
            return cp.value
    return None

---

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

[Copilot]

For arrays of objects, _get_type_name returns prop.items.name without normalizing it. However, to_protobuf_message() will UpperCamel-case the message name, which can cause a type mismatch (e.g., items.name=user → nested message User {} but field type repeated user). Normalize items.name the same way as message declarations (e.g., via _snake_to_upper_camel) so field references always match generated type names.

Suggested change

	# Use as-is (might already be in UpperCamelCase)
	return prop.items.name
	# Normalize items.name the same way as message declarations
	return _snake_to_upper_camel(prop.items.name)

[Schokuroff]

I am not sure, we need it

[Copilot]

_get_field_declaration unconditionally prefixes optional for required: false non-array fields. In proto3, optional is only valid for scalar/enum fields; applying it to message/object-typed fields can produce an invalid .proto. Gate this so optional is added only for scalar/enum field declarations (and never for message types).

Suggested change

	# Add 'optional' for non-required fields that are not arrays
	if (hasattr(prop, 'required') and prop.required is False and
	not (prop.logicalType and prop.logicalType.lower() == "array")):
	logical_type = (prop.logicalType or "").lower()
	is_array = logical_type == "array"
	is_message_type = logical_type in ["object", "record", "struct"]
	# Add 'optional' only for non-required, non-array, non-message fields (scalars/enums)
	if hasattr(prop, "required") and prop.required is False and not is_array and not is_message_type:

[Copilot]

There’s leftover commented-out code (# field_type = ...) and a non-English inline comment (# Новая функция) in the message generation loop. Please remove the dead/commented code and use a consistent language for comments to keep the exporter maintainable.

Suggested change

	# field_type = _get_field_type(prop) # includes "repeated" if needed
	field_decl = _get_field_declaration(prop) # Новая функция
	field_decl = _get_field_declaration(prop) # Use helper to build field declaration (handles optional/repeated)