feat(python): add model_dump_json override to handle FieldMetadata aliases

[tjb9dc]

Description

Refs extend-hq/extend-python-sdk#14

Adds a model_dump_json() override to UniversalBaseModel so that JSON serialization properly respects FieldMetadata aliases (e.g., snake_case → camelCase).

Background: The existing dict() method already handles FieldMetadata aliases via convert_and_respect_annotation_metadata(). However, model_dump_json() was calling Pydantic's native implementation directly, which doesn't know about FieldMetadata annotations. This caused JSON output to use Python field names instead of the expected wire-format aliases.

Requested by: [email protected] (@tjb9dc)
Link to Devin run: https://app.devin.ai/sessions/078ffa2dec06476a9c041a29f6f37cf4

Changes Made

• Added model_dump_json() override to UniversalBaseModel in pydantic_utilities.py
• Added same override to with_pydantic_v1_on_v2/pydantic_utilities.py variant
• Updated versions.yml with changelog entry for v4.43.0

The new method delegates to our custom dict() (which handles aliases correctly) and then serializes to JSON:

• Pydantic v2: Uses to_jsonable_python(data, fallback=encode_by_type) for proper v2 JSON normalization
• Pydantic v1: Uses json.dumps(data, default=encode_by_type) directly

Note: The with_pydantic_aliases variants were intentionally NOT updated because they use native Pydantic Field(alias=...) where by_alias=True already works correctly. The with_pydantic_v1_on_v2 variant uses the simpler v1-style implementation since it's explicitly v1-only.

Updates Since Last Revision

• Added proper v1/v2 branching in the main pydantic_utilities.py to use to_jsonable_python for Pydantic v2, matching the existing pattern used by to_jsonable_with_fallback in the codebase

Testing

• Unit tests added/updated
• Lint checks pass (pnpm run check)
• Seed tests (will run in CI)

Human Review Checklist

• Verify v1/v2 branching is correct (v2 uses to_jsonable_python, v1 uses json.dumps with default)
• Verify encode_by_type handles all necessary types (datetime, UUID, Decimal, Enum, etc.) - it uses Pydantic's ENCODERS_BY_TYPE
• Confirm seed tests adequately cover model_dump_json() with aliased fields
• Verify with_pydantic_v1_on_v2 variant correctly uses simpler implementation (no IS_PYDANTIC_V2 check needed since it's v1-only)

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring