feat: support for Solido RAG endpoints

[ajcasagrande]

Summary by CodeRabbit

• New Features

  • Added support for SOLIDO RAG endpoint integration
  • Implemented retrieval-augmented generation response handling with source attribution capabilities

• Tests

  • Added comprehensive test suite covering endpoint metadata, payload formatting, response parsing, and end-to-end integration scenarios

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[coderabbitai]

Walkthrough

Introduces SOLIDO RAG endpoint support by adding a new SOLIDO enum member to EndpointType, creating a RAGResponseData model with optional sources, implementing the SolidoEndpoint handler, and exposing these through public APIs alongside comprehensive test coverage.

Changes

Cohort / File(s)	Summary
Enum & Model Definitions src/aiperf/common/enums/plugin_enums.py, src/aiperf/common/models/record_models.py	Added SOLIDO = "solido" to EndpointType enum; introduced RAGResponseData class extending TextResponseData with optional sources field; updated ParsedResponse.data union to include RAGResponseData.
Model API Export src/aiperf/common/models/__init__.py	Added RAGResponseData to public API exports.
Endpoint Implementation src/aiperf/endpoints/solido_rag.py, src/aiperf/endpoints/base_endpoint.py	Created new SolidoEndpoint class implementing SOLIDO RAG endpoint protocol with metadata, payload formatting, response parsing, and data extraction; added make_rag_response_data() static helper to BaseEndpoint.
Endpoint API Exposure src/aiperf/endpoints/__init__.py	Imported and exposed SolidoEndpoint in public API.
Test Suite tests/endpoints/test_solido_rag.py	Added comprehensive test coverage for SolidoEndpoint including metadata validation, payload formatting, error handling, response parsing, and end-to-end integration scenarios.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

The changes introduce multiple new entity types (enum member, model class, endpoint class) with heterogeneous logic across different layers. While individual components are straightforward, the interconnected nature across models, endpoints, and tests requires careful verification of consistency in data flow and protocol compliance.

Poem

🐰 SOLIDO hops into the fold,
RAGResponseData sources bold,
New endpoint paths and parsing grace,
Tests ensure a proper place!
Sources tracked, responses neat—
The RAG adventure's now complete! 🌟

Pre-merge checks

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title Check	✅ Passed	The pull request title "feat: support for Solido RAG endpoints" accurately captures the primary objective of the changeset. The changes include introducing a new SOLIDO enum member, creating a RAGResponseData model with an optional sources field, implementing a new SolidoEndpoint class with complete request/response handling, and adding comprehensive test coverage. All modifications are cohesively aligned with adding support for Solido RAG endpoints, and the title clearly communicates this main purpose. The title is concise, specific, and uses standard conventional commit formatting, making it easily scannable in project history.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

---

📜 Recent review details

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 8ddb6b4 and 10e9597.

📒 Files selected for processing (7)

• src/aiperf/common/enums/plugin_enums.py (1 hunks)
• src/aiperf/common/models/__init__.py (2 hunks)
• src/aiperf/common/models/record_models.py (2 hunks)
• src/aiperf/endpoints/__init__.py (1 hunks)
• src/aiperf/endpoints/base_endpoint.py (2 hunks)
• src/aiperf/endpoints/solido_rag.py (1 hunks)
• tests/endpoints/test_solido_rag.py (1 hunks)

🧰 Additional context used

🧬 Code graph analysis (5)

src/aiperf/common/models/__init__.py (1)

src/aiperf/common/models/record_models.py (1)

• RAGResponseData (578-583)

src/aiperf/endpoints/base_endpoint.py (1)

src/aiperf/common/models/record_models.py (2)

• ParsedResponse (610-621)
• RAGResponseData (578-583)

src/aiperf/endpoints/__init__.py (1)

src/aiperf/endpoints/solido_rag.py (1)

• SolidoEndpoint (24-121)

src/aiperf/endpoints/solido_rag.py (7)

src/aiperf/common/enums/plugin_enums.py (1)

• EndpointType (19-30)

src/aiperf/common/factories.py (1)

• EndpointFactory (452-470)

src/aiperf/common/models/record_models.py (11)

• BaseResponseData (545-550)
• ParsedResponse (610-621)
• RequestInfo (706-759)
• get_json (212-214)
• get_json (237-244)
• get_json (332-341)
• get_raw (202-204)
• get_raw (229-231)
• get_raw (322-324)
• get (155-160)
• get (172-174)

src/aiperf/endpoints/base_endpoint.py (5)

• metadata (40-41)
• BaseEndpoint (28-103)
• format_payload (66-70)
• parse_response (73-76)
• make_rag_response_data (99-103)

src/aiperf/common/models/metadata.py (1)

• EndpointMetadata (11-41)

src/aiperf/common/protocols.py (3)

• EndpointProtocol (342-353)
• InferenceServerResponse (357-398)
• debug (71-71)

src/aiperf/common/models/model_endpoint_info.py (1)

• primary_model_name (148-150)

tests/endpoints/test_solido_rag.py (6)

src/aiperf/common/enums/plugin_enums.py (1)

• EndpointType (19-30)

src/aiperf/common/models/record_models.py (8)

• RequestInfo (706-759)
• TextResponseData (553-560)
• get_json (212-214)
• get_json (237-244)
• get_json (332-341)
• get_raw (202-204)
• get_raw (229-231)
• get_raw (322-324)

src/aiperf/common/protocols.py (1)

• InferenceServerResponse (357-398)

src/aiperf/endpoints/solido_rag.py (4)

• SolidoEndpoint (24-121)
• metadata (33-41)
• format_payload (43-77)
• parse_response (79-98)

tests/endpoints/conftest.py (3)

• create_endpoint_with_mock_transport (43-49)
• create_model_endpoint (21-40)
• create_request_info (52-66)

src/aiperf/endpoints/base_endpoint.py (4)

• metadata (40-41)
• format_payload (66-70)
• parse_response (73-76)
• make_text_response_data (94-96)

🪛 Ruff (0.14.1)

src/aiperf/endpoints/solido_rag.py

53-53: Avoid specifying long messages outside the exception class

(TRY003)

tests/endpoints/test_solido_rag.py

246-246: Unused method argument: description

(ARG002)

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (9)

• GitHub Check: build (ubuntu-latest, 3.12)
• GitHub Check: build (macos-latest, 3.12)
• GitHub Check: build (macos-latest, 3.10)
• GitHub Check: build (ubuntu-latest, 3.10)
• GitHub Check: build (ubuntu-latest, 3.11)
• GitHub Check: build (macos-latest, 3.11)
• GitHub Check: integration-tests (ubuntu-latest, 3.12)
• GitHub Check: integration-tests (ubuntu-latest, 3.11)
• GitHub Check: integration-tests (ubuntu-latest, 3.10)

🔇 Additional comments (17)

src/aiperf/endpoints/__init__.py (1)

19-21: LGTM! Clean integration of the new endpoint.

The import and export of SolidoEndpoint follows the existing pattern consistently with other endpoints in the module.

Also applies to: 29-29

src/aiperf/common/models/__init__.py (1)

80-80: LGTM! Proper export of the RAGResponseData model.

The export follows the existing pattern and correctly integrates RAGResponseData into the public API surface.

Also applies to: 162-162

src/aiperf/common/enums/plugin_enums.py (1)

30-30: LGTM! Clean enum addition.

The SOLIDO endpoint type follows the existing naming and value conventions.

src/aiperf/endpoints/base_endpoint.py (2)

7-7: LGTM! Proper imports for RAG support.

The addition of Any type hint and RAGResponseData import enables the new RAG response data helper.

Also applies to: 15-15

---

98-103: LGTM! Well-structured helper method.

The make_rag_response_data helper follows the same pattern as make_text_response_data, providing a consistent API for creating RAG response objects. The implementation correctly returns None when text is empty, maintaining the expected behavior.

src/aiperf/common/models/record_models.py (2)

578-584: LGTM! Well-designed RAG response data model.

The RAGResponseData class appropriately extends TextResponseData to add RAG-specific metadata. The sources field is correctly typed as optional and uses a flexible list[Any] to accommodate various source formats.

---

616-616: LGTM! Proper integration into the response data union.

The addition of RAGResponseData to the union type maintains the type hierarchy correctly, positioned between ReasoningResponseData and TextResponseData.

tests/endpoints/test_solido_rag.py (5)

23-30: LGTM! Useful test helper.

The create_mock_response helper provides a clean way to construct mock InferenceServerResponse objects for testing.

---

32-44: LGTM! Thorough metadata validation.

The test comprehensively validates all metadata fields for the SOLIDO endpoint.

---

46-195: LGTM! Excellent payload formatting test coverage.

The test suite thoroughly covers:

• Basic payload formatting
• Multiple text contents
• Empty content filtering
• Model selection precedence
• Extra parameter merging
• Custom filter overrides
• Error handling for missing turns
• Multi-turn handling (last turn selection)

---

197-271: LGTM! Comprehensive response parsing tests.

The test suite covers valid responses, additional fields handling, invalid responses, and various content types including multi-line, special characters, and Unicode.

Note: The static analysis warning about the unused description parameter on line 246 is a false positive - the parameter is intentionally used only in the @pytest.mark.parametrize decorator for test case identification.

---

273-327: LGTM! Solid integration tests.

The integration tests validate the full request-response cycle and the helper method behavior with various inputs.

src/aiperf/endpoints/solido_rag.py (5)

22-30: LGTM! Proper endpoint registration.

The class is correctly decorated with @implements_protocol(EndpointProtocol) and registered with @EndpointFactory.register(EndpointType.SOLIDO). The docstring clearly describes the endpoint's purpose and capabilities.

---

32-41: LGTM! Comprehensive metadata configuration.

The metadata correctly configures the endpoint with:

• Proper endpoint path for SOLIDO RAG API
• Streaming support enabled
• Token production and input tokenization enabled
• Descriptive metrics title

---

43-77: LGTM! Well-structured payload formatting.

The implementation correctly:

• Validates that at least one turn exists
• Extracts query text from the last turn (multi-turn support)
• Applies default filters for SOLIDO
• Handles model selection with proper precedence (turn model → endpoint model)
• Merges extra parameters from the endpoint configuration
• Includes debug logging

The design intentionally allows extra parameters to override defaults (including filters), as confirmed by the test suite.

---

79-98: LGTM! Clean response parsing logic.

The method properly:

• Extracts JSON from the response
• Logs when JSON is missing
• Delegates to the extraction helper
• Returns None for invalid responses

---

100-121: LGTM! Appropriate RAG data extraction.

The implementation correctly:

• Extracts the content field as the primary response text
• Returns None when content is missing
• Optionally extracts sources (logs debug message but continues when missing)
• Uses the make_rag_response_data helper to construct the response

The design appropriately treats sources as optional, which aligns with RAG use cases where sources may not always be available.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}