release: v0.3.0 — Fase B, capa REST completa - #111
Merged
Merged
Conversation
* feat(api): add REST API contract and models for clickbait analysis feat(nlp): update model cards with dimensions for signal classification test(nlp): enhance model cards tests to validate dimensions and signal registration * feat(api): implement orchestration for POST /analyze endpoint * feat(api): enhance headline validation to prevent empty strings in AnalyzeRequest * feat(ci): update branch filter to include dev for pull requests and pushes * feat(tests): add comprehensive tests for /analyze orchestration with mock dependencies
* feat(requirements): add fastapi and uvicorn to requirements * docs: registrar H1 y aclarar la organización del trabajo Sección de H1 en Fase B: contrato de POST /analyze — los tres principios del esquema, por qué el veredicto va por dimensiones y no por mayoría, el tono como señal que se muestra pero no vota, y el desajuste del campo signal que destapó el diseño. Incluye los límites: analyze() no es alcanzable por HTTP y los tests usan dobles. Convenciones: el README decía 'minor = épica' mientras la tabla de hitos asignaba v0.3 a H2, llevando dos ejes de agrupación a la vez. En Fase B el hito sustituye a la épica —ya trae dominio y fecha— y las épicas quedan como histórico de Fase A. La tabla de ramas documentaba 'feature/' cuando la práctica es 'feat/<nº issue>-<slug> * feat(api): app FastAPI con /analyze y /health App REST como segundo entry point, independiente del servidor MCP: se arranca con `uvicorn backend.api.app:app`. Ambos son fachadas sobre el mismo núcleo, no una encima de la otra. - POST /analyze delega en la orquestación ya existente. Devuelve 200 aunque alguna señal falle; cada una lleva su propio status. - GET /health reutiliza el mismo sondeo que la tool MCP: check_health sale de register() para que las dos fachadas no puedan divergir. - CORS con orígenes por setting. allow_credentials en False: autoriza cookies y autenticación HTTP, que aquí no existen, y no es lo que necesitan los tokens Bearer. - configure_logging en el lifespan, no al importar, para que los tests no hereden la configuración global de structlog. 12 tests de la capa HTTP: validación, delegación, CORS y OpenAPI. No repiten la orquestación, que ya cubre test_analyze.py. Validado en vivo: análisis real por HTTP en 0,4-0,7 s en caliente. Discrimina — "Federal Reserve Holds Interest Rates Steady" sale factual con las cuatro señales de acuerdo; "17 Things Nobody Tells You About Moving Abroad" sale ambiguo porque el zero-shot lo lee como factual y léxico y lineal lo marcan. * feat(docs): update requirements and README with deployment architecture and API behavior
* chore: add Ruff configuration file for linter and formatter and fixed bugs found by it. * chore: add unused numpy import for testing Ruff checks * chore: remove unused numpy import for Ruff checks
* feat(config): make MCP server transport configurable with default to stdio * feat(tests): automate HTTP server verification for MCP transport
…quisitos para mayor claridad (#96)
* feat(integrations): implement automatic discovery and registration of integrations * ruff check
* feat(metadata): add tool metadata handling for categories and origins * feat(api): implement GET /tools endpoint for dynamic tool catalog retrieval
* feat(core): relaunch exceptions in log_tool_invocation to ensure proper error handling * feat(mcp): contrato de retorno estructurado en las tools * refactor(api): refactor MCP session handling into a dedicated module * fix(api): correct import and type hint for async session handling * feat(api): add execute tool endpoint with argument validation and error handling * docs: registrar #100 en el README * feat(spike): add Spike #82 for revalidating selection with real descriptions
…#104) * feat(history): implement persistent history storage for analysis and tool executions * feat(tests): implement history tests. * feat(history): refine SQL query to specify columns in history retrieval feat(bench): add benchmarks for database connection and write performance feat(bench): measure impact of connection management on SQLite performance
* feat(api): implement history retention policies and indexing for performance * feat(api): add filtering options for history retrieval and include retention policy in response * Refine TFG requirements, enhance Ruff configuration, and add performance benchmarks - Updated requirements in `docs/requisitos.md` to clarify filtering criteria for execution history in the Backend API. - Enhanced `ruff.toml` to include detection of blocking calls within async functions and added comments on limitations. - Introduced new benchmarking scripts in `spikes/bench_poda_convergencia.py` and `spikes/bench_poda_coste_e_indice.py` to evaluate pruning strategies and their costs. - Created `spikes/bench_poda_limite_fijo_y_count.py` to address design flaws in previous benchmarks and assess the impact of fixed limits on pruning performance. - Added `spikes/check_fechas_lexicograficas.py` to verify the correctness of date comparisons in SQLite, ensuring lexicographic order aligns with chronological order.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
v0.3.0— Fase B: la capa REST completaRelease de
devamain. Recoge todo lo construido desdev0.2.0, que cerró laFase A: 80 ficheros, 8.593 líneas.
Qué entra
POST /analyze/analyze, CORS y OpenAPIstdio|streamable-http)GET /toolspor handshake MCPPOST /tools/{name}/executey contrato de retorno estructuradoGET /historycon paginaciónRequisitos que cierra
R4 (API REST), R5 (catálogo y transparencia) y R9 (persistencia e
historial). Con ellos queda cerrado el hito H2.