refactor: tighten backend, flows, frontend, tests; push flows to Langflow

This commit is contained in:
Jonas Miederer 2026-09-28 15:58:00 +02:00
parent 6688fb08c6
commit 06ab25fb47
31 changed files with 783 additions and 358 deletions

View file

@ -5,6 +5,11 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
## [Unreleased]
### Changed
- Declutter: shared `BaseApiClient` (LFX/Langflow), single `get_db`, `model_validate` mappers, unified LFX/Langflow error mapping, stream logic moved to `runner.py`, `sync_full_pass` single orchestration (API+CLI), shared model base (`utcnow`/`ts_column`), unified auth counters, `httpx` in `deploy_flow.py`.
- Forms: shared `components/kirche/_form_core.py` single source (codegen-inlined for LFX), structured `visible_if` key alongside legacy markers, `LangflowClient.upsert_flow` + `scripts/push_flows_to_langflow.py` (6 canonical flows pushed to Langflow UI).
- Tests (full real LFX kept, 146 green): shared `conftest` helpers (`load_flow_fixture`, `flow_helpers`, `lfx_client`, `anon_client`), `tests/_harness.py` shared with smoke script, parametrized `filename_stem`.
- Frontend: shared `formatApiError`, `<ErrorAlert/>`, `<RunsTable/>`, `URLSearchParams` queries, localized run dates.
### Added
- Branched questionnaires: `[visible_if feld=Wert]` markers in a
`CustomInputForm` field's `info` (code editor) show the field only when