Replace the single [llm] block in settings.toml with a curated catalog
of LLM "backends" (provider + model + sampling params + system prompt)
in server/src/analyze/backend.rs. Two backends ship initially:
gpt_oss_120b (default, reasoning_effort=medium) and llama_3_1_405b
(uses response_format: json_schema to sidestep the <|eot_id|> leak).
The case page now renders one submit button per available backend; the
clicked button's name=backend value rides through AnalysisInput.backend_id
to the worker, which looks it up via find_backend() with default fallback.
A submitted unknown backend id is rejected as 400. .analysis_failed.json
records the failed backend and the failure banner surfaces its label.
The API key now comes from IONOS_API_KEY (env), not settings.toml; the
[llm] section is read into a discarded field so old configs still load.
Backends without a satisfied requires_api_key are filtered from the UI
(any_backend_available()). Three end-to-end tests that mocked the LLM
endpoint via settings.llm.url are #[ignore]'d with a clear note — they
need per-test backend injection (Arc<Vec<LlmBackend>> through the worker)
before they can be re-enabled.
Config sheds its 14 Bucket-B fields (retention, whisper, ollama, llm) and
gains a sibling Arc<Settings> in AppState, loaded from settings.toml at
startup via Settings::load_or_default. Workers (transcribe, analyze,
recovery, auto_trigger) and routes (health, bulk, case_actions,
user_web) take settings as a separate parameter or State<> extractor;
Config::llm_configured moves to Settings::llm_configured.
TestConfig::build_pair() returns (Arc<Config>, Arc<Settings>); the new
create_router_with_settings / create_router_and_session_store_with_settings
helpers wire both into integration tests that exercise LLM/Whisper/Ollama.
The legacy single-Arc create_router falls back to Settings::default()
so the 17 non-LLM tests stay untouched.
Verified: cargo build clean, clippy --all-targets clean, 318 tests pass.
Finishes the lift of shared helpers into `tests/common/`. Covered:
- health, web, oneliners_api, upload (31 tests; upload exercises the
new `multipart_upload_body` helper driven by doctate_common field
constants)
- sse_integration, sse_cleanup, transcribe, oneliner_heal_decoupled,
silent_case_empty, failed_only_case_empty_oneliner,
transient_failure_retries (24 tests)
Also applied cargo fmt across the test tree and fixed one clippy
needless_borrows_for_generic_args warning in analyze_test.
All 387 tests pass; 3 ignored (as before).
Side effect: health_test previously used a hardcoded `/tmp/doctate-test`
data path, which parallel `cargo test` runs could collide on. The
migration replaces it with the common unique-tmpdir pattern, removing
a latent flake.
Lift duplicated test-harness code (User factories, TestConfig builder,
login/CSRF flow, form/json helpers, case-directory seeding, URL path
builders) into `server/tests/common/` so future API changes touch one
file instead of every integration test.
Migrated batches: csrf_attack, auth, login, magic_link, security_headers
(28 tests); analyze, case_page, delete_recording, close_watermark,
retention_sweep (67 tests). All 95 tests still green.
Net line delta across these 10 files: +1113 / -1896 (~780 lines removed),
plus ~500 lines of new shared infrastructure under tests/common/.
Brummel