test: +13 more tests (99→112), incl. live Ollama smoke (ignored by default)
text_utils (was 5, now 12 tests, +7)
- empty_input_yields_empty_output
- handles_usa_uk_sos_brands — verifies all brand replacements survive
- collapse_two_letter_acronym_not_changed — 2-letter dotted patterns stay
- short_url_under_threshold_not_stripped — gate is len > 7
- smart_quotes_softened — «»/"" stripped
- collapses_repeated_whitespace — no 3+ spaces in row
- idempotent_when_already_clean — sanitize(sanitize(x)) == sanitize(x)
llm::client (was 4, now 10 tests + 1 ignored, +6 + 1 smoke)
- chat_message_serde_round_trip — pins JSON shape {role, content}
- ollama_client_has_no_api_key — uses LlmBackend::Ollama
- groq_client_carries_token — verifies token + default model
- ollama_default_model_can_be_overridden — OLLAMA_MODEL env honoured
- parses_response_with_unicode_content — Russian Cyrillic survives JSON
- empty_choices_yields_empty_response_err — graceful on []
- ollama_smoke (#[ignore]) — actual API call when Ollama is running locally.
Run with: cargo test --lib llm::client::tests::ollama_smoke -- --ignored
Total: 112 passing + 1 ignored. Smoke test verified locally — Ollama not
running (expected), test correctly skipped.
Build: cargo build --release -p jarvis-app -p jarvis-gui green.
20b6b06b5d