import json import pytest from estimator.__main__ import main from estimator.config import get_settings @pytest.fixture def env(monkeypatch, dsn): monkeypatch.setenv("EST_DATABASE__DSN", dsn) monkeypatch.setenv("EST_LLM__ANTHROPIC__API_KEY", "sk-test") monkeypatch.setenv("EST_RAG__PROVIDER", "hash") get_settings.cache_clear() yield get_settings.cache_clear() def write_csv(tmp_path, rows, name="refs.csv"): path = tmp_path / name lines = ["project,title,description,role,hours,hours_kind"] + [",".join(r) for r in rows] path.write_text("\n".join(lines) + "\n", encoding="utf-8") return path GOOD = [ ["P1", "Интеграция amoCRM", "сделки и контакты", "backend", "18", "actual"], ["P1", "Адаптер MAX", "", "backend", "12", "estimate"], ["P2", "Тестирование сценариев", "", "qa", "20", "actual"], ] def test_import_stats_search_delete_cycle(env, tmp_path, capsys): path = write_csv(tmp_path, GOOD) assert main(["rag", "import", str(path)]) == 0 out = capsys.readouterr().out assert "rows 3, valid 3, rejected 0" in out and "inserted 3" in out and "embedded 3 with hash-v1" in out assert main(["rag", "stats"]) == 0 out = capsys.readouterr().out assert "tasks 3 in 2 projects (2 with actual hours, 1 estimates)" in out assert "need (re)indexing: 0" in out and "retrieval is OFF" in out assert main(["rag", "search", "интеграция с CRM", "--min-similarity", "0.3"]) == 0 out = capsys.readouterr().out assert "Интеграция amoCRM" in out and "18 h actual" in out and "P1" in out assert main(["rag", "search", "ничего общего", "--min-similarity", "0.9"]) == 0 assert "nothing above the similarity threshold" in capsys.readouterr().out assert main(["rag", "search", "тестирование", "--role", "backend", "--min-similarity", "0.2"]) == 0 assert "Тестирование сценариев" not in capsys.readouterr().out # a QA task is not shown for a backend query assert main(["rag", "delete", "--project", "P2"]) == 0 assert "deleted 1 tasks of 'P2'" in capsys.readouterr().out main(["rag", "stats"]) assert "tasks 2 in 1 projects" in capsys.readouterr().out def test_dry_run_validates_without_writing(env, tmp_path, capsys): path = write_csv(tmp_path, GOOD) assert main(["rag", "import", str(path), "--dry-run"]) == 0 assert "dry run: nothing was written" in capsys.readouterr().out main(["rag", "stats"]) assert "tasks 0 in 0 projects" in capsys.readouterr().out def test_bad_rows_are_listed_and_do_not_block_the_good_ones(env, tmp_path, capsys): path = write_csv(tmp_path, [*GOOD, ["P1", "Плохая", "", "wizard", "5", ""], ["P1", "", "", "", "5", ""]]) assert main(["rag", "import", str(path)]) == 0 out = capsys.readouterr().out assert "row 4: unknown role 'wizard'" in out and "row 5: title" in out and "valid 3, rejected 2" in out def test_a_file_with_only_bad_rows_fails(env, tmp_path, capsys): path = write_csv(tmp_path, [["P1", "", "", "", "x", ""]]) assert main(["rag", "import", str(path)]) == 1 def test_default_project_and_json_input(env, tmp_path, capsys): path = tmp_path / "refs.json" path.write_text(json.dumps([{"title": "Задача без проекта", "hours": 4}]), encoding="utf-8") assert main(["rag", "import", str(path)]) == 1 # no project in the file and no default: nothing valid assert "project is required" in capsys.readouterr().out assert main(["rag", "import", str(path), "--project", "Старый проект"]) == 0 assert "inserted 1" in capsys.readouterr().out def test_without_an_embeddings_key_the_openai_provider_is_refused(monkeypatch, dsn, capsys): monkeypatch.setenv("EST_DATABASE__DSN", dsn) monkeypatch.setenv("EST_LLM__ANTHROPIC__API_KEY", "sk-test") get_settings.cache_clear() try: assert main(["rag", "stats"]) == 2 finally: get_settings.cache_clear() assert "No embeddings API key" in capsys.readouterr().err def test_the_commands_work_while_retrieval_is_still_switched_off(env, tmp_path, capsys): """An administrator loads the corpus first and enables retrieval afterwards.""" main(["rag", "import", str(write_csv(tmp_path, GOOD))]) capsys.readouterr() main(["rag", "stats"]) assert "retrieval is OFF" in capsys.readouterr().out