arm-tabular-parser/api/tests/unit/parser/test_cli.py

"""CLI разбора файла `python -m src.parser.cli` (Task 10; spec tabular-recognition-service, README «Запуск на файле»).

Коды возврата: 0 — `done`, 1 — `needs_check`, 2 — `error`, 3 — файл не найден / не читается / ошибка использования.
Синтетические книги собираются openpyxl в памяти (`test_book_parser.xlsx`) и пишутся в `tmp_path`; реальный файл —
запуск как модуля через subprocess (маркер `fixtures`).
"""

import json
import os
import subprocess
import sys
from pathlib import Path
from typing import Any

import pytest
from result_contract import schema_errors
from test_book_parser import mini_osv, xlsx

from src.contract import RecognitionResult
from src.parser import cli

API_DIR = Path(__file__).resolve().parents[3]
REAL_OSV_66 = Path("osv") / "rb" / "2024-FY" / "ОСВ_66 за 2024.xls"
SPOILED = (60, None, 20, 10, 71, None)
EXIT_DONE, EXIT_NEEDS_CHECK, EXIT_ERROR, EXIT_USAGE = 0, 1, 2, 3


def _write(tmp_path: Path, name: str, data: bytes) -> Path:
    path = tmp_path / name
    path.write_bytes(data)
    return path


def _run(capsys: pytest.CaptureFixture[str], *argv: str) -> tuple[int, str, str]:
    code = cli.main(list(argv))
    captured = capsys.readouterr()
    return code, captured.out, captured.err


def test_done_book_returns_0_and_prints_result_json(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    path = _write(tmp_path, "osv.xlsx", xlsx({"ОСВ 60": mini_osv()}))

    code, out, _err = _run(capsys, str(path))

    assert code == EXIT_DONE
    payload: dict[str, Any] = json.loads(out)
    assert payload["status"] == "done"
    assert payload["schema_version"] == "2.0"
    assert schema_errors(payload) == []
    assert RecognitionResult.model_validate(payload).to_wire() == payload
    assert payload["documents"][0]["header"]["account_code_raw"] == "60"
    assert "ОСВ 60" in out  # не ASCII-экранирование
    assert out.endswith("\n")


def test_output_is_indented_and_keeps_model_field_order(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    path = _write(tmp_path, "osv.xlsx", xlsx({"ОСВ 60": mini_osv()}))

    _code, out, _err = _run(capsys, str(path))

    assert out.startswith('{\n  "schema_version": "2.0",\n')
    keys = list(json.loads(out))
    assert keys != sorted(keys)  # sort_keys не используется: порядок полей модели значим для golden


def test_money_is_printed_as_strings(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    path = _write(tmp_path, "osv.xlsx", xlsx({"ОСВ 60": mini_osv()}))

    _code, out, _err = _run(capsys, str(path))

    row = json.loads(out)["documents"][0]["rows"][1]
    assert row["opening_debit"] == "60.00"


def test_needs_check_book_returns_1(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    path = _write(tmp_path, "spoiled.xlsx", xlsx({"ОСВ 60": mini_osv(SPOILED)}))

    code, out, _err = _run(capsys, str(path))

    assert code == EXIT_NEEDS_CHECK
    payload = json.loads(out)
    assert payload["status"] == "needs_check"
    assert schema_errors(payload) == []


def test_error_outcome_returns_2_with_error_json(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    path = _write(tmp_path, "broken.xlsx", b"PK\x03\x04 broken zip")

    code, out, err = _run(capsys, str(path))

    assert code == EXIT_ERROR
    payload = json.loads(out)
    assert payload["status"] == "error"
    assert payload["error"]["code"] == "UNREADABLE"
    assert payload["error"]["message"]
    assert "UNREADABLE" in err


def test_missing_file_returns_3(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    code, out, err = _run(capsys, str(tmp_path / "nope.xlsx"))

    assert code == EXIT_USAGE
    assert out == ""
    assert "nope.xlsx" in err


def test_directory_instead_of_file_returns_3(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    code, out, _err = _run(capsys, str(tmp_path))

    assert code == EXIT_USAGE
    assert out == ""


def test_unknown_extension_without_file_type_returns_3(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    path = _write(tmp_path, "osv.dat", xlsx({"ОСВ 60": mini_osv()}))

    code, out, err = _run(capsys, str(path))

    assert code == EXIT_USAGE
    assert out == ""
    assert "--file-type" in err


def test_explicit_file_type_overrides_extension(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    path = _write(tmp_path, "osv.dat", xlsx({"ОСВ 60": mini_osv()}))

    code, out, _err = _run(capsys, str(path), "--file-type", "XLSX")

    assert code == EXIT_DONE
    assert json.loads(out)["status"] == "done"


def test_extension_is_case_insensitive(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    path = _write(tmp_path, "OSV.XLSX", xlsx({"ОСВ 60": mini_osv()}))

    code, _out, _err = _run(capsys, str(path))

    assert code == EXIT_DONE


def test_invalid_file_type_value_returns_3_not_argparse_2(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    path = _write(tmp_path, "osv.xlsx", xlsx({"ОСВ 60": mini_osv()}))

    code, out, err = _run(capsys, str(path), "--file-type", "PDF")

    assert code == EXIT_USAGE
    assert out == ""
    assert "PDF" in err


def test_no_arguments_returns_3(capsys: pytest.CaptureFixture[str]) -> None:
    code, out, _err = _run(capsys)

    assert code == EXIT_USAGE
    assert out == ""


def test_output_is_deterministic_except_duration(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
    path = _write(tmp_path, "osv.xlsx", xlsx({"ОСВ 60": mini_osv()}))

    first = json.loads(_run(capsys, str(path))[1])
    second = json.loads(_run(capsys, str(path))[1])

    for payload in (first, second):
        payload["metrics"].pop("duration_ms")
    assert first == second


@pytest.mark.fixtures
def test_module_run_on_real_file_prints_valid_json(fixtures_dir: Path) -> None:
    """`python -m src.parser.cli <файл>` из `api/`: stdout — JSON результата по схеме, код возврата 0."""
    completed = subprocess.run(
        [sys.executable, "-m", "src.parser.cli", str(fixtures_dir / REAL_OSV_66)],
        cwd=API_DIR,
        env={**os.environ, "ENV_FILE": ".env.test"},  # импорт пакета `src` читает настройки сервиса
        capture_output=True,
        check=False,
        timeout=120,
    )

    assert completed.returncode == EXIT_DONE, completed.stderr.decode("utf-8", "replace")
    payload = json.loads(completed.stdout.decode("utf-8"))
    assert payload["status"] == "done"
    assert schema_errors(payload) == []
    assert payload["documents"][0]["header"]["account_code_raw"] == "66"