tests/test_ajimee_cli.py

from __future__ import annotations

import io
import json

from deberta_ime.ajimee import AjimeeArtifact
from deberta_ime.ajimee_cli import run
from deberta_ime.mozc import build_mozc_index


def test_ajimee_cli_writes_aggregate_boundary_audit(tmp_path) -> None:
    dictionary_dir = tmp_path / "dictionary_oss"
    dictionary_dir.mkdir()
    (dictionary_dir / "dictionary00.txt").write_text(
        "はし\t1\t1\t3500\t箸\nはし\t1\t1\t3800\t橋\n",
        encoding="utf-8",
    )
    index_path = tmp_path / "mozc.sqlite3"
    build_mozc_index(dictionary_dir, index_path, source_revision="fixture")
    items = (
        {
            "index": "1",
            "context_text": "川に架かる",
            "input": "ハシ",
            "expected_output": ["橋"],
            "original_text": "川に架かる橋",
            "splitted_input_for_limited_input_length": [],
        },
    )
    artifact = AjimeeArtifact(
        path=tmp_path / "items.json",
        url="https://example.invalid/items.json",
        revision="fixture-revision",
        sha256="d" * 64,
        size_bytes=100,
        items=items,
    )
    stdout = io.StringIO()

    exit_code = run(
        [
            "--index",
            str(index_path),
            "--output-dir",
            str(tmp_path / "outputs"),
            "--stem",
            "audit_fixture",
        ],
        stdout=stdout,
        artifact_loader=lambda data_dir: artifact,
    )

    summary = json.loads(stdout.getvalue())
    report = json.loads((tmp_path / "outputs" / "audit_fixture.json").read_text("utf-8"))
    assert exit_code == 0
    assert summary["ok"] is True
    assert report["status"] == "LOCAL_INPUT_BOUNDARY_AUDIT"
    assert report["audit"]["sequence_accuracy"] is None
    assert (tmp_path / "outputs" / "audit_fixture.md").is_file()