"""B-2 KOSHA 사망사고 속보(callApiId=1040) — 순수 파서 fixture 테스트 (plan safety-library-1).

fixture = 2026-06-13 data.go.kr 라이브 박제 (serviceKey 응답 본문 미포함 확인,
tests/fixtures/kosha_fatal_response.json). _fatal_fields/_items 는 순수 함수라 DB/httpx
호출 없이 검증 — [[feedback_external_api_fixture_first]].
"""

import json
from datetime import date
from pathlib import Path

from workers.kosha_collector import _fatal_fields, _items

FIXTURE = Path(__file__).parent / "fixtures" / "kosha_fatal_response.json"


def _payload() -> dict:
    return json.loads(FIXTURE.read_text(encoding="utf-8"))


def test_items_envelope_parse():
    """body.items.item 봉투 파싱 — 재해사례와 동일 envelope."""
    items = _items(_payload())
    assert len(items) == 3
    assert all({"arno", "keyword", "contents"} <= set(it) for it in items)


def test_fatal_fields_basic_mapping():
    item = _items(_payload())[0]
    f = _fatal_fields(item)
    assert f is not None
    assert f["arno"] == "20260611111536KIZXJ8"
    assert f["title"].startswith("[6/9, 부산 사상구]")
    # HTML 태그 + 이미지 서버 URL 노이즈 완전 제거 (검색/임베딩 본문 정화)
    assert "<" not in f["text"]
    assert "portal.kosha.or.kr" not in f["text"]
    assert "data-filename" not in f["text"]
    # 본문 텍스트는 보존
    assert "(사망 2명)" in f["text"]
    assert "베란다 난간" in f["text"]
    # published_date = arno 접두 8자리(KST 등록일), reg_dt = 14자리 등록시각 원문
    assert f["published_date"] == date(2026, 6, 11)
    assert f["reg_dt"] == "20260611111536"


def test_fatal_fields_all_three_items_well_formed():
    for item in _items(_payload()):
        f = _fatal_fields(item)
        assert f is not None
        assert f["published_date"] == date(2026, 6, 11)  # 3건 모두 06-11 등록
        assert f["reg_dt"] is not None
        assert f["text"] and "<" not in f["text"]


def test_fatal_fields_skips_missing_required():
    assert _fatal_fields({"arno": "20260611111536XX", "contents": "x"}) is None  # keyword 부재
    assert _fatal_fields({"keyword": "제목만", "contents": "x"}) is None           # arno 부재
    assert _fatal_fields({"arno": "  ", "keyword": "  ", "contents": "x"}) is None  # 공백뿐


def test_fatal_fields_malformed_arno_date_is_fail_quiet():
    # arno 접두가 8자리 날짜로 안 풀리면 published_date/reg_dt = None (보조 축이라 fail-quiet)
    f = _fatal_fields({"arno": "ABC123", "keyword": "제목", "contents": "<p>본문</p>"})
    assert f is not None
    assert f["published_date"] is None
    assert f["reg_dt"] is None
    assert f["text"] == "본문"