hyungi_document_server/tests/test_crawl_cycle2_shapes.py

"""crawl-24x7 사이클 2 — 순수 함수/형태 회귀 테스트 (DB 불요).

Guardian 호출 형태 + fixture 응답 파싱 + 채널 정체성 + B-5 quirk.
fixture = tests/fixtures/guardian_open_platform_search_response.json
(2026-06-10 실키 live 박제, api-key 응답 본문 미포함 확인 — [[feedback_external_api_fixture_first]]).
"""

import json
import re
from pathlib import Path

from workers.news_collector import (
    _article_hash,
    _doc_identity,
    _guardian_request,
    _normalize_category,
)

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


def _make_source(**kw):
    """ORM 인스턴스 없이 속성만 흉내 (식별성 함수는 속성 접근만 사용)."""
    class S:
        pass
    s = S()
    s.source_channel = kw.get("source_channel", "news")
    s.parser_quirk = kw.get("parser_quirk")
    return s


class TestGuardianCallShape:
    def test_request_shape_matches_fixture_recipe(self):
        """fixture 박제 시 사용한 호출과 단일 source-of-truth 정합
        ([[feedback_fixture_first_call_shape]])."""
        endpoint, params = _guardian_request(
            "https://content.guardianapis.com/search?section=world", "KEY"
        )
        assert endpoint == "https://content.guardianapis.com/search"
        assert params["section"] == "world"
        assert params["show-fields"] == "bodyText,trailText"
        assert params["order-by"] == "newest"
        assert params["api-key"] == "KEY"

    def test_feed_url_query_overridden_by_fixed_fields(self):
        # feed_url 에 show-fields 가 잘못 박혀 있어도 고정 필드가 이긴다 (dict merge 순서)
        _, params = _guardian_request(
            "https://content.guardianapis.com/search?section=world&show-fields=headline", "K"
        )
        assert params["show-fields"] == "bodyText,trailText"


class TestGuardianFixtureParsing:
    def test_fixture_response_shape(self):
        payload = json.loads(FIXTURE.read_text())["response"]
        assert payload["status"] == "ok"
        assert payload["results"], "fixture 에 결과 0건"
        for item in payload["results"]:
            assert item["webTitle"].strip()
            assert item["webUrl"].startswith("https://")
            assert "webPublicationDate" in item
            assert "sectionName" in item
            fields = item.get("fields") or {}
            assert "bodyText" in fields and "trailText" in fields

    def test_fixture_bodytext_is_fulltext_grade(self):
        payload = json.loads(FIXTURE.read_text())["response"]
        # 전문 게이트(200자)를 fixture 가 통과해야 어댑터 is_full 경로가 산다
        assert any(len(i["fields"]["bodyText"]) >= 200 for i in payload["results"])

    def test_fixture_contains_no_api_key(self):
        assert "api-key" not in FIXTURE.read_text()


class TestChannelIdentity:
    def test_news_channel_unchanged(self):
        ident = _doc_identity(_make_source(source_channel="news"), "경향신문", "Society")
        assert ident == {
            "path_prefix": "news",
            "ai_domain": "News",
            "ai_tags": ["News/경향신문/Society"],
        }

    def test_crawl_channel_domain_identity(self):
        ident = _doc_identity(_make_source(source_channel="crawl"), "TWI", "Engineering")
        assert ident["path_prefix"] == "crawl"
        assert ident["ai_domain"] == "Engineering"
        assert ident["ai_tags"] == ["Engineering/TWI"]

    def test_crawl_channel_unknown_category_falls_back(self):
        ident = _doc_identity(_make_source(source_channel="crawl"), "X", "Other")
        assert ident["ai_domain"] == "Domain"

    def test_category_map_has_domain_axes(self):
        assert _normalize_category("안전") == "Safety"
        assert _normalize_category("Engineering") == "Engineering"
        assert _normalize_category("철학") == "Philosophy"


class TestSkipVideoQuirk:
    PATTERN = re.compile(r"/videos?/")

    def test_video_urls_match(self):
        assert self.PATTERN.search("https://psyche.co/videos/some-film")
        assert self.PATTERN.search("https://aeon.co/video/another")

    def test_article_urls_pass(self):
        assert not self.PATTERN.search("https://psyche.co/ideas/how-to-think")


class TestRedirect304Distinction:
    """httpx is_redirect 가 304(3xx 전체)에 True 라 redirect 로 오인 → 조건부 GET
    안정 피드가 'redirect 3회 초과'로 전멸하던 버그. has_redirect_location 으로 구분."""

    def test_304_is_not_a_redirect_location(self):
        import httpx
        r = httpx.Response(304, request=httpx.Request("GET", "https://x/"))
        assert r.is_redirect is True          # httpx 함정: 304 도 is_redirect
        assert r.has_redirect_location is False  # 우리가 써야 하는 정확한 판별

    def test_real_redirect_has_location(self):
        import httpx
        r = httpx.Response(301, headers={"location": "https://y/"},
                           request=httpx.Request("GET", "https://x/"))
        assert r.has_redirect_location is True

    def test_collector_uses_has_redirect_location(self):
        import inspect
        from workers import news_collector
        src = inspect.getsource(news_collector._fetch_rss)
        assert "has_redirect_location" in src
        assert "while resp.is_redirect" not in src  # 옛 버그 패턴 부재


class TestArticleHashStability:
    def test_static_corpus_hash_deterministic(self):
        a = _article_hash("Creep and Creep Failures", "static", "National Board 기술 아티클")
        b = _article_hash("Creep and Creep Failures", "static", "National Board 기술 아티클")
        assert a == b and len(a) == 32