o2o-site-AEO/backend/tests/test_vision_api.py

"""사진 분석(VISION) — 잡 배선과 ★ 신뢰도 게이트.

★ 이 잡이 절대 하면 안 되는 것:
   - 신뢰도 낮은 라벨을 자동 반영하는 것 (사람 확인 큐를 건너뛰는 것)
   - 한 장 실패로 잡 전체를 실패시키는 것
   - 같은 사진을 재분석해 요금을 두 번 내는 것
"""
import uuid

import pytest
from sqlalchemy import text

from common.enums import ErrorType, JobStatus, JobType, MediaStatus, PlaceCategory, SourceType
from crud.job_crud import JobQueue
from services.external import gemini
from worker.handlers import build_handler
from worker.runner import Worker


async def _place_with_media(client, h, db_engine, n=3, kakao="v1"):
    pid = (await client.post("/v1/place", headers=h, json={"name": "비전펜션", "category": 1})).json()["place"]["place_id"]
    await client.post(f"/v1/place/{pid}/verify", headers=h, json={"external_place_id": kakao, "source": 1})
    async with db_engine.begin() as c:
        for i in range(n):
            await c.execute(
                text("INSERT INTO media (media_id, place_id, url, origin_url, source_type, status, sort_order) "
                     "VALUES (:m, :p, :u, :u, :s, :st, :o)"),
                {"m": uuid.uuid4(), "p": uuid.UUID(pid), "u": f"https://cdn.test/{pid}/{i}.jpg",
                 "s": SourceType.CRAWL.value, "st": MediaStatus.PENDING_REVIEW.value, "o": i},
            )
    return pid


async def _media_rows(db_engine, pid):
    async with db_engine.begin() as c:
        return (await c.execute(
            text("SELECT origin_url, label, alt_text, vision_confidence, status FROM media "
                 "WHERE place_id = :p ORDER BY sort_order"),
            {"p": uuid.UUID(pid)},
        )).all()


def _fake_analyze(results):
    """gemini.analyze_images 를 대체 — 실제 API 를 때리지 않는다."""
    async def _fn(images, **kw):
        out = []
        for img in images:
            r = results.get(img.origin_url)
            out.append(r if r else gemini.VisionResult(
                origin_url=img.origin_url, label=None, alt_text=None,
                confidence=0.0, needs_review=True, ok=False, error="no result"))
        return out
    return _fn


async def test_high_confidence_is_auto_applied(auth_headers, client, db_engine, monkeypatch):
    """검증: 신뢰도 높은 분석 결과.
    기대결과: 라벨·alt 가 반영되고 status 가 APPROVED 로 올라간다."""
    h = await auth_headers("u1")
    pid = await _place_with_media(client, h, db_engine, n=2)
    urls = [r[0] for r in await _media_rows(db_engine, pid)]

    monkeypatch.setattr(gemini, "is_configured", lambda: True)
    monkeypatch.setattr(gemini, "analyze_images", _fake_analyze({
        urls[0]: gemini.VisionResult(urls[0], "A동 침실", "침대와 창문이 있는 객실", 0.93, False, True, None),
        urls[1]: gemini.VisionResult(urls[1], "외관", "2층 건물 외관", 0.88, False, True, None),
    }))

    job_id = (await client.post(f"/v1/place/{pid}/vision", headers=h, json={})).json()["job_id"]
    await Worker("w", JobQueue(), build_handler(), job_deadline_sec=30).process_one()

    job = (await client.get(f"/v1/job/{job_id}", headers=h)).json()["job"]
    assert job["status"] == JobStatus.DONE.value, job.get("last_error")
    assert job["result"]["approved"] == 2

    rows = await _media_rows(db_engine, pid)
    assert rows[0][1] == "A동 침실" and rows[0][2] == "침대와 창문이 있는 객실"
    assert all(r[4] == MediaStatus.APPROVED.value for r in rows)


async def test_low_confidence_goes_to_review_queue(auth_headers, client, db_engine, monkeypatch):
    """검증: 신뢰도가 임계값 미만인 결과.
    기대결과: ★ 라벨은 저장되지만 status 는 PENDING_REVIEW 로 남는다 — 자동 반영하지 않는다."""
    h = await auth_headers("u1")
    pid = await _place_with_media(client, h, db_engine, n=1, kakao="v2")
    url = (await _media_rows(db_engine, pid))[0][0]

    monkeypatch.setattr(gemini, "is_configured", lambda: True)
    monkeypatch.setattr(gemini, "analyze_images", _fake_analyze({
        url: gemini.VisionResult(url, "수영장", "물이 있는 공간", 0.31, True, True, None),
    }))

    job_id = (await client.post(f"/v1/place/{pid}/vision", headers=h, json={})).json()["job_id"]
    await Worker("w", JobQueue(), build_handler(), job_deadline_sec=30).process_one()

    job = (await client.get(f"/v1/job/{job_id}", headers=h)).json()["job"]
    assert job["result"]["needs_review"] == 1
    assert job["result"]["approved"] == 0

    row = (await _media_rows(db_engine, pid))[0]
    assert row[1] == "수영장", "사람이 고칠 재료는 저장돼야 한다"
    assert row[4] == MediaStatus.PENDING_REVIEW.value, "★ 신뢰도 낮은 결과가 자동 승인됐다"


async def test_partial_failure_does_not_fail_the_job(auth_headers, client, db_engine, monkeypatch):
    """검증: 3장 중 1장만 분석에 실패한다.
    기대결과: 잡은 DONE, 실패한 장만 확인 큐에 남는다 — 한 장이 나머지를 죽이지 않는다."""
    h = await auth_headers("u1")
    pid = await _place_with_media(client, h, db_engine, n=3, kakao="v3")
    urls = [r[0] for r in await _media_rows(db_engine, pid)]

    monkeypatch.setattr(gemini, "is_configured", lambda: True)
    monkeypatch.setattr(gemini, "analyze_images", _fake_analyze({
        urls[0]: gemini.VisionResult(urls[0], "외관", "건물 외관", 0.9, False, True, None),
        urls[1]: gemini.VisionResult(urls[1], None, None, 0.0, True, False, "download failed"),
        urls[2]: gemini.VisionResult(urls[2], "거실", "소파가 있는 거실", 0.85, False, True, None),
    }))

    job_id = (await client.post(f"/v1/place/{pid}/vision", headers=h, json={})).json()["job_id"]
    await Worker("w", JobQueue(), build_handler(), job_deadline_sec=30).process_one()

    job = (await client.get(f"/v1/job/{job_id}", headers=h)).json()["job"]
    assert job["status"] == JobStatus.DONE.value
    assert job["result"]["approved"] == 2
    assert job["result"]["failed"] == 1


async def test_second_run_skips_already_analyzed(auth_headers, client, db_engine, monkeypatch):
    """검증: 분석이 끝난 사업장에 다시 분석을 건다.
    기대결과: MEDIA_NOT_FOUND — 같은 사진 재분석은 요금만 나간다. force 로만 다시 돈다."""
    h = await auth_headers("u1")
    pid = await _place_with_media(client, h, db_engine, n=1, kakao="v4")
    url = (await _media_rows(db_engine, pid))[0][0]

    monkeypatch.setattr(gemini, "is_configured", lambda: True)
    monkeypatch.setattr(gemini, "analyze_images", _fake_analyze({
        url: gemini.VisionResult(url, "외관", "건물 외관", 0.9, False, True, None),
    }))
    await client.post(f"/v1/place/{pid}/vision", headers=h, json={})
    await Worker("w", JobQueue(), build_handler(), job_deadline_sec=30).process_one()

    r = await client.post(f"/v1/place/{pid}/vision", headers=h, json={})
    assert r.json()["result"]["code"] == ErrorType.MEDIA_NOT_FOUND.value

    forced = await client.post(f"/v1/place/{pid}/vision", headers=h, json={"force": True})
    assert forced.json()["result"]["success"] is True


async def test_vision_requires_api_key(auth_headers, client, db_engine, monkeypatch):
    """검증: GEMINI_API_KEY 없이 분석을 건다.
    기대결과: GENERATOR_NOT_CONFIGURED — 잡을 만들지 않는다(만들어봐야 DEAD 로 간다)."""
    h = await auth_headers("u1")
    pid = await _place_with_media(client, h, db_engine, n=1, kakao="v5")
    monkeypatch.setattr(gemini, "is_configured", lambda: False)

    r = await client.post(f"/v1/place/{pid}/vision", headers=h, json={})
    assert r.json()["result"]["code"] == ErrorType.GENERATOR_NOT_CONFIGURED.value


async def test_collect_chains_vision_job(auth_headers, client, monkeypatch):
    """검증: 수집이 사진을 저장하면 사진 분석 잡이 이어서 걸리는가.
    기대결과: 수집 잡 result 에 vision_job_id 가 담기고 그 잡이 큐에 있다 — 파이프라인이 안 끊긴다."""
    from services.collector import MockAdapter

    monkeypatch.setattr(gemini, "is_configured", lambda: True)
    h = await auth_headers("u1")
    pid = (await client.post("/v1/place", headers=h, json={"name": "체인펜션", "category": 1})).json()["place"]["place_id"]
    await client.post(f"/v1/place/{pid}/verify", headers=h, json={"external_place_id": "v6", "source": 1})
    url = MockAdapter.url_for(PlaceCategory.LODGING, pid, channel="naver_place")
    lid = (await client.post(f"/v1/place/{pid}/link", headers=h, json={"channel": 3, "url": url})).json()["link"]["link_id"]
    await client.post(f"/v1/place/{pid}/link/{lid}/confirm", headers=h)

    job_id = (await client.post(f"/v1/place/{pid}/collect", headers=h, json={})).json()["job_id"]
    await Worker("w", JobQueue(), build_handler(), job_deadline_sec=120).process_one()

    job = (await client.get(f"/v1/job/{job_id}", headers=h)).json()["job"]
    vision_id = job["result"].get("vision_job_id")
    assert vision_id, "사진을 저장했는데 분석 잡이 안 걸렸다"
    chained = await JobQueue().get(vision_id)
    assert chained["job_type"] == JobType.VISION.value
    assert chained["status"] == JobStatus.PENDING.value