backend/tests/test_datasets_api.py

"""Tests for the admin dataset endpoints (spec §2) incl. transactional cascade purge."""
import os
from pathlib import Path

from sqlalchemy import select

from app.storage import get_storage

from app.db import SessionLocal
from app.models import Case, Dataset, Embedding, KnownDog, Match, Picture, UnknownDog, User
from app.models.base import SubjectType
from app.services.batch_loader import load_dataset
from scripts.make_sample_images import make_image
from scripts.prepare_test_data import prepare
from tests.conftest import auth


def _build_input(tmp: Path, layout: dict[str, int], *, identical: bool = False) -> Path:
    root = tmp / "input"
    for folder, count in layout.items():
        d = root / folder
        d.mkdir(parents=True)
        shared = make_image(abs(hash(folder)) % 1000)
        for i in range(count):
            (d / f"img{i}.jpg").write_bytes(shared if identical else make_image((abs(hash(folder)) + i) % 1000))
    return root


def _load(folder: Path, csv_name: str, dtype: str, name: str, **kw) -> int:
    db = SessionLocal()
    try:
        res = load_dataset(db, folder=folder, dataset_type=dtype, name=name,
                           description=None, csv_path=folder / csv_name, **kw)
        return res.dataset_id
    finally:
        db.close()


def test_list_requires_admin(client, owner_token):
    assert client.get("/admin/datasets", headers=auth(owner_token)).status_code == 403


def test_list_and_detail(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2, "dogB": 2})  # 2 folders -> 2 dogs
    prepare(root, root, holdout=1, seed=1)
    dsid = _load(root, "known_dogs.csv", "known", "My Known", mark_lost=True, mark_lost_pct=100)

    lst = client.get("/admin/datasets", headers=auth(admin_token))
    assert lst.status_code == 200
    row = next(d for d in lst.json() if d["id"] == dsid)
    assert row["name"] == "My Known"
    assert row["type"] == "known"
    assert row["dog_count"] == 2
    assert row["case_count"] == 2  # mark_lost opened a lost case per dog

    detail = client.get(f"/admin/datasets/{dsid}", headers=auth(admin_token))
    assert detail.status_code == 200
    stats = detail.json()["stats"]
    assert stats["known_dog_count"] == 2
    assert stats["case_count"] == 2
    assert stats["picture_count"] == 2


def test_image_dataset_lookup(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2})
    prepare(root, root, holdout=1, seed=1)
    dsid = _load(root, "known_dogs.csv", "known", "K")
    db = SessionLocal()
    pic_id = db.execute(select(Picture.id)).scalars().first()
    db.close()
    r = client.get(f"/admin/image/{pic_id}/dataset", headers=auth(admin_token))
    assert r.status_code == 200
    assert r.json()["dataset_id"] == dsid
    assert r.json()["dataset_name"] == "K"


def test_embed_all_idempotent_then_reembeds(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2})
    prepare(root, root, holdout=1, seed=1)
    dsid = _load(root, "known_dogs.csv", "known", "K")

    # Already embedded during load -> nothing new.
    r1 = client.post(f"/admin/datasets/{dsid}/embed-all", headers=auth(admin_token)).json()
    assert r1["embedded"] == 0
    assert r1["skipped"] == r1["total_pictures"] >= 1

    # Wipe embeddings, then embed-all should regenerate them.
    db = SessionLocal()
    db.execute(Embedding.__table__.delete())
    db.commit()
    db.close()
    r2 = client.post(f"/admin/datasets/{dsid}/embed-all", headers=auth(admin_token)).json()
    assert r2["embedded"] == r2["total_pictures"] >= 1
    assert r2["errors"] == 0


def test_embed_all_reports_progress(tmp_path):
    """The service streams progress events and commits incrementally over the run."""
    from app.models import Dataset
    from app.services import datasets as ds

    root = _build_input(tmp_path, {"dogA": 3, "dogB": 3})
    prepare(root, root, holdout=1, seed=1)
    dsid = _load(root, "known_dogs.csv", "known", "K", skip_embeddings=True)

    events: list[dict] = []
    db = SessionLocal()
    try:
        dataset = db.get(Dataset, dsid)
        result = ds.embed_all(db, dataset, progress=events.append, commit_every=2)
    finally:
        db.close()

    assert result["embedded"] == result["total_pictures"] >= 4
    assert events[0]["processed"] == 0                       # start
    assert events[-1]["processed"] == result["total_pictures"]  # finish
    assert all(e["total"] == result["total_pictures"] for e in events)
    assert [e["processed"] for e in events] == sorted(e["processed"] for e in events)  # monotonic


def test_embed_all_job_runs_in_background(client, admin_token, tmp_path):
    """POST embed-all-job returns a job id; polling it reaches done with the encode counts."""
    import time

    root = _build_input(tmp_path, {"dogA": 2})
    prepare(root, root, holdout=1, seed=1)
    dsid = _load(root, "known_dogs.csv", "known", "K", skip_embeddings=True)

    start = client.post(f"/admin/datasets/{dsid}/embed-all-job", headers=auth(admin_token))
    assert start.status_code == 202
    job_id = start.json()["job_id"]

    job = None
    for _ in range(50):  # up to ~5s
        job = client.get(f"/admin/jobs/{job_id}", headers=auth(admin_token)).json()
        if job["status"] in ("done", "error"):
            break
        time.sleep(0.1)
    assert job and job["status"] == "done", job
    assert job["result"]["embedded"] == job["result"]["total_pictures"] >= 1
    assert client.post("/admin/datasets/9999/embed-all-job", headers=auth(admin_token)).status_code == 404


def test_match_endpoint_creates_matches(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2}, identical=True)
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K", mark_lost=True, mark_lost_pct=100)
    u_id = _load(root, "found_dogs.csv", "unknown", "U")

    r = client.post(f"/admin/datasets/{u_id}/match", headers=auth(admin_token))
    assert r.status_code == 200
    assert r.json()["matches_created"] >= 1


def test_purge_cascade_deletes_everything(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2}, identical=True)
    prepare(root, root, holdout=1, seed=1)
    k_id = _load(root, "known_dogs.csv", "known", "K", mark_lost=True, mark_lost_pct=100)
    u_id = _load(root, "found_dogs.csv", "unknown", "U")
    client.post(f"/admin/datasets/{u_id}/match", headers=auth(admin_token))

    # Sanity: matches exist before purge.
    db = SessionLocal()
    assert db.execute(select(Match)).scalars().first() is not None
    db.close()

    # Purge the unknown dataset: its unknown dog, found case, and produced matches go.
    res = client.delete(f"/admin/datasets/{u_id}", headers=auth(admin_token))
    assert res.status_code == 200
    deleted = res.json()["deleted"]
    assert deleted["unknown_dogs"] == 1
    assert deleted["cases"] == 1
    assert deleted["matches"] >= 1
    assert deleted["datasets"] == 1

    db = SessionLocal()
    assert db.execute(select(UnknownDog).where(UnknownDog.dataset_id == u_id)).scalars().first() is None
    assert db.execute(select(Match)).scalars().first() is None
    assert db.get(Dataset, u_id) is None
    db.close()

    # Purge the known dataset: owner users, known dogs, lost cases go.
    res2 = client.delete(f"/admin/datasets/{k_id}", headers=auth(admin_token))
    deleted2 = res2.json()["deleted"]
    # dogA has 2 images -> 1 registration (known) + 1 holdout (found).
    assert deleted2["known_dogs"] == 1
    assert deleted2["users"] == 1  # the single owner
    assert deleted2["cases"] == 1  # the lost case from mark_lost

    db = SessionLocal()
    assert db.execute(select(KnownDog).where(KnownDog.dataset_id == k_id)).scalars().first() is None
    assert db.execute(select(User).where(User.dataset_id == k_id)).scalars().first() is None
    # No orphan pictures left for the purged dogs.
    assert db.execute(select(Picture)).scalars().first() is None
    db.close()


def test_admin_all_cases_and_owners(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2}, identical=True)
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K", mark_lost=True, mark_lost_pct=100)

    cases = client.get("/admin/cases", headers=auth(admin_token)).json()
    assert cases["total"] >= 1
    row = next(c for c in cases["items"] if c["type"] == "lost")
    assert row["dog"]["kind"] == "known" and row["person"] is not None
    assert "match_count" in row
    # filter by type
    assert all(c["type"] == "lost" for c in
               client.get("/admin/cases?kind=lost", headers=auth(admin_token)).json()["items"])

    owners = client.get("/admin/owners", headers=auth(admin_token)).json()
    assert owners["total"] >= 1
    assert owners["items"][0]["dog_count"] >= 1 and owners["items"][0]["dogs"]
    # non-admin denied
    assert client.get("/admin/cases").status_code in (401, 403)


def test_admin_set_dog_status(client, admin_token, owner_token):
    # A found dog dropped at a shelter -> starts 'at_shelter'.
    client.post(
        "/cases/found",
        data={
            "event_zip": "20001", "event_date": "2026-06-03", "finder_email": "f@example.com",
            "current_location_detail": "Happy Paws Shelter",
        },
        files={"files": ("f.jpg", make_image(5), "image/jpeg")},
    )
    uid = client.get("/admin/dogs?kind=unknown", headers=auth(admin_token)).json()["items"][0]["id"]

    r = client.post(
        f"/admin/dogs/unknown/{uid}/status", headers=auth(admin_token), json={"status": "reunited"}
    )
    assert r.status_code == 200 and r.json()["profile"]["status"] == "reunited"

    # A known-dog status isn't valid for an unknown dog.
    assert client.post(
        f"/admin/dogs/unknown/{uid}/status", headers=auth(admin_token), json={"status": "home"}
    ).status_code == 400
    # Bad kind / missing dog.
    assert client.post(
        f"/admin/dogs/banana/{uid}/status", headers=auth(admin_token), json={"status": "reunited"}
    ).status_code == 400
    assert client.post(
        "/admin/dogs/unknown/999999/status", headers=auth(admin_token), json={"status": "reunited"}
    ).status_code == 404
    # Non-admin denied.
    assert client.post(
        f"/admin/dogs/unknown/{uid}/status", headers=auth(owner_token), json={"status": "reunited"}
    ).status_code == 403


def test_admin_delete_dog(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2}, identical=True)
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K", mark_lost=True, mark_lost_pct=100)
    did = client.get("/admin/dogs?kind=known", headers=auth(admin_token)).json()["items"][0]["id"]

    r = client.delete(f"/admin/dogs/known/{did}", headers=auth(admin_token))
    assert r.status_code == 200 and r.json()["deleted"]["dogs"] == 1
    assert client.get(f"/admin/dog/known/{did}", headers=auth(admin_token)).status_code == 404
    assert client.delete("/admin/dogs/known/99999", headers=auth(admin_token)).status_code == 404
    assert client.delete("/admin/dogs/banana/1", headers=auth(admin_token)).status_code == 400


def test_admin_owners_search_and_recent_first(client, admin_token):
    client.post("/auth/register", json={
        "name": "Aaron First", "email": "aaron@example.com", "password": "password123", "zip": "20001"})
    client.post("/auth/register", json={
        "name": "Zelda Last", "email": "zelda@example.com", "password": "password123", "zip": "20001"})

    names = [o["name"] for o in client.get("/admin/owners", headers=auth(admin_token)).json()["items"]]
    assert names.index("Zelda Last") < names.index("Aaron First")  # newest first

    hit = client.get("/admin/owners?q=zelda", headers=auth(admin_token)).json()
    assert hit["total"] == 1 and hit["items"][0]["name"] == "Zelda Last"
    assert client.get("/admin/owners?q=nobodyxyz", headers=auth(admin_token)).json()["total"] == 0


def test_admin_owner_detail(client, admin_token, owner_token):
    # Owner registers a dog with a photo and opens a lost case.
    dog = client.post("/dogs", headers=auth(owner_token), json={"name": "Rex"}).json()
    client.post(
        f"/dogs/{dog['id']}/photos", headers=auth(owner_token),
        files={"files": ("d.jpg", make_image(7), "image/jpeg")},
    )
    client.post(
        "/cases/lost", headers=auth(owner_token),
        json={"known_dog_id": dog["id"], "event_zip": "77002", "event_date": "2026-06-01"},
    )
    uid = client.get("/auth/me", headers=auth(owner_token)).json()["id"]

    body = client.get(f"/admin/owners/{uid}", headers=auth(admin_token)).json()
    assert body["owner"]["id"] == uid and body["owner"]["dog_count"] == 1
    assert len(body["dogs"]) == 1 and body["dogs"][0]["name"] == "Rex"
    assert body["dogs"][0]["picture_count"] == 1
    assert len(body["cases"]) == 1 and body["cases"][0]["type"] == "lost"

    # Missing user, and a non-owner (the admin), both 404 — the view is owner-only.
    assert client.get("/admin/owners/999999", headers=auth(admin_token)).status_code == 404
    admin_id = client.get("/auth/me", headers=auth(admin_token)).json()["id"]
    assert client.get(f"/admin/owners/{admin_id}", headers=auth(admin_token)).status_code == 404
    # Non-admin denied.
    assert client.get(f"/admin/owners/{uid}", headers=auth(owner_token)).status_code == 403


def test_admin_delete_person(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2}, identical=True)
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K", mark_lost=True, mark_lost_pct=100)
    owner = client.get("/admin/owners", headers=auth(admin_token)).json()["items"][0]
    uid, dog_count = owner["id"], owner["dog_count"]

    r = client.delete(f"/admin/owners/{uid}", headers=auth(admin_token))
    assert r.status_code == 200
    assert r.json()["deleted"]["users"] == 1 and r.json()["deleted"]["known_dogs"] == dog_count
    # owner and their dogs are gone
    assert all(o["id"] != uid for o in client.get("/admin/owners", headers=auth(admin_token)).json()["items"])
    assert client.delete(f"/admin/owners/{uid}", headers=auth(admin_token)).status_code == 404
    # refuse to delete an admin
    from app.db import SessionLocal
    from app.models import User
    from app.models.base import UserRole
    from sqlalchemy import select
    db = SessionLocal()
    admin_id = db.execute(select(User.id).where(User.role == UserRole.admin)).scalars().first()
    db.close()
    assert client.delete(f"/admin/owners/{admin_id}", headers=auth(admin_token)).status_code == 400


def test_admin_delete_case(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2}, identical=True)
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K", mark_lost=True, mark_lost_pct=100)
    cid = client.get("/admin/cases?kind=lost", headers=auth(admin_token)).json()["items"][0]["id"]

    r = client.delete(f"/admin/cases/{cid}", headers=auth(admin_token))
    assert r.status_code == 200 and r.json()["deleted"]["cases"] == 1
    remaining = client.get("/admin/cases", headers=auth(admin_token)).json()["items"]
    assert all(c["id"] != cid for c in remaining)
    # the dog itself is untouched
    assert client.get("/admin/dogs?kind=known", headers=auth(admin_token)).json()["total"] == 1
    assert client.delete("/admin/cases/99999", headers=auth(admin_token)).status_code == 404


def test_admin_case_detail(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2}, identical=True)
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K", mark_lost=True, mark_lost_pct=100)
    cid = client.get("/admin/cases?kind=lost", headers=auth(admin_token)).json()["items"][0]["id"]

    d = client.get(f"/admin/cases/{cid}", headers=auth(admin_token)).json()
    assert d["case"]["id"] == cid
    assert d["dog"]["profile"]["kind"] == "known"
    assert isinstance(d["dog"]["photos"], list) and isinstance(d["matches"], list)
    assert client.get("/admin/cases/99999", headers=auth(admin_token)).status_code == 404


def test_admin_run_case_match(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2}, identical=True)
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K", mark_lost=True, mark_lost_pct=100)
    _load(root, "found_dogs.csv", "unknown", "U")

    lost = client.get("/admin/cases?kind=lost", headers=auth(admin_token)).json()["items"]
    cid = lost[0]["id"]
    matches = client.post(f"/admin/cases/{cid}/run-match", headers=auth(admin_token)).json()
    assert len(matches) >= 1 and matches[0]["candidate"]["type"] == "unknown"
    assert client.post("/admin/cases/99999/run-match", headers=auth(admin_token)).status_code == 404


def test_purge_retains_user_shared_with_another_dataset():
    """A user shared across datasets (reused email in test CSVs) still owns a dog in the OTHER
    dataset. Purging one dataset must NOT fail on the FK / delete that user — it should keep the
    user (detached: dataset_id nulled) and leave the other dataset intact."""
    from app.models.base import DatasetType, UserRole
    from app.services.datasets import purge_dataset

    db = SessionLocal()
    try:
        d1 = Dataset(name="A", type=DatasetType.known)
        d2 = Dataset(name="B", type=DatasetType.known)
        db.add_all([d1, d2])
        db.flush()
        # One user registered under d1 but owning a dog in BOTH datasets (the shared-email case).
        shared = User(name="o", email="shared@example.com", zip="20001",
                      password_hash="x", role=UserRole.owner, dataset_id=d1.id)
        db.add(shared)
        db.flush()
        dog_a = KnownDog(owner_id=shared.id, name="A0", description="", dataset_id=d1.id)
        dog_b = KnownDog(owner_id=shared.id, name="B0", description="", dataset_id=d2.id)
        db.add_all([dog_a, dog_b])
        db.commit()
        a_id, b_id, dog_a_id, dog_b_id, uid = d1.id, d2.id, dog_a.id, dog_b.id, shared.id

        counts = purge_dataset(db, d1)
        db.expire_all()  # purge detaches users via a core UPDATE; drop stale identity-map state

        # d1 and its dog are gone; the FK error no longer aborts the purge.
        assert db.get(Dataset, a_id) is None
        assert db.get(KnownDog, dog_a_id) is None
        assert counts["users"] == 0 and counts["users_detached"] == 1
        # The shared user survives (detached), and d2 + its dog are untouched.
        survivor = db.get(User, uid)
        assert survivor is not None and survivor.dataset_id is None
        assert db.get(Dataset, b_id) is not None
        assert db.get(KnownDog, dog_b_id) is not None
    finally:
        db.close()


def test_purge_missing_dataset_404(client, admin_token):
    assert client.delete("/admin/datasets/99999", headers=auth(admin_token)).status_code == 404


def test_purge_deletes_media_files(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2, "dogB": 2})
    prepare(root, root, holdout=1, seed=1)
    dsid = _load(root, "known_dogs.csv", "known", "K")

    db = SessionLocal()
    storage = get_storage()
    paths: set[str] = set()  # thumb_path may == file_path (full-image thumbs) -> dedupe
    for fp, tp in db.execute(select(Picture.file_path, Picture.thumb_path)).all():
        paths.add(storage.abs_path(fp))
        if tp:
            paths.add(storage.abs_path(tp))
    db.close()
    assert paths and all(os.path.exists(p) for p in paths)  # files exist before purge

    res = client.delete(f"/admin/datasets/{dsid}", headers=auth(admin_token)).json()
    assert res["deleted"]["media_files"] == len(paths)
    assert all(not os.path.exists(p) for p in paths)  # and gone after


def test_list_all_dogs_grouped(client, admin_token, owner_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2, "dogB": 2})
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K")
    _load(root, "found_dogs.csv", "unknown", "U")

    # auth required
    assert client.get("/admin/dogs", headers=auth(owner_token)).status_code == 403

    allres = client.get("/admin/dogs", headers=auth(admin_token)).json()
    assert allres["counts"] == {"known": 2, "unknown": 2}
    assert allres["total"] == 4
    assert {i["kind"] for i in allres["items"]} == {"known", "unknown"}
    # profile fields present
    first = allres["items"][0]
    for field in ("kind", "id", "name", "status", "picture_count", "dataset_name", "thumb_url"):
        assert field in first

    known = client.get("/admin/dogs?kind=known", headers=auth(admin_token)).json()
    assert known["total"] == 2
    assert all(i["kind"] == "known" for i in known["items"])
    assert known["items"][0]["dataset_name"] == "K"

    unknown = client.get("/admin/dogs?kind=unknown", headers=auth(admin_token)).json()
    assert unknown["total"] == 2
    assert all(i["kind"] == "unknown" for i in unknown["items"])


def test_test_match_finds_counterpart(client, admin_token, owner_token, tmp_path):
    # Identical images per folder -> a known dog's found counterpart scores ~1.0 (mock embedder).
    root = _build_input(tmp_path, {"dogA": 2, "dogB": 2}, identical=True)
    prepare(root, root, holdout=1, seed=1)
    k_id = _load(root, "known_dogs.csv", "known", "K")
    _load(root, "found_dogs.csv", "unknown", "U")

    db = SessionLocal()
    known_dog_id = db.execute(
        select(KnownDog.id).where(KnownDog.dataset_id == k_id).order_by(KnownDog.id)
    ).scalars().first()
    db.close()

    # auth required
    assert client.get(
        f"/admin/test-match?kind=known&dog_id={known_dog_id}", headers=auth(owner_token)
    ).status_code == 403

    r = client.get(f"/admin/test-match?kind=known&dog_id={known_dog_id}", headers=auth(admin_token))
    assert r.status_code == 200, r.text
    body = r.json()
    assert body["query"]["kind"] == "known"
    assert body["query_embedded"] is True
    assert body["candidate_count"] == 2  # two unknown dogs considered
    assert len(body["results"]) >= 1
    top = body["results"][0]
    assert top["kind"] == "unknown"  # opposite category
    assert top["score"] > 0.99  # its identical-image counterpart


def test_test_match_unembedded_dog(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2})
    prepare(root, root, holdout=1, seed=1)
    k_id = _load(root, "known_dogs.csv", "known", "K", skip_embeddings=True)
    db = SessionLocal()
    dog_id = db.execute(select(KnownDog.id).where(KnownDog.dataset_id == k_id)).scalars().first()
    db.close()
    body = client.get(f"/admin/test-match?kind=known&dog_id={dog_id}", headers=auth(admin_token)).json()
    assert body["query_embedded"] is False
    assert body["results"] == []


def test_test_match_404_and_bad_kind(client, admin_token):
    assert client.get("/admin/test-match?kind=known&dog_id=99999", headers=auth(admin_token)).status_code == 404
    assert client.get("/admin/test-match?kind=banana&dog_id=1", headers=auth(admin_token)).status_code == 400


def test_dog_detail_returns_all_photos(client, admin_token, owner_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 5})  # 5 imgs -> holdout 2 -> 3 registration photos
    prepare(root, root, holdout=None, seed=1)
    k_id = _load(root, "known_dogs.csv", "known", "K")
    db = SessionLocal()
    dog_id = db.execute(select(KnownDog.id).where(KnownDog.dataset_id == k_id)).scalars().first()
    db.close()

    assert client.get(f"/admin/dog/known/{dog_id}", headers=auth(owner_token)).status_code == 403

    r = client.get(f"/admin/dog/known/{dog_id}", headers=auth(admin_token))
    assert r.status_code == 200, r.text
    body = r.json()
    assert body["profile"]["id"] == dog_id
    assert len(body["photos"]) == 3  # all registration photos for this grouped dog
    assert all(p["url"] for p in body["photos"])
    assert client.get(f"/admin/dog/known/99999", headers=auth(admin_token)).status_code == 404
    assert client.get(f"/admin/dog/banana/1", headers=auth(admin_token)).status_code == 400


def test_breed_filter(client, admin_token, tmp_path):
    # Mock breed classifier assigns deterministic labels at load time; identical images share them.
    root = _build_input(tmp_path, {"dogA": 2, "dogB": 2}, identical=True)
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K")

    # /admin/breeds lists available labels for the active (mock) model.
    breeds = client.get("/admin/breeds", headers=auth(admin_token)).json()
    assert breeds["breeds"], breeds
    label = breeds["breeds"][0]["label"]
    assert label.startswith("mock-breed-")

    # A dog's profile carries its predicted breeds.
    page = client.get("/admin/dogs?kind=known", headers=auth(admin_token)).json()
    assert page["items"]
    assert "predicted_breeds" in page["items"][0]

    # Filtering by a real top-1 label returns only dogs that have it in their top breed.
    dog0 = page["items"][0]
    top_label = dog0["predicted_breeds"][0]
    filtered = client.get(
        f"/admin/dogs?kind=known&breed={top_label}&breed_k=1", headers=auth(admin_token)
    ).json()
    assert all(top_label in d["predicted_breeds"] for d in filtered["items"])
    assert filtered["total"] >= 1

    # A nonsense breed returns nothing.
    none = client.get("/admin/dogs?breed=not-a-real-breed&breed_k=10", headers=auth(admin_token)).json()
    assert none["total"] == 0


def test_estimated_breeds_blend_frequency_and_confidence(client):
    # Combined metric = SUM of confidence across a dog's photos, blending frequency with confidence.
    # Here "beagle" is the top-1 pick in 2 of 3 photos (pure frequency would pick it), but "labrador"
    # has far more total confidence (one very strong photo + solid elsewhere), so the blend wins it.
    from app.models import BreedPrediction, User
    from app.models.base import UserRole
    from app.services.datasets import aggregated_breeds

    db = SessionLocal()
    try:
        owner = User(name="O", email="breedvote@example.com", zip="20001", role=UserRole.owner)
        db.add(owner)
        db.flush()
        dog = KnownDog(owner_id=owner.id, name="Rex")
        db.add(dog)
        db.flush()
        model = ("test-breed", "v1")
        per_photo = [
            [("beagle", 0.36), ("labrador", 0.34)],   # beagle top-1
            [("beagle", 0.35), ("labrador", 0.33)],   # beagle top-1  -> beagle wins pure frequency
            [("labrador", 0.95), ("beagle", 0.05)],    # labrador very confident
        ]
        # Totals: labrador 0.34+0.33+0.95 = 1.62  vs  beagle 0.36+0.35+0.05 = 0.76.
        for preds in per_photo:
            pic = Picture(
                subject_type=SubjectType.known, subject_id=dog.id,
                file_path="x.jpg", mime_type="image/jpeg",
            )
            db.add(pic)
            db.flush()
            for rank, (label, score) in enumerate(preds):
                db.add(BreedPrediction(
                    picture_id=pic.id, model_name=model[0], model_version=model[1],
                    rank=rank, label=label, score=score,
                ))
        db.commit()

        result = aggregated_breeds(db, SubjectType.known, dog.id, model, limit=3)
        assert result[0] == "labrador"  # blend of frequency + confidence beats top-1 frequency alone
        assert set(result[:2]) == {"labrador", "beagle"}
    finally:
        db.close()


def test_list_all_dogs_ordering_and_pagination(client, admin_token, tmp_path):
    root = _build_input(tmp_path, {"dogA": 2, "dogB": 2, "dogC": 2})
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K")  # 3 known created first (oldest)
    _load(root, "found_dogs.csv", "unknown", "U")  # 3 unknown created later (newest)

    # Default is newest-first: [U, U, U, K, K, K].
    newest = client.get("/admin/dogs?kind=all", headers=auth(admin_token)).json()
    assert [i["kind"] for i in newest["items"]] == ["unknown"] * 3 + ["known"] * 3
    assert newest["total"] == 6

    # Pagination spans the known/unknown boundary: offset 2 => 1 unknown + 3 known.
    page = client.get("/admin/dogs?kind=all&limit=4&offset=2", headers=auth(admin_token)).json()
    assert [i["kind"] for i in page["items"]] == ["unknown", "known", "known", "known"]

    # sort=oldest reverses the order.
    oldest = client.get("/admin/dogs?kind=all&sort=oldest", headers=auth(admin_token)).json()
    assert oldest["items"][0]["kind"] == "known"
    assert [i["id"] for i in oldest["items"]] == [i["id"] for i in reversed(newest["items"])]


def test_list_all_dogs_added_date_filter(client, admin_token, tmp_path):
    from datetime import date, timedelta

    root = _build_input(tmp_path, {"dogA": 2, "dogB": 2})
    prepare(root, root, holdout=1, seed=1)
    _load(root, "known_dogs.csv", "known", "K")  # created today

    today = date.today().isoformat()
    tomorrow = (date.today() + timedelta(days=1)).isoformat()
    yesterday = (date.today() - timedelta(days=1)).isoformat()

    assert client.get(f"/admin/dogs?added_to={today}", headers=auth(admin_token)).json()["total"] >= 2
    assert client.get(f"/admin/dogs?added_from={tomorrow}", headers=auth(admin_token)).json()["total"] == 0
    assert client.get(f"/admin/dogs?added_to={yesterday}", headers=auth(admin_token)).json()["total"] == 0


def test_list_all_dogs_zip_prefix_filter(client, admin_token):
    # Two found dogs in different metros; the ZIP-prefix filter keeps only the Houston one.
    for zip_code in ("77002", "20001"):
        client.post(
            "/cases/found",
            data={"event_zip": zip_code, "event_date": "2026-06-03", "finder_email": "f@example.com"},
            files={"files": ("f.jpg", make_image(101), "image/jpeg")},
        )
    page = client.get("/admin/dogs?kind=unknown&zip=770", headers=auth(admin_token)).json()
    assert page["total"] == 1
    assert page["counts"]["unknown"] == 1
    assert all(d["zip"].startswith("770") for d in page["items"])