enduro-trails/tests/api/test_gps_tracks_dedup.py

"""Unit тесты для дедупликации GPS-треков (ET-008).

U-10: два трека с одинаковым bbox+length+date → один ключ
U-11: разные даты → разные ключи
U-12: bbox-округление до 0.01°
U-13: merge sources при upsert
U-14: merge external_urls
"""
import json
import pytest

from src.api.gps_tracks.dedup import compute_dedup_key
from src.api.gps_tracks.db import open_db, init_db, upsert_track
from src.api.gps_tracks.models import TrackInsert


def _make_track(
    external_id="T1",
    source_id="osm",
    length_m=5000.0,
    created_at="2024-05-12T10:00:00Z",
    min_lon=37.61,
    min_lat=55.75,
    max_lon=37.62,
    max_lat=55.76,
    external_url=None,
    name=None,
    source_priority=50,
) -> TrackInsert:
    """Хелпер для создания TrackInsert с тестовой WKB геометрией."""
    from shapely.geometry import LineString
    from shapely import wkb

    coords = [(min_lon, min_lat), (max_lon, max_lat)]
    geom_wkb = wkb.dumps(LineString(coords))

    return TrackInsert(
        external_id=external_id,
        source_id=source_id,
        external_url=external_url,
        name=name,
        description=None,
        activity_type="other",
        user=None,
        created_at=created_at,
        length_m=length_m,
        points_count=2,
        geom_wkb=geom_wkb,
        min_lon=min_lon,
        min_lat=min_lat,
        max_lon=max_lon,
        max_lat=max_lat,
        tags=[],
        source_priority=source_priority,
    )


@pytest.fixture
def db(tmp_path):
    """Создаёт изолированную БД в tmp_path."""
    db_path = str(tmp_path / "test.sqlite")
    conn = open_db(db_path)
    init_db(conn)
    yield conn
    conn.close()


# ─── U-10: одинаковый bbox+length+date → один ключ ───────────────────────────

def test_u10_same_key_for_same_track():
    """U-10: два трека с одинаковым bbox+length+date дают одинаковый ключ."""
    bounds = (37.61, 55.75, 37.62, 55.76)
    meta = {"length_m": 5000.0, "created_at": "2024-05-12T10:00:00Z"}

    key1 = compute_dedup_key(bounds, meta)
    key2 = compute_dedup_key(bounds, meta)

    assert key1 == key2


# ─── U-11: разные даты → разные ключи ────────────────────────────────────────

def test_u11_different_dates_give_different_keys():
    """U-11: треки с разными датами дают разные ключи."""
    bounds = (37.61, 55.75, 37.62, 55.76)

    key1 = compute_dedup_key(bounds, {"length_m": 5000.0, "created_at": "2024-05-12"})
    key2 = compute_dedup_key(bounds, {"length_m": 5000.0, "created_at": "2024-05-13"})

    assert key1 != key2


# ─── U-12: bbox-округление до 0.01° ─────────────────────────────────────────

def test_u12_bbox_rounding_to_2_decimals():
    """U-12: bbox округляется до 0.01°, незначительные отличия игнорируются."""
    # Оба варианта округляются к (37.61, 55.75, 37.62, 55.76)
    # Используем значения в середине диапазона, гарантированно округляемые одинаково
    bounds1 = (37.6111, 55.7512, 37.6192, 55.7563)
    bounds2 = (37.6144, 55.7533, 37.6188, 55.7571)

    meta = {"length_m": 5000.0, "created_at": "2024-05-12"}

    key1 = compute_dedup_key(bounds1, meta)
    key2 = compute_dedup_key(bounds2, meta)

    # Оба bbox округляются к (37.61, 55.75, 37.62, 55.76) — ключи одинаковы
    assert key1 == key2


def test_u12_significantly_different_bbox_gives_different_key():
    """U-12: существенно разные bbox дают разные ключи."""
    bounds1 = (37.61, 55.75, 37.62, 55.76)
    bounds2 = (38.00, 56.00, 38.10, 56.10)

    meta = {"length_m": 5000.0, "created_at": "2024-05-12"}

    key1 = compute_dedup_key(bounds1, meta)
    key2 = compute_dedup_key(bounds2, meta)

    assert key1 != key2


# ─── U-13: merge sources при upsert ──────────────────────────────────────────

def test_u13_merge_sources_on_upsert(db):
    """U-13: при upsert с тем же dedup_key sources мержатся (union без дублей)."""
    bounds = (37.61, 55.75, 37.62, 55.76)
    meta = {"length_m": 5000.0, "created_at": "2024-05-12T10:00:00Z"}
    dedup_key = compute_dedup_key(bounds, meta)

    # Первая вставка — от osm
    track1 = _make_track(external_id="T1", source_id="osm", source_priority=50)
    result1 = upsert_track(db, track1, dedup_key, source_priority=50)
    assert result1 == "inserted"

    # Вторая вставка — от другого источника с тем же dedup_key
    track2 = _make_track(external_id="T2", source_id="enduro_russia", source_priority=10)
    result2 = upsert_track(db, track2, dedup_key, source_priority=10)
    assert result2 == "updated"

    # Проверяем merged sources
    cur = db.cursor()
    cur.execute("SELECT sources_json FROM tracks WHERE dedup_key = ?", (dedup_key,))
    row = cur.fetchone()
    sources = json.loads(row["sources_json"])

    assert "osm" in sources
    assert "enduro_russia" in sources
    assert len(sources) == 2  # без дублей


def test_u13_no_duplicate_sources_on_repeated_upsert(db):
    """U-13: повторный upsert от того же источника не создаёт дублей в sources."""
    bounds = (37.61, 55.75, 37.62, 55.76)
    meta = {"length_m": 5000.0, "created_at": "2024-05-12T10:00:00Z"}
    dedup_key = compute_dedup_key(bounds, meta)

    track = _make_track(external_id="T1", source_id="osm")
    upsert_track(db, track, dedup_key, source_priority=50)
    upsert_track(db, track, dedup_key, source_priority=50)
    upsert_track(db, track, dedup_key, source_priority=50)

    cur = db.cursor()
    cur.execute("SELECT sources_json FROM tracks WHERE dedup_key = ?", (dedup_key,))
    row = cur.fetchone()
    sources = json.loads(row["sources_json"])

    assert sources.count("osm") == 1


# ─── U-14: merge external_urls ───────────────────────────────────────────────

def test_u14_merge_external_urls_on_upsert(db):
    """U-14: external_urls мержатся без дублей при upsert."""
    bounds = (37.61, 55.75, 37.62, 55.76)
    meta = {"length_m": 5000.0, "created_at": "2024-05-12T10:00:00Z"}
    dedup_key = compute_dedup_key(bounds, meta)

    url1 = "https://www.openstreetmap.org/user/alice/traces/12345"
    url2 = "https://enduro-russia.ru/track/99"

    track1 = _make_track(external_id="T1", source_id="osm", external_url=url1)
    upsert_track(db, track1, dedup_key, source_priority=50)

    track2 = _make_track(external_id="T2", source_id="enduro_russia", external_url=url2)
    upsert_track(db, track2, dedup_key, source_priority=10)

    cur = db.cursor()
    cur.execute("SELECT external_urls_json FROM tracks WHERE dedup_key = ?", (dedup_key,))
    row = cur.fetchone()
    urls = json.loads(row["external_urls_json"])

    assert url1 in urls
    assert url2 in urls
    assert len(urls) == 2


def test_u14_no_duplicate_urls_on_repeated_upsert(db):
    """U-14: повторный upsert с тем же URL не дублирует его."""
    bounds = (37.61, 55.75, 37.62, 55.76)
    meta = {"length_m": 5000.0, "created_at": "2024-05-12T10:00:00Z"}
    dedup_key = compute_dedup_key(bounds, meta)

    url = "https://www.openstreetmap.org/user/alice/traces/12345"

    track = _make_track(external_id="T1", source_id="osm", external_url=url)
    upsert_track(db, track, dedup_key, source_priority=50)
    upsert_track(db, track, dedup_key, source_priority=50)

    cur = db.cursor()
    cur.execute("SELECT external_urls_json FROM tracks WHERE dedup_key = ?", (dedup_key,))
    row = cur.fetchone()
    urls = json.loads(row["external_urls_json"])

    assert urls.count(url) == 1