fangke/server/tracker/tests/test_ingest.py

from datetime import timedelta
from django.utils import timezone
from tracker.models import Site, Visitor, VisitSession, Event
from tracker.services.geo import resolver as _geo
from tracker.services.ingest import ingest_batch

NOW = timezone.now()

def _batch(events, vid="v1", sid="s1"):
    return {"visitor_id": vid, "session_id": sid, "context": {
        "referrer_domain": "google.com", "utm": {"utm_source": "google"},
        "path": "/landing", "device_type": "PC", "os": "Windows",
        "browser": "Chrome", "language": "en-US", "screen": "1920x1080"},
        "events": events}

def _ev(i, name="page_view", **kw):
    d = dict(event_id=f"e{i}", event_name=name, occurred_at=NOW,
             path="/p", title="T", properties={})
    d.update(kw)
    return d

# 1. 首批建档
def test_first_batch_creates_visitor_session_events(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    r = ingest_batch(site, _batch([_ev(1), _ev(2)]), ip="114.114.114.114")
    assert (r["accepted"], r["duplicated"], r["rejected"]) == (2, 0, [])
    v = Visitor.objects.get(site=site, visitor_id="v1")
    assert v.first_referrer_domain == "google.com" and v.visit_count == 1
    assert v.active_days == 1 and v.last_ip == "114.114.114.114"
    if _geo._searcher:    # 地理断言仅在有真实 xdb 时执行，CI/fresh clone 不因缺库变红
        assert v.country == "中国" and v.city == "南京市"    # 114 DNS（江苏南京）
    assert Event.objects.filter(site=site).count() == 2

# 2. 重放幂等 + 不抬高会话聚合
def test_duplicate_events_never_inflate_aggregates(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    ingest_batch(site, _batch([_ev(1), _ev(2, "page_stay", properties={"active_ms": 3000, "path": "/p"})]))
    r = ingest_batch(site, _batch([_ev(1)]))     # 重放
    assert r["duplicated"] == 1 and Event.objects.count() == 2
    sess = VisitSession.objects.get()
    assert sess.event_count == 2 and sess.stay_ms == 3000   # 重放没有重复加分

# 3. 新会话递增 visit_count（F 表达式）
def test_new_session_increments_visit_count(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    ingest_batch(site, _batch([_ev(1)], sid="s1"))
    ingest_batch(site, _batch([_ev(2)], sid="s2"))
    v = Visitor.objects.get(visitor_id="v1")
    assert v.visit_count == 2 and VisitSession.objects.count() == 2

# 4. 服务端 30 分钟兜底切分（同 sid 间隔>30min 切出 s1#2，且同批后续事件复用同段）
def test_server_backstop_splits_idle_gap(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    ingest_batch(site, _batch([_ev(1, occurred_at=NOW - timedelta(minutes=90))]))
    r = ingest_batch(site, _batch([_ev(2), _ev(3)]))       # 两条同批
    assert VisitSession.objects.count() == 2               # s1 与 s1#2，不会切成 3 段
    assert r["accepted"] == 2

# 5. 拒绝事件带原因上报
def test_rejected_events_reported_with_reason(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    r = ingest_batch(site, _batch([_ev(1, event_name="bad"), _ev(2)]))
    assert r["accepted"] == 1 and r["rejected"][0]["reason"] == "event_name_not_allowed"

# 6. page_stay 的 active_ms 累计进 stay_ms
def test_stay_ms_accumulates(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    ingest_batch(site, _batch([
        _ev(1), _ev(2, "page_stay", properties={"active_ms": 5000, "path": "/p"})]))
    ingest_batch(site, _batch([
        _ev(3, occurred_at=NOW + timedelta(minutes=1)),
        _ev(4, "page_stay", properties={"active_ms": 7000, "path": "/p"},
            occurred_at=NOW + timedelta(minutes=1))]))
    assert VisitSession.objects.get().stay_ms == 12000

# 7. 爬虫 UA 标记（访客+事件）
def test_bot_ua_marked(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    ingest_batch(site, _batch([_ev(1)]), ip="66.249.66.1", ua="Mozilla/5.0 (compatible; Googlebot/2.1)")
    assert Visitor.objects.get().is_bot is True
    assert Event.objects.get().is_bot is True

# 8. 超批拒绝
def test_batch_size_invalid(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    r = ingest_batch(site, _batch([_ev(i) for i in range(60)]))
    assert r["rejected"] == [{"reason": "batch_size_invalid"}]

# 9. 同批重复 event_id：去重后只入一份，聚合不灌倍
def test_intra_batch_duplicate_event_id_deduped(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    evs = [_ev(1, "page_stay", properties={"active_ms": 1000, "path": "/p"})
           for _ in range(3)]                            # 3 份同 id
    r = ingest_batch(site, _batch(evs))
    assert (r["accepted"], r["duplicated"]) == (1, 2)
    sess = VisitSession.objects.get()
    assert sess.event_count == 1 and sess.stay_ms == 1000

# 10. 切分后同 sid 小间隔连发：空闲比较取会话族最大 ended_at，不再每批新开段
def test_split_then_small_gap_reuses_segment(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    ingest_batch(site, _batch([_ev(1, occurred_at=NOW - timedelta(minutes=90))]))
    ingest_batch(site, _batch([_ev(2)]))                 # 90min 间隔 → 切出 s1#2
    ingest_batch(site, _batch([_ev(3, occurred_at=NOW + timedelta(minutes=1))]))
    ingest_batch(site, _batch([_ev(4, occurred_at=NOW + timedelta(minutes=2))]))
    assert VisitSession.objects.count() == 2             # 保持 s1 与 s1#2，不切 s1#3/s1#4
    assert Visitor.objects.get(visitor_id="v1").visit_count == 2

# 11. invariant 4：乱序批次（后入更早发生）ended_at 不回退
def test_ended_at_never_regresses_on_out_of_order_batch(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    ingest_batch(site, _batch([_ev(1, occurred_at=NOW + timedelta(minutes=1))]))
    ingest_batch(site, _batch([_ev(2, occurred_at=NOW)]))
    sess = VisitSession.objects.get()
    assert abs((sess.ended_at - (NOW + timedelta(minutes=1))).total_seconds()) < 1

# 12. 首触持久：第二批换 referrer 不改首触字段
def test_first_touch_immutable_across_batches(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    ingest_batch(site, _batch([_ev(1)]))
    b = _batch([_ev(2)])
    b["context"]["referrer_domain"] = "bing.com"
    ingest_batch(site, b)
    v = Visitor.objects.get(visitor_id="v1")
    assert v.first_referrer_domain == "google.com"
    assert v.first_utm == {"utm_source": "google"}

# 13. payload 缺 visitor_id/session_id 提前拒绝
def test_missing_payload_fields_rejected(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    p = _batch([_ev(1)])
    del p["visitor_id"]
    assert ingest_batch(site, p)["rejected"] == [{"reason": "invalid_payload"}]
    p = _batch([_ev(1)])
    del p["session_id"]
    assert ingest_batch(site, p)["rejected"] == [{"reason": "invalid_payload"}]

# 14. SDK context 的 timezone 落库 Visitor.timezone_name（Task16 联调发现：键名映射缺口）
def test_context_timezone_persisted_to_visitor(db):
    site = Site.objects.create(name="s", domains=["a.com"])
    b = _batch([_ev(1)])
    b["context"]["timezone"] = "Asia/Shanghai"
    ingest_batch(site, b)
    assert Visitor.objects.get(visitor_id="v1").timezone_name == "Asia/Shanghai"

# 15. 归因分类 best-effort：引擎抛异常不影响采集入库（first_touch/session_entry 双路）
def test_attribution_failure_never_breaks_ingest(db, monkeypatch):
    from tracker.services import ingest as ingest_mod
    site = Site.objects.create(name="s", domains=["a.com"])
    monkeypatch.setattr(ingest_mod, "classify", lambda **kw: (_ for _ in ()).throw(RuntimeError("boom")))
    r = ingest_batch(site, _batch([_ev(1)]))
    assert r["accepted"] == 1 and Event.objects.count() == 1
    assert Visitor.objects.count() == 1 and VisitSession.objects.count() == 1

# 16. 首触分类用本批第一条 page_view 的完整 referrer（搜索词可恢复），非 ctx 裸域名
def test_first_touch_classification_uses_full_pageview_referrer(db):
    from tracker.models import SourceClassification
    site = Site.objects.create(name="s", domains=["a.com"])
    b = _batch([_ev(1, properties={"referrer": "https://m.baidu.com/s?wd=挖掘机"})])
    b["context"]["utm"] = {}     # 默认 fixture 带 utm_source=google，与百度矛盾 → conflict
    ingest_batch(site, b)
    ft = SourceClassification.objects.get(scope="first_touch")
    assert ft.channel == "natural_search" and ft.platform == "百度"
    assert ft.search_term == "挖掘机" and ft.resolution == "resolved"