analytics: filter known-bot User-Agents at /api/events (honest visitor counts)

Many modern crawlers (AI scrapers, headless Chrome, link-preview fetchers) run JS and
fire the visit/summary_viewed beacon, inflating "visitors" even though there's no
human discovery channel. Apply queries.is_bot_ua() at /api/events — the same filter
the load-error beacon uses — so honest bot UAs (GPTBot, AhrefsBot, headless Chrome,
python/curl, …) are dropped before recording. Response is identical so a bot can't
detect it. Counts read lower but truer going forward (past rows unchanged). Won't catch
UA-spoofing bots; that needs a heavier heuristic. Tests: bot UAs dropped, real browser
counted; existing event tests send a real UA (default client UA contains "python").

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

tests/test_events.py

@@ -1,6 +1,12 @@
 import pytest
 from fastapi.testclient import TestClient
 
+# The events beacon now drops known-bot User-Agents (queries.is_bot_ua), and the test
+# client's default UA contains "python" → would be filtered. Send a real browser UA so
+# these record like a genuine visitor; the bot case is covered explicitly below.
+_BROWSER = {"user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 "
+                          "(KHTML, like Gecko) Chrome/120.0 Safari/537.36"}
+
 
 @pytest.fixture
 def app_db(tmp_path, monkeypatch):
@@ -28,7 +34,7 @@ def _count(db, **where):
 
 def test_event_recorded_and_deduped(app_db):
     app, db = app_db
-    tc = TestClient(app)
+    tc = TestClient(app, headers=_BROWSER)
     for _ in range(3):  # same (kind, article, visitor, day) → one row
         assert tc.post("/api/events", json={"kind": "open", "article_id": 5, "visitor": "tok"}).json() == {"ok": True}
     assert _count(db, kind="open", article_id=5) == 1
@@ -39,7 +45,7 @@ def test_event_recorded_and_deduped(app_db):
 
 def test_visitor_token_is_hashed_not_stored_raw(app_db):
     app, db = app_db
-    TestClient(app).post("/api/events", json={"kind": "visit", "visitor": "secret-token"})
+    TestClient(app, headers=_BROWSER).post("/api/events", json={"kind": "visit", "visitor": "secret-token"})
     from goodnews.db import connect
     c = connect(str(db))
     vh = c.execute("SELECT visitor_hash FROM events").fetchone()[0]
@@ -49,13 +55,28 @@ def test_visitor_token_is_hashed_not_stored_raw(app_db):
 
 def test_unknown_kind_is_ignored(app_db):
     app, db = app_db
-    assert TestClient(app).post("/api/events", json={"kind": "evil", "visitor": "x"}).json() == {"ok": True}
+    assert TestClient(app, headers=_BROWSER).post("/api/events", json={"kind": "evil", "visitor": "x"}).json() == {"ok": True}
     assert _count(db) == 0
 
 
+def test_bot_user_agents_are_not_counted(app_db):
+    """JS-capable crawlers fire this beacon too; honest bot UAs must not inflate counts.
+    Response stays {ok:true} so a bot can't tell it was dropped."""
+    app, db = app_db
+    for bot_ua in ("Mozilla/5.0 (compatible; GPTBot/1.0; +https://openai.com/gptbot)",
+                   "Mozilla/5.0 (compatible; AhrefsBot/7.0; +http://ahrefs.com/robot/)",
+                   "Mozilla/5.0 (X11; Linux x86_64) HeadlessChrome/120.0 Safari/537.36"):
+        tc = TestClient(app, headers={"user-agent": bot_ua})
+        assert tc.post("/api/events", json={"kind": "visit", "visitor": "b"}).json() == {"ok": True}
+    assert _count(db, kind="visit") == 0                      # none recorded
+    # a real browser on the same beacon IS counted
+    TestClient(app, headers=_BROWSER).post("/api/events", json={"kind": "visit", "visitor": "human"})
+    assert _count(db, kind="visit") == 1
+
+
 def test_game_event_kinds_are_allowed(app_db):
     app, db = app_db
-    tc = TestClient(app)
+    tc = TestClient(app, headers=_BROWSER)
     # the per-game funnel kinds (incl. the share-loop arrival) pass the allowlist
     for kind in ("word_started", "word_completed", "word_shared", "word_arrival", "match_arrival"):
         assert tc.post("/api/events", json={"kind": kind, "article_id": 0, "visitor": "t"}).json() == {"ok": True}
@@ -67,7 +88,7 @@ def test_game_event_kinds_are_allowed(app_db):
 
 def test_admin_stats_games_funnel_aggregates(app_db):
     app, db = app_db
-    tc = TestClient(app)
+    tc = TestClient(app, headers=_BROWSER)
     # two visitors arrive at Daily Word via a shared link; one engages + shares; a Match completes
     for v in ("a", "b"):
         tc.post("/api/events", json={"kind": "word_arrival", "article_id": 0, "visitor": v})