From d39ee838ff76bf359aeebc705ba10ff39673f9cf Mon Sep 17 00:00:00 2001
From: Josue Zamudio <josuezamudio1@gmail.com>
Date: Sun, 23 Aug 2026 21:29:01 +0000
Subject: [PATCH] api: dedup ghost history rows (int vs string ep_id);
 normalize ep_id on write

---
 server/app.py     | 19 +++++++++++++++++--
 tests/test_api.py | 17 +++++++++++++++--
 2 files changed, 32 insertions(+), 4 deletions(-)

diff --git a/server/app.py b/server/app.py
index 238d038..05af4fe 100644
--- a/server/app.py
+++ b/server/app.py
@@ -125,9 +125,17 @@ HIST_DIR = Path(os.environ.get("ANIWEB_DATA", Path(__file__).parent / "data")) /
 
 def _load_hist(user: str):
     try:
-        return json.loads((HIST_DIR / f"{user}.json").read_text())
+        entries = json.loads((HIST_DIR / f"{user}.json").read_text())
     except (FileNotFoundError, json.JSONDecodeError):
         return []
+    # ponytail: dedup legacy ghost rows — the same episode was stored with ep_id
+    # as int AND string; keep the newest per (slug_id, str(ep_id))
+    best = {}
+    for e in entries:
+        k = (str(e.get("slug_id")), str(e.get("ep_id")))
+        if e.get("t", 0) >= best.get(k, {"t": 0})["t"]:
+            best[k] = e
+    return sorted(best.values(), key=lambda e: e.get("t", 0), reverse=True)
 
 
 def _save_hist(user: str, entries):
@@ -162,11 +170,18 @@ def api_history_delete(request: Request, slug_id: str, ep_id: str):
 def api_history_post(entry: dict = Body(...), request: Request = None):
     if "ep_id" not in entry:
         raise HTTPException(422, "ep_id required")
+    # normalize: ghost rows were born when ep_id arrived as str against an
+    # int row (or vice versa) — int() makes the stored type stable
+    try:
+        entry["ep_id"] = int(entry["ep_id"])
+    except (TypeError, ValueError):
+        raise HTTPException(422, "ep_id must be an integer")
     entry["t"] = time.time()
     u = request.state.user
     entries = _load_hist(u)
     for i, e in enumerate(entries):
-        if e.get("slug_id") == entry.get("slug_id") and e.get("ep_id") == entry.get("ep_id"):
+        # loose str() match: merge onto a legacy ghost row of the other type
+        if e.get("slug_id") == entry.get("slug_id") and str(e.get("ep_id")) == str(entry["ep_id"]):
             entries[i] = {**e, **entry}
             break
     else:
diff --git a/tests/test_api.py b/tests/test_api.py
index c9d9396..aa50409 100644
--- a/tests/test_api.py
+++ b/tests/test_api.py
@@ -111,8 +111,9 @@ def test_history_per_user(base=BASE):
     assert req("POST", "/api/history", entry, jar=a, base=base) == 200
     got_a = _json("GET", "/api/history", jar=a, base=base)
     got_b = _json("GET", "/api/history", jar=b, base=base)
-    assert any(e["ep_id"] == "111" for e in got_a)
-    assert all(e.get("ep_id") != "111" for e in got_b), "leak: admin entry in alice's history"
+    # ep_id is normalized to int on write (ghost-row fix)
+    assert any(e["ep_id"] == 111 for e in got_a)
+    assert all(e.get("ep_id") != 111 for e in got_b), "leak: admin entry in alice's history"
 
 
 def test_history_all_and_delete(base=BASE):
@@ -138,6 +139,18 @@ def test_history_all_and_delete(base=BASE):
                  jar=jar, base=base)["removed"] == 0
 
 
+def test_history_no_ghost_duplicates(base=BASE):
+    """same episode posted with int and string ep_id -> exactly one row"""
+    jar = login("alice", "alicepw12345", base)
+    for ep in (301, "301"):
+        assert req("POST", "/api/history",
+                   {"ep_id": ep, "slug_id": "ghost-a", "title": "Ghost", "season_id": 1,
+                    "ep": "1", "lang": "eng", "pos": 5}, jar=jar, base=base) == 200
+    rows = [e for e in _json("GET", "/api/history?all=1", jar=jar, base=base)
+            if e["slug_id"] == "ghost-a"]
+    assert len(rows) == 1 and rows[0]["ep_id"] == 301, rows
+
+
 def test_recent_searches(base=BASE):
     a = login("alice", "alicepw12345", base)
     assert req("POST", "/api/searches", {"q": "naruto"}, jar=a, base=base) == 200