From 65761885ba35644dfa995f4abdf7feac2e382d60 Mon Sep 17 00:00:00 2001
From: Roman <roman@lanro.eu>
Date: Wed, 23 Sep 2026 12:42:24 +0000
Subject: [PATCH] =?UTF-8?q?Corrections=20:=20test=20Torznab=20p=C3=A9rim?=
 =?UTF-8?q?=C3=A9=20et=20retries=20sur=20erreurs=20d=C3=A9finitives?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- test_torznab_search_requires_query attendait un 400 mais le commit
  853b4e0 a volontairement fait servir les nouveautés quand q est
  absent (RSS sync Sonarr) — le test valide désormais ce comportement.
- fetch() ne retente plus les 4xx définitifs (sauf 408/429) : un
  lecteur VoirAnime en échec 422 coûtait 2 requêtes et 3 s de pauses
  avant de basculer sur le lecteur suivant.
- 4 nouveaux tests unitaires du client HTTP (fail-fast + retries).
---
 app/scrapers/http.py  |  8 ++++++++
 tests/test_core.py    | 43 +++++++++++++++++++++++++++++++++++++++++++
 tests/test_torznab.py | 21 +++++++++++++++++++--
 3 files changed, 70 insertions(+), 2 deletions(-)

diff --git a/app/scrapers/http.py b/app/scrapers/http.py
index 8655a44..269e41e 100644
--- a/app/scrapers/http.py
+++ b/app/scrapers/http.py
@@ -53,6 +53,14 @@ async def fetch(
             response = await get_client().get(url, headers=request_headers)
             response.raise_for_status()
             return response.text
+        except httpx.HTTPStatusError as exc:
+            status = exc.response.status_code
+            if 400 <= status < 500 and status not in (408, 429):
+                raise ScrapeError(f"Échec de récupération de {url} : HTTP {status} (définitif)") from exc
+            last_error = exc
+            logger.warning("fetch %s — HTTP %d, tentative %d/%d", url, status, attempt + 1, retries + 1)
+            if attempt < retries:
+                await asyncio.sleep(1.0 * (attempt + 1))
         except (httpx.HTTPError, httpx.InvalidURL) as exc:
             last_error = exc
             logger.warning("fetch %s — tentative %d/%d : %s", url, attempt + 1, retries + 1, exc)
diff --git a/tests/test_core.py b/tests/test_core.py
index 0fa804d..383ad04 100644
--- a/tests/test_core.py
+++ b/tests/test_core.py
@@ -1,9 +1,12 @@
 """Tests unitaires : format interne, sanitisation, auth, registres scrapers."""
 
+import httpx
 import pytest
 
 from app import auth
+from app.scrapers import http as http_module
 from app.scrapers.base import (
+    ScrapeError,
     decode_internal_url,
     encode_internal_url,
     get_source,
@@ -133,3 +136,43 @@ def test_hoster_resolution():
     assert resolve_hoster("https://vidmoly.to/embed-x.html").name == "vidmoly"
     assert resolve_hoster("https://sendvid.com/embed/x").name == "sendvid"
     assert resolve_hoster("https://inconnu.example.com/v/1") is None
+
+
+# ------------------------------------------------------------ client HTTP
+
+
+class _StatusClient:
+    """Fake client HTTP : sert une série de codes statut puis un corps 200."""
+
+    def __init__(self, statuses: list[int]):
+        self._statuses = statuses
+        self.calls = 0
+
+    async def get(self, url, headers=None):
+        self.calls += 1
+        request = httpx.Request("GET", url)
+        status = self._statuses.pop(0) if self._statuses else 200
+        response = httpx.Response(status, request=request, text="ok")
+        response.raise_for_status()
+        return response
+
+
+async def test_fetch_fails_fast_on_definitive_4xx(monkeypatch):
+    client = _StatusClient([422])
+    monkeypatch.setattr("app.scrapers.http.get_client", lambda: client)
+    with pytest.raises(ScrapeError, match="422"):
+        await http_module.fetch("https://site.example/lecteur/prepare/1?content=2&episode=3")
+    assert client.calls == 1  # aucun retry sur erreur définitive
+
+
+@pytest.mark.parametrize("status", [429, 500, 503])
+async def test_fetch_retries_transitory_errors(monkeypatch, status):
+    client = _StatusClient([status, 200])
+    monkeypatch.setattr("app.scrapers.http.get_client", lambda: client)
+
+    async def no_sleep(_delay):
+        return None
+
+    monkeypatch.setattr("app.scrapers.http.asyncio.sleep", no_sleep)
+    assert await http_module.fetch("https://site.example/page", retries=2) == "ok"
+    assert client.calls == 2
diff --git a/tests/test_torznab.py b/tests/test_torznab.py
index 37d24d9..a7808e1 100644
--- a/tests/test_torznab.py
+++ b/tests/test_torznab.py
@@ -128,9 +128,26 @@ async def test_torznab_tvsearch_all_episodes(client, apikey, fake_source):
     assert "Frieren S02E01 VOSTFR WEB-DL" in titles
 
 
-async def test_torznab_search_requires_query(client, apikey):
+async def test_torznab_search_without_query_serves_latest(client, apikey, fake_source):
+    # Sans q : flux RSS des nouveautés (RSS sync / test Sonarr), pas une erreur 400.
+    async def fake_latest():
+        return [
+            SearchResult(
+                source="fake",
+                source_id="frieren-1",
+                title="Frieren",
+                url="https://fake.example/frieren",
+            )
+        ]
+
+    fake_source.latest = fake_latest
     r = await client.get("/torznab/api", params={"t": "search", "apikey": apikey})
-    assert r.status_code == 400
+    assert r.status_code == 200
+    root = ET.fromstring(r.text)
+    items = root.findall(".//item")
+    assert len(items) == 1
+    # seul le dernier épisode de la nouveauté est publié (S02E01 > S01E02 > S01E01)
+    assert items[0].find("title").text == "Frieren S02E01 VOSTFR WEB-DL"
 
 
 # ---------------------------------------------------------------- grab