Merge remote-tracking branch 'upstream/dev' into fix/onboarding

surfsense_backend/tests/unit/capabilities/instagram/test_schemas.py

@@ -66,6 +66,4 @@ def test_details_wraps_profile_items():
 
 def test_details_rejects_both_sources():
     with pytest.raises(ValidationError):
-        DetailsInput(
-            urls=["https://www.instagram.com/natgeo/"], search_queries=["x"]
-        )
+        DetailsInput(urls=["https://www.instagram.com/natgeo/"], search_queries=["x"])

surfsense_backend/tests/unit/capabilities/tiktok/scrape/test_executor.py

@@ -54,7 +54,6 @@ async def test_forwards_typed_sources_and_limit():
         ScrapeInput(
             profiles=["nasa"],
             hashtags=["food"],
-            search_queries=["cats"],
             results_per_page=7,
             max_items=25,
         )
@@ -63,7 +62,6 @@ async def test_forwards_typed_sources_and_limit():
     (actor_input, limit) = scraper.calls[0]
     assert actor_input.profiles == ["nasa"]
     assert actor_input.hashtags == ["food"]
-    assert actor_input.searchQueries == ["cats"]
     assert actor_input.resultsPerPage == 7
     # The outer collection limit is the caller's total-item cap.
     assert limit == 25

surfsense_backend/tests/unit/platforms/instagram/test_budget.py

@@ -42,8 +42,7 @@ def _profile_payload(n: int) -> dict:
                 "edge_owner_to_timeline_media": {
                     "count": n,
                     "edges": [
-                        {"node": {"id": str(i), "shortcode": f"S{i}"}}
-                        for i in range(n)
+                        {"node": {"id": str(i), "shortcode": f"S{i}"}} for i in range(n)
                     ],
                 },
             }

surfsense_backend/tests/unit/platforms/instagram/test_discovery.py

@@ -33,9 +33,7 @@ async def test_google_discovery_keeps_only_profiles(monkeypatch):
             "https://example.com/not-instagram",
         ),
     )
-    targets = await scraper._discover(
-        "nat geo photos", search_type="profile", limit=10
-    )
+    targets = await scraper._discover("nat geo photos", search_type="profile", limit=10)
     assert [(t.kind, t.value) for t in targets] == [("profile", "natgeo")]
 
 
@@ -48,9 +46,7 @@ async def test_google_discovery_dedupes(monkeypatch):
             "https://www.instagram.com/natgeo/",
         ),
     )
-    targets = await scraper._discover(
-        "nat geo photos", search_type="profile", limit=10
-    )
+    targets = await scraper._discover("nat geo photos", search_type="profile", limit=10)
     assert len(targets) == 1
 
 

surfsense_backend/tests/unit/platforms/instagram/test_fetch_resilience.py

@@ -102,7 +102,9 @@ async def test_warms_then_returns_json():
     holder = _FakeHolder([_FakeSession(200, csrftoken=True)])
     token = _current_session.set(holder)
     try:
-        result = await fetch_json("api/v1/users/web_profile_info/", {"username": "natgeo"})
+        result = await fetch_json(
+            "api/v1/users/web_profile_info/", {"username": "natgeo"}
+        )
     finally:
         _current_session.reset(token)
     assert result == _PAYLOAD

surfsense_backend/tests/unit/platforms/instagram/test_parsers.py

@@ -175,8 +175,14 @@ def test_parse_post_prefers_relay_json():
                 "image_versions2": {"candidates": [{"url": "https://cdn/c2.jpg"}]},
             },
         ],
-        "usertags": {"in": [{"position": [0.5, 0.5], "user": {"username": "tagged1", "id": "77"}}]},
-        "coauthor_producers": [{"username": "coauthor1", "id": "88", "is_verified": True}],
+        "usertags": {
+            "in": [
+                {"position": [0.5, 0.5], "user": {"username": "tagged1", "id": "77"}}
+            ]
+        },
+        "coauthor_producers": [
+            {"username": "coauthor1", "id": "88", "is_verified": True}
+        ],
         "location": {"id": "123", "name": "Bali"},
     }
     html = (

surfsense_backend/tests/unit/platforms/tiktok/test_fetch_resilience.py

@@ -84,7 +84,9 @@ async def test_warms_then_returns_html():
 
 
 async def test_rotates_when_warm_fails_then_succeeds():
-    holder = _FakeHolder([_FakeSession(200, warms=False), _FakeSession(200, warms=True)])
+    holder = _FakeHolder(
+        [_FakeSession(200, warms=False), _FakeSession(200, warms=True)]
+    )
     token = _current_session.set(holder)
     try:
         result = await client.fetch_html("https://www.tiktok.com/@scout2015")
@@ -119,9 +121,7 @@ async def test_rotates_and_rewarms_on_403():
 
 async def test_persistent_403_raises_blocked(monkeypatch):
     _no_sleep(monkeypatch)
-    holder = _FakeHolder(
-        [_FakeSession(403) for _ in range(client._MAX_ROTATIONS + 1)]
-    )
+    holder = _FakeHolder([_FakeSession(403) for _ in range(client._MAX_ROTATIONS + 1)])
     token = _current_session.set(holder)
     try:
         raised = False

surfsense_backend/tests/unit/platforms/tiktok/test_target_resolver.py

@@ -6,7 +6,9 @@ from app.proprietary.platforms.tiktok.targets import resolve_target
 
 
 def test_resolve_video_carries_username_and_id():
-    target = resolve_target("https://www.tiktok.com/@scout2015/video/6718335390845095173")
+    target = resolve_target(
+        "https://www.tiktok.com/@scout2015/video/6718335390845095173"
+    )
     assert target is not None
     assert target.kind == "video"
     assert target.value == "6718335390845095173"

surfsense_backend/tests/unit/platforms/tiktok/test_user_search.py

@@ -28,9 +28,7 @@ async def test_user_search_parses_dedupes_and_caps():
     async def fake_fetch(_url: str, _cap: int) -> list[dict]:
         return [_user("1", "nasa"), _user("1", "nasa"), _user("2", "nasa2")]
 
-    items = await search_tiktok_users(
-        ["nasa"], per_query=2, fetch_users=fake_fetch
-    )
+    items = await search_tiktok_users(["nasa"], per_query=2, fetch_users=fake_fetch)
 
     assert [i["id"] for i in items] == ["1", "2"]
     first = items[0]
@@ -49,9 +47,7 @@ async def test_user_search_empty_query_emits_error_item():
     async def fake_fetch(_url: str, _cap: int) -> list[dict]:
         return []
 
-    items = await search_tiktok_users(
-        ["ghost"], per_query=5, fetch_users=fake_fetch
-    )
+    items = await search_tiktok_users(["ghost"], per_query=5, fetch_users=fake_fetch)
 
     assert len(items) == 1
     assert items[0]["errorCode"] == "no_users"

surfsense_backend/tests/unit/services/test_model_connections.py

@@ -64,6 +64,22 @@ def test_openai_compatible_resolver_uses_explicit_api_base() -> None:
     assert ensure_v1("http://example.com/v1") == "http://example.com/v1"
 
 
+def test_openai_compatible_raw_resolver_does_not_append_v1() -> None:
+    model, kwargs = to_litellm(
+        {
+            "provider": "openai_compatible_raw",
+            "base_url": "https://ark.cn-beijing.volces.com/api/v3",
+            "api_key": "ark-key",
+            "extra": {},
+        },
+        "ep-20260101000000-test",
+    )
+
+    assert model == "openai/ep-20260101000000-test"
+    assert kwargs["api_base"] == "https://ark.cn-beijing.volces.com/api/v3"
+    assert kwargs["api_key"] == "ark-key"
+
+
 def test_ollama_resolver_uses_native_api_base() -> None:
     model, kwargs = to_litellm(
         {

surfsense_backend/tests/unit/services/test_requesty_model_normalizer.py

@@ -0,0 +1,103 @@
+"""Unit tests for Requesty model normalization.
+
+Mirrors the OpenRouter normalizer coverage but exercises Requesty's flat
+boolean capability fields (``supports_tool_calling`` / ``supports_vision``)
+and ``context_window`` sizing.
+"""
+
+from __future__ import annotations
+
+import pytest
+
+from app.services.requesty_model_normalizer import (
+    is_requesty_chat_model,
+    is_requesty_image_model,
+    normalize_requesty_models,
+    supports_image_input,
+    supports_tool_calling,
+)
+
+pytestmark = pytest.mark.unit
+
+
+def _requesty_model(
+    *,
+    model_id: str,
+    context_window: int = 128_000,
+    tools: bool = True,
+    vision: bool = False,
+    image_generation: bool = False,
+    name: str | None = None,
+) -> dict:
+    """Return a synthetic Requesty ``/v1/models`` entry.
+
+    Only the fields the normalizer inspects are populated; the live payload
+    carries many more (pricing, ``supports_caching``, ``description``, ...).
+    """
+    return {
+        "id": model_id,
+        "name": name or model_id,
+        "api": "chat",
+        "object": "model",
+        "context_window": context_window,
+        "supports_tool_calling": tools,
+        "supports_vision": vision,
+        "supports_image_generation": image_generation,
+    }
+
+
+def test_chat_model_requires_slash_tools_and_context():
+    assert is_requesty_chat_model(_requesty_model(model_id="openai/gpt-4o-mini"))
+    assert not is_requesty_chat_model(
+        _requesty_model(model_id="openai/gpt-4o-mini", tools=False)
+    )
+    assert not is_requesty_chat_model(
+        _requesty_model(model_id="openai/gpt-4o-mini", context_window=8_000)
+    )
+    assert not is_requesty_chat_model(_requesty_model(model_id="bare-model"))
+
+
+def test_excluded_provider_slug_is_filtered():
+    assert not is_requesty_chat_model(_requesty_model(model_id="amazon/nova-pro-v1"))
+
+
+def test_image_generation_models_excluded_from_chat_and_flagged():
+    image_model = _requesty_model(
+        model_id="google/gemini-2.5-flash-image", image_generation=True
+    )
+    assert not is_requesty_chat_model(image_model)
+    assert is_requesty_image_model(image_model)
+
+
+def test_capability_helpers_read_flat_booleans():
+    model = _requesty_model(
+        model_id="anthropic/claude-sonnet-4-5", vision=True, tools=True
+    )
+    assert supports_image_input(model) is True
+    assert supports_tool_calling(model) is True
+
+
+def test_normalize_maps_context_window_and_capabilities():
+    normalized = normalize_requesty_models(
+        [
+            _requesty_model(
+                model_id="openai/gpt-4o-mini",
+                context_window=128_000,
+                vision=True,
+                name="GPT-4o mini",
+            ),
+            _requesty_model(model_id="openai/gpt-4o-mini", tools=False),
+            _requesty_model(model_id="black-forest-labs/flux", image_generation=True),
+        ]
+    )
+
+    assert len(normalized) == 1
+    entry = normalized[0]
+    assert entry["model_id"] == "openai/gpt-4o-mini"
+    assert entry["display_name"] == "GPT-4o mini"
+    assert entry["supports_chat"] is True
+    assert entry["max_input_tokens"] == 128_000
+    assert entry["supports_image_input"] is True
+    assert entry["supports_tools"] is True
+    assert entry["supports_image_generation"] is False
+    assert entry["metadata"]["id"] == "openai/gpt-4o-mini"