feat: enhance search with include/exclude tokens and improved sorting

- Add token parsing to support include/exclude search terms using "-" prefix
- Implement token-based matching logic for relative path searches
- Improve search result sorting by prioritizing prefix matches and match position
- Add frontend test for multi-token highlighting with exclusion support

tests/frontend/components/autocomplete.behavior.test.js

@@ -136,4 +136,23 @@ describe('AutoComplete widget interactions', () => {
     expect(input.focus).toHaveBeenCalled();
     expect(input.setSelectionRange).toHaveBeenCalled();
   });
+
+  it('highlights multiple include tokens while ignoring excluded ones', async () => {
+    const input = document.createElement('textarea');
+    document.body.append(input);
+
+    const { AutoComplete } = await import(AUTOCOMPLETE_MODULE);
+    const autoComplete = new AutoComplete(input, 'loras', { showPreview: false });
+
+    const highlighted = autoComplete.highlightMatch(
+      'models/flux/beta-detail.safetensors',
+      'flux detail -beta',
+    );
+
+    const highlightCount = (highlighted.match(/<span/g) || []).length;
+    expect(highlightCount).toBe(2);
+    expect(highlighted).toContain('flux');
+    expect(highlighted).toContain('detail');
+    expect(highlighted).not.toMatch(/beta<\/span>/i);
+  });
 });

tests/services/test_relative_path_search.py

@@ -0,0 +1,63 @@
+import pytest
+
+from py.services.base_model_service import BaseModelService
+from py.utils.models import BaseModelMetadata
+
+
+class DummyService(BaseModelService):
+    async def format_response(self, model_data):
+        return model_data
+
+
+class FakeCache:
+    def __init__(self, raw_data):
+        self.raw_data = list(raw_data)
+
+
+class FakeScanner:
+    def __init__(self, raw_data, roots):
+        self._cache = FakeCache(raw_data)
+        self._roots = list(roots)
+
+    async def get_cached_data(self, *_args, **_kwargs):
+        return self._cache
+
+    def get_model_roots(self):
+        return list(self._roots)
+
+
+@pytest.mark.asyncio
+async def test_search_relative_paths_supports_multiple_tokens():
+    scanner = FakeScanner(
+        [
+            {"file_path": "/models/flux/detail-model.safetensors"},
+            {"file_path": "/models/flux/only-flux.safetensors"},
+            {"file_path": "/models/detail/flux-trained.safetensors"},
+            {"file_path": "/models/detail/standalone.safetensors"},
+        ],
+        ["/models"],
+    )
+    service = DummyService("stub", scanner, BaseModelMetadata)
+
+    matching = await service.search_relative_paths("flux detail")
+
+    assert matching == [
+        "flux/detail-model.safetensors",
+        "detail/flux-trained.safetensors",
+    ]
+
+
+@pytest.mark.asyncio
+async def test_search_relative_paths_excludes_tokens():
+    scanner = FakeScanner(
+        [
+            {"file_path": "/models/flux/detail-model.safetensors"},
+            {"file_path": "/models/flux/keep-me.safetensors"},
+        ],
+        ["/models"],
+    )
+    service = DummyService("stub", scanner, BaseModelMetadata)
+
+    matching = await service.search_relative_paths("flux -detail")
+
+    assert matching == ["flux/keep-me.safetensors"]