Add semantic-index service, deployment assets, and tests

2026-05-04 09:50:03 -04:00
parent faad70872b
commit b305544f63
42 changed files with 5059 additions and 0 deletions
@@ -0,0 +1,115 @@
+import unittest
+from pathlib import Path
+
+from semantic_index.app import create_app
+from semantic_index.config import Settings
+from semantic_index.models import SearchResult
+
+
+class FakeSearchService:
+    def search(self, query):
+        return [
+            SearchResult(
+                id="redmine:issue:1:chunk:0",
+                score=0.8,
+                text="Snippet text",
+                payload={
+                    "source": "redmine",
+                    "project_identifier": "customer-service",
+                    "doc_type": "issue",
+                    "issue_id": 1,
+                    "redmine_url": "http://redmine/issues/1",
+                    "source_record_id": "issue:1",
+                },
+            )
+        ]
+
+    def get_document(self, document_id):
+        return {"id": document_id, "text": "Full text", "payload": {}}
+
+
+class FakeStore:
+    def list_projects(self, source=None, limit=1000):
+        return [{"project_identifier": "customer-service", "document_count": 10}]
+
+
+class FakeRefreshService:
+    def __init__(self):
+        self.calls = []
+
+    def refresh_redmine_project_limits(self, project_limits, dry_run=False, force_rebuild=False, overlap_minutes=15):
+        self.calls.append((project_limits, dry_run, force_rebuild, overlap_minutes))
+        return {"source": "redmine", "projects": len(project_limits), "dry_run": dry_run}
+
+
+def fake_services():
+    refresh = FakeRefreshService()
+    return {
+        "settings": Settings(
+            openai_api_key="",
+            qdrant_url="http://qdrant",
+            qdrant_api_key=None,
+            qdrant_collection="semantic",
+            redmine_url="http://redmine",
+            redmine_api_key="",
+            redmine_project_identifier=None,
+            sample_limit=50,
+            bind_host="127.0.0.1",
+            bind_port=8787,
+            service_api_key=None,
+            refresh_state_path=Path(".cache/semantic_index/refresh_state.json"),
+        ),
+        "search": FakeSearchService(),
+        "store": FakeStore(),
+        "refresh": refresh,
+    }
+
+
+class SemanticIndexAppTest(unittest.TestCase):
+    def test_health_does_not_build_live_services(self):
+        def broken_builder():
+            raise AssertionError("health should not build live clients")
+
+        app = create_app(service_builder=broken_builder)
+        routes = {route.path: route.endpoint for route in app.routes}
+
+        self.assertEqual({"status": "ok"}, routes["/health"]())
+
+    def test_search_endpoint_returns_normalized_agent_response(self):
+        app = create_app(service_builder=fake_services)
+        routes = {route.path: route.endpoint for route in app.routes}
+
+        response = routes["/search"]({"query": "printer", "project_identifier": "customer-service", "limit": 3})
+
+        self.assertEqual("printer", response["query"])
+        self.assertEqual("customer-service", response["filters"]["project_identifier"])
+        self.assertEqual("customer-service", response["results"][0]["citation"]["project_identifier"])
+
+    def test_projects_endpoint_lists_indexed_projects(self):
+        app = create_app(service_builder=fake_services)
+        routes = {route.path: route.endpoint for route in app.routes}
+
+        response = routes["/projects"]()
+
+        self.assertEqual("customer-service", response["projects"][0]["project_identifier"])
+
+    def test_refresh_endpoint_passes_project_limits_and_cost_flags(self):
+        services = fake_services()
+        app = create_app(service_builder=lambda: services)
+        routes = {route.path: route.endpoint for route in app.routes}
+
+        response = routes["/sources/redmine/refresh"](
+            {
+                "project_limits": {"customer-service": 5},
+                "dry_run": True,
+                "force_rebuild": False,
+                "overlap_minutes": 30,
+            }
+        )
+
+        self.assertTrue(response["dry_run"])
+        self.assertEqual(({"customer-service": 5}, True, False, 30), services["refresh"].calls[0])
+
+
+if __name__ == "__main__":
+    unittest.main()