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 +1,85 @@
+import unittest
+
+from semantic_index.models import IndexDocument, SearchQuery, SearchResult
+from semantic_index.qdrant_store import build_filter, point_id_for_document
+from semantic_index.search import HybridSearchService, keyword_boost
+
+
+class FakeEmbedder:
+    def embed_query(self, text):
+        return [0.1, 0.2, 0.3]
+
+
+class FakeStore:
+    def __init__(self):
+        self.query = None
+
+    def search(self, vector, query, limit):
+        self.query = query
+        return [
+            SearchResult(
+                id="weak",
+                score=0.7,
+                text="general support text",
+                payload={"redmine_url": "http://redmine/issues/1"},
+            ),
+            SearchResult(
+                id="strong",
+                score=0.6,
+                text="Customer ada@example.com asked about ORD-12345",
+                payload={"redmine_url": "http://redmine/issues/2"},
+            ),
+        ][:limit]
+
+
+class SearchTest(unittest.TestCase):
+    def test_qdrant_point_id_is_deterministic_uuid_for_stable_document_id(self):
+        first = point_id_for_document("redmine:issue:42:journal:5:chunk:0")
+        second = point_id_for_document("redmine:issue:42:journal:5:chunk:0")
+
+        self.assertEqual(first, second)
+        self.assertRegex(first, r"^[0-9a-f-]{36}$")
+
+    def test_filter_maps_supported_metadata(self):
+        query = SearchQuery(
+            text="printer",
+            source="redmine",
+            project_identifier="fud-helpdesk",
+            doc_type="message",
+            issue_id=42,
+            contact_email="ada@example.com",
+            date_from="2026-04-01T00:00:00Z",
+            date_to="2026-04-30T23:59:59Z",
+        )
+
+        qfilter = build_filter(query)
+
+        self.assertEqual(
+            [
+                {"key": "source", "match": {"value": "redmine"}},
+                {"key": "project_identifier", "match": {"value": "fud-helpdesk"}},
+                {"key": "doc_type", "match": {"value": "message"}},
+                {"key": "issue_id", "match": {"value": 42}},
+                {"key": "contact_email", "match": {"value": "ada@example.com"}},
+                {"key": "created_on", "range": {"gte": "2026-04-01T00:00:00Z", "lte": "2026-04-30T23:59:59Z"}},
+            ],
+            qfilter["must"],
+        )
+
+    def test_keyword_boost_prioritizes_exact_email_and_order_matches(self):
+        weak = SearchResult(id="weak", score=0.7, text="general support text", payload={})
+        strong = SearchResult(id="strong", score=0.6, text="Customer ada@example.com asked about ORD-12345", payload={})
+
+        self.assertGreater(
+            keyword_boost('ada@example.com "ORD-12345"', strong),
+            keyword_boost('ada@example.com "ORD-12345"', weak),
+        )
+
+        service = HybridSearchService(embedder=FakeEmbedder(), store=FakeStore())
+        results = service.search(SearchQuery(text='ada@example.com "ORD-12345"', limit=2))
+        self.assertEqual("strong", results[0].id)
+        self.assertEqual("http://redmine/issues/2", results[0].citation["url"])
+
+
+if __name__ == "__main__":
+    unittest.main()