sync: 鍚屾灞€鍩熺綉鏈€鏂颁唬鐮?(source master @ b69568e4)

offline-backend-20260512/backend/tests/test_knowledge_live_embedding.py

@@ -0,0 +1,62 @@
+"""Opt-in acceptance against an actual OpenAI-compatible embedding endpoint."""
+
+import os
+
+import pytest
+
+from deerflow.config.llmwiki_config import LocalWikiEmbeddingConfig, LocalWikiIndexConfig
+from deerflow.integrations.weknora.local_index.embedding import StrictWikiEmbeddingClient
+from deerflow.integrations.weknora.local_index.normalize import normalize_wiki_page
+from deerflow.integrations.weknora.local_index.search import WikiVectorSearchService
+from deerflow.integrations.weknora.local_index.vector_cache import WikiVectorCache
+from deerflow.integrations.weknora.local_index.vector_codec import encode_vector
+from deerflow.persistence.llmwiki_index.memory import MemoryLlmWikiIndexStore
+
+
+@pytest.mark.asyncio
+@pytest.mark.skipif(not os.getenv("KNOWLEDGE_TEST_EMBEDDING_KEY"), reason="Live embedding credentials not configured")
+async def test_real_embedding_search_preserves_library_scope_and_excludes_drafts():
+    client = StrictWikiEmbeddingClient(
+        LocalWikiEmbeddingConfig(
+            provider="openai_compatible",
+            base_url=os.getenv("KNOWLEDGE_TEST_EMBEDDING_URL", "https://api.siliconflow.cn/v1"),
+            model=os.getenv("KNOWLEDGE_TEST_EMBEDDING_MODEL", "BAAI/bge-m3"),
+            api_key=os.environ["KNOWLEDGE_TEST_EMBEDDING_KEY"],
+            dimensions=1024,
+        )
+    )
+    content = "设备过热时应先断电，再检查风道和散热风扇。"
+    try:
+        vectors = await client.embed_texts([content, "读书交流活动通过分享心得促进阅读。"])
+        assert [len(vector) for vector in vectors] == [1024, 1024]
+        store = MemoryLlmWikiIndexStore()
+        for base_id in ("selected", "other"):
+            await store.try_acquire_sync_lease(base_id, owner="test", sync_id="sync", lease_seconds=60, fingerprint=client.fingerprint)
+            for slug, status, index in (("cooling", "published", 0), ("books", "published", 1), ("draft", "draft", 0)):
+                page = normalize_wiki_page(base_id, {"slug": slug, "title": slug, "content": content if index == 0 else "读书交流活动", "status": status})
+                blob, _ = encode_vector(vectors[index], dimensions=1024)
+                await store.replace_page_vectors(
+                    page=page,
+                    vectors=[
+                        {
+                            "section_index": 0,
+                            "heading": slug,
+                            "section_content": content if index == 0 else "读书交流活动",
+                            "content_hash": slug,
+                            "embedding_dimensions": 1024,
+                            "vector_blob": blob,
+                        }
+                    ],
+                    fingerprint=client.fingerprint,
+                    sync_id="sync",
+                )
+        service = WikiVectorSearchService(LocalWikiIndexConfig(min_similarity=0, top_k_sections=10, top_k_pages=8), client, WikiVectorCache(store, revision_check_seconds=0))
+        result = await service.search("设备太热应该如何处理？", [{"id": "selected", "name": "验收设备库"}])
+        assert result["results"][0]["wiki_slug"] == "cooling"
+        assert {row["wiki_slug"] for row in result["results"]} == {"cooling", "books"}
+        assert all("selected" in row["url"] for row in result["results"])
+        overview = await service.search("此知识库都讲了什么？", [{"id": "selected", "name": "验收设备库"}])
+        assert overview["mode"] == "overview"
+        assert overview["results"]
+    finally:
+        await client.close()