feat: 增强知识库功能,优化索引和RAG检索

This commit is contained in:
caoxiaozhu
2026-05-18 02:49:39 +00:00
parent 55e0591a5e
commit 4414ffb34c
18 changed files with 5656 additions and 4659 deletions

View File

@@ -74,6 +74,28 @@ def test_build_hits_prioritizes_answer_clue_appendix_for_rule_queries() -> None:
assert [item["candidate_id"] for item in hits] == ["clue-1", "plain-1"]
def test_build_hits_demotes_chapter_navigation_for_specific_rule_queries() -> None:
hits = KnowledgeRagService._build_hits_from_query_data(
query="探亲差旅归哪个部门管理?",
chunks=[
{
"chunk_id": "toc-1",
"file_path": "/tmp/doc-1__费用制度.md",
"content": "# 章节导航\n\n- 第一章 总则\n- 第二章 职责分工\n- 第三章 支出归口",
},
{
"chunk_id": "body-1",
"file_path": "/tmp/doc-1__费用制度.md",
"content": "附表3支出归口管理部门与归口业务范围\n组织人事部:探亲差旅、条件艰苦及安全风险较高区域补助等支出。",
},
],
entities=[],
limit=2,
)
assert [item["candidate_id"] for item in hits] == ["body-1", "toc-1"]
def test_resolve_default_qdrant_url_prefers_container_host(monkeypatch) -> None:
monkeypatch.setattr(
knowledge_rag_module.socket,
@@ -93,3 +115,29 @@ def test_resolve_default_qdrant_url_falls_back_to_loopback(monkeypatch) -> None:
monkeypatch.setattr(knowledge_rag_module.socket, "getaddrinfo", raise_lookup_error)
assert knowledge_rag_module._resolve_default_qdrant_url() == "http://127.0.0.1:6333"
def test_is_query_ready_status_rejects_failed_status_even_with_chunks() -> None:
assert (
KnowledgeRagService.is_query_ready_status(
{
"status": "failed",
"chunks_count": 11,
"chunks_list": ["chunk-1"],
}
)
is False
)
def test_is_query_ready_status_rejects_processing_status_even_with_chunks() -> None:
assert (
KnowledgeRagService.is_query_ready_status(
{
"status": "processing",
"chunks_count": 11,
"chunks_list": ["chunk-1"],
}
)
is False
)