YG-Soft/X-Financial
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(server): 重构知识库服务和路由配置，优化LLM维基知识管理接口，增强知识检索能力

Browse Source
This commit is contained in:
caoxiaozhu
2026-05-15 09:33:59 +00:00
parent 7a3feb14a0
commit 6793b6f832
7 changed files with 812 additions and 123 deletions
Download Patch File Download Diff File Expand all files Collapse all files

186
server/tests/test_llm_wiki_service.py
View File

@@ -16,7 +16,7 @@ from app.core.agent_enums import AgentReviewStatus, AgentRunSource, AgentRunStat
from app.db.base import Base
from app.main import create_app
from app.schemas.agent_asset import AgentAssetReviewCreate
from app.schemas.knowledge import LlmWikiSummaryUpdateWrite, LlmWikiSyncRead
from app.schemas.knowledge import LlmWikiSummaryUpdateWrite
from app.services.agent_assets import AgentAssetService
from app.services.agent_runs import AgentRunService
from app.services.knowledge import (
@@ -25,7 +25,7 @@ from app.services.knowledge import (
KNOWLEDGE_INGEST_STATUS_PUBLISHED,
KnowledgeService,
)
from app.services.llm_wiki import LlmWikiService
from app.services.llm_wiki import CandidateModelAttempt, LlmWikiService
def build_session() -> Session:
@@ -86,6 +86,36 @@ def upload_policy_document(storage_root: Path, *, filename: str = "公司差旅
return document.id
def upload_multipage_policy_document(storage_root: Path, *, filename: str = "公司支出管理办法.txt") -> str:
service = KnowledgeService(storage_root=storage_root)
service.ensure_library_ready()
document = service.upload_document(
folder="报销制度",
filename=filename,
content=(
"商密【中】\n"
"关于颁布《公司支出管理办法》的通知\n"
"特此通知。\n"
"\f"
"目录\n"
"第一章 总则................................4\n"
"第二章 报销审批................................7\n"
"\f"
"第一条 报销申请\n"
"员工提交报销申请时，应附发票、行程单和审批说明。\n"
"第二条 报销审批\n"
"住宿费超过制度标准时，必须升级至总经理审批。\n"
"第三条 附件补充\n"
"缺少附件时不得提交报销。\n"
"\f"
"第四条 财务复核\n"
"财务复核时应校验预算、发票真伪和审批链完整性。\n"
).encode("utf-8"),
current_user=build_admin_user(),
)
return document.id
def build_candidate_payload(chunk_id: str, *, summary: str = "住宿费超过标准时必须升级审批。") -> dict[str, object]:
return {
"knowledge_candidates": [
@@ -222,10 +252,14 @@ def test_llm_wiki_sync_creates_artifacts_and_draft_rule(tmp_path, monkeypatch) -
document_payload = json.loads((document_dir / "document.json").read_text(encoding="utf-8"))
assert document_payload["sync_reason"] == "initial_build"
assert document_payload["quality_status"] == "formal"
assert document_payload["formal_knowledge_candidate_count"] == 1
assert document_payload["fallback_knowledge_candidate_count"] == 0
detail = service.get_document_detail(document_id)
assert "公司差旅报销制度.txt 知识总结" in detail.knowledge_summary_markdown
assert "住宿费升级审批要求" in detail.knowledge_summary_markdown
assert detail.quality_status == "formal"
asset = AgentAssetService(db).get_asset(result.generated_rule_asset_ids[0])
assert asset is not None
@@ -386,9 +420,91 @@ def test_llm_wiki_sync_uses_fallback_candidates_when_system_hermes_times_out(
assert result.knowledge_candidate_count >= 1
assert runtime_called["count"] == 0
detail = KnowledgeService(storage_root=tmp_path).get_document_detail(document_id)
assert detail.stateCode == KNOWLEDGE_INGEST_STATUS_INGESTED
assert detail.state == "已归纳"
knowledge_service = KnowledgeService(storage_root=tmp_path)
detail = knowledge_service.get_document_detail(document_id)
assert detail.stateCode == KNOWLEDGE_INGEST_STATUS_FAILED
assert detail.state == "归纳失败"
assert detail.llmWikiAvailable is True
assert detail.llmWikiQualityStatus == "fallback_only"
document_payload = json.loads(
(
tmp_path
/ "knowledge"
/ ".llm_wiki"
/ "documents"
/ document_id
/ "document.json"
).read_text(encoding="utf-8")
)
assert document_payload["quality_status"] == "fallback_only"
assert document_payload["formal_knowledge_candidate_count"] == 0
assert document_payload["fallback_knowledge_candidate_count"] == 1
candidates_payload = json.loads(
(
tmp_path
/ "knowledge"
/ ".llm_wiki"
/ "documents"
/ document_id
/ "knowledge_candidates.json"
).read_text(encoding="utf-8")
)
assert candidates_payload[0]["extraction_mode"] == "fallback"
assert "fallback_only" in candidates_payload[0]["quality_flags"]
def test_llm_wiki_sync_continues_after_single_group_failure(tmp_path, monkeypatch) -> None:
document_id = upload_multipage_policy_document(tmp_path, filename="多页支出制度.txt")
call_count = {"count": 0}
def fake_call_candidate_model(self, *, entry, chunk_group):
call_count["count"] += 1
if call_count["count"] == 1:
return CandidateModelAttempt(
payload={},
source="hermes",
ok=False,
failure_reason="simulated_timeout",
)
return build_candidate_payload(chunk_group[0]["chunk_id"])
monkeypatch.setattr(LlmWikiService, "_call_candidate_model", fake_call_candidate_model)
with build_session() as db:
service = LlmWikiService(db, storage_root=tmp_path)
result = service.sync_folder(
folder="报销制度",
current_user=build_admin_user(),
document_ids=[document_id],
)
detail = service.get_document_detail(document_id)
assert result.document_count == 1
assert call_count["count"] >= 2
assert detail.quality_status == "partial_degraded"
assert detail.successful_group_count >= 1
assert detail.failed_group_count >= 1
assert detail.formal_knowledge_candidate_count >= 1
knowledge_detail = KnowledgeService(storage_root=tmp_path).get_document_detail(document_id)
assert knowledge_detail.stateCode == KNOWLEDGE_INGEST_STATUS_INGESTED
assert knowledge_detail.llmWikiQualityStatus == "partial_degraded"
def test_llm_wiki_filters_cover_and_catalog_chunks_before_candidate_extraction(tmp_path) -> None:
document_id = upload_multipage_policy_document(tmp_path, filename="封面目录过滤测试.txt")
with build_session() as db:
service = LlmWikiService(db, storage_root=tmp_path)
text = service.knowledge_service.extract_document_text(document_id)
chunks = service._build_chunks(document_id=document_id, text=text)
candidate_chunks = service._select_candidate_chunks(chunks)
assert len(chunks) > len(candidate_chunks)
assert candidate_chunks
assert all(int(item.get("source_page") or 0) >= 3 for item in candidate_chunks)
def test_llm_wiki_sync_skips_unchanged_and_rebuilds_on_updated_at_change(tmp_path, monkeypatch) -> None:
@@ -475,22 +591,46 @@ def test_llm_wiki_sync_does_not_overwrite_active_rule(tmp_path, monkeypatch) ->
def test_llm_wiki_sync_endpoint_records_agent_run(monkeypatch) -> None:
def fake_sync_folder(self, *, folder="报销制度", current_user, document_ids=None, force=False):
return LlmWikiSyncRead(
ok=True,
run_id="wiki_test_sync",
folder=folder,
document_count=1,
knowledge_candidate_count=2,
rule_candidate_count=1,
generated_rule_count=1,
generated_rule_asset_ids=["asset-rule-1"],
summary="已完成 Hermes LLM Wiki 同步。",
)
monkeypatch.setattr(LlmWikiService, "sync_folder", fake_sync_folder)
client, session_factory = build_client()
def fake_submit_sync(*, agent_run_id, folder, current_user, document_ids=None, force=False):
with session_factory() as db:
service = AgentRunService(db)
service.record_tool_call(
run_id=agent_run_id,
tool_type="llm",
tool_name="system_hermes_llm_wiki_sync",
request_json={
"folder": folder,
"document_ids": list(document_ids or []),
"force": force,
},
response_json={"run_id": "wiki_test_sync"},
status="succeeded",
duration_ms=0,
)
service.merge_route_json(
agent_run_id,
{
"phase": "succeeded",
"sync_run_id": "wiki_test_sync",
"progress": {
"total_documents": len(document_ids or []),
"completed_documents": len(document_ids or []),
"failed_documents": 0,
"skipped_documents": 0,
"percent": 100,
},
},
status=AgentRunStatus.SUCCEEDED.value,
result_summary="已完成 Hermes LLM Wiki 同步。",
)
monkeypatch.setattr(
"app.services.llm_wiki_tasks.llm_wiki_task_manager.submit_sync",
fake_submit_sync,
)
with session_factory() as db:
before_count = len(AgentRunService(db).list_runs(limit=100))
@@ -506,8 +646,8 @@ def test_llm_wiki_sync_endpoint_records_agent_run(monkeypatch) -> None:
assert response.status_code == 200
payload = response.json()
assert payload["run_id"] == "wiki_test_sync"
assert payload["generated_rule_count"] == 1
assert payload["agent_run_id"].startswith("run_")
assert payload["status"] == AgentRunStatus.RUNNING.value
with session_factory() as db:
service = AgentRunService(db)
@@ -521,4 +661,4 @@ def test_llm_wiki_sync_endpoint_records_agent_run(monkeypatch) -> None:
assert latest_run.tool_calls
assert latest_run.tool_calls[0].tool_name == "system_hermes_llm_wiki_sync"
assert latest_run.tool_calls[0].status == "succeeded"
assert latest_run.tool_calls[0].response_json["run_id"] == "wiki_test_sync"
assert latest_run.route_json["sync_run_id"] == "wiki_test_sync"