fix: KB删除时清理VDB/Graph/physical文件

- 修复物理文件删除:路径不以/idfile/开头,改为检查/开头
- 新增VDB向量删除:调用vectordb API删除该KB的所有向量
- 新增Graph图数据库清理
- 新增entities/entity_relations表清理
This commit is contained in:
ymq 2026-08-11 17:03:33 +08:00
parent 3b1cda8d2f
commit 239b82c85d

View File

@ -9,10 +9,10 @@ if not kb_id:
env = request._run_ns
userorgid = await env.get_userorgid()
deleted = {"documents": 0, "chunks": 0, "tags": 0, "media_tags": 0, "files": 0}
deleted = {"documents": 0, "chunks": 0, "tags": 0, "media_tags": 0,
"files": 0, "vdb": 0, "graph": 0, "entities": 0}
async with get_sor_context(env, 'rag') as sor:
# Check owner
kb = await sor.sqlExe(
"SELECT id, name FROM knowledge_bases WHERE id=${id}$ AND (org_id=${org_id}$ OR org_id IS NULL)",
{"id": kb_id, "org_id": userorgid})
@ -20,52 +20,78 @@ async with get_sor_context(env, 'rag') as sor:
return {"widgettype": "urlwidget",
"options": {"url": entire_url('/rag/knowledge_bases_list/index.ui')}}
# Get all doc IDs and file paths for physical cleanup
docs = await sor.sqlExe(
"SELECT id, file_path FROM documents WHERE kb_id=${kb_id}$",
{"kb_id": kb_id})
doc_ids = [d.id for d in docs]
# 1. VDB cleanup
if doc_ids:
# Delete chunks
await sor.sqlExe(
"DELETE FROM document_chunks WHERE doc_id IN (${ids}$)",
chunks = await sor.sqlExe(
"SELECT vector_id FROM document_chunks WHERE doc_id IN (${ids}$) AND vector_id IS NOT NULL AND vector_id != ''",
{"ids": doc_ids})
deleted["chunks"] = len(doc_ids)
vector_ids = [c.vector_id for c in chunks]
if vector_ids:
try:
client = StreamHttpClient()
await client.request('POST',
'https://vectordb.opencomputing.net:10443/v1/delete',
json={"colname": kb_id, "ids": vector_ids})
deleted["vdb"] = len(vector_ids)
except Exception as e:
info(f'delete_kb vdb error: {e}')
# Delete media_tags
mt = await sor.sqlExe(
"SELECT COUNT(*) AS cnt FROM media_tags WHERE kb_id=${kb_id}$",
{"kb_id": kb_id})
await sor.sqlExe(
"DELETE FROM media_tags WHERE kb_id=${kb_id}$",
{"kb_id": kb_id})
# 2. Graph DB cleanup
try:
client2 = StreamHttpClient()
await client2.request('POST',
'https://graph.opencomputing.net:10443/api/graph/delete',
json={"graph": kb_id})
deleted["graph"] = 1
except Exception as e:
info(f'delete_kb graph error: {e}')
# 3. Entities & relations
er = await sor.sqlExe(
"SELECT COUNT(*) AS cnt FROM entity_relations WHERE kb_id=${kb_id}$",
{"kb_id": kb_id})
await sor.sqlExe("DELETE FROM entity_relations WHERE kb_id=${kb_id}$", {"kb_id": kb_id})
deleted["entities"] += er[0].cnt if er else 0
ent = await sor.sqlExe(
"SELECT COUNT(*) AS cnt FROM entities WHERE kb_id=${kb_id}$",
{"kb_id": kb_id})
await sor.sqlExe("DELETE FROM entities WHERE kb_id=${kb_id}$", {"kb_id": kb_id})
deleted["entities"] += ent[0].cnt if ent else 0
# 4. Chunks
ch = await sor.sqlExe(
"SELECT COUNT(*) AS cnt FROM document_chunks WHERE doc_id IN (${ids}$)",
{"ids": doc_ids})
await sor.sqlExe("DELETE FROM document_chunks WHERE doc_id IN (${ids}$)", {"ids": doc_ids})
deleted["chunks"] = ch[0].cnt if ch else 0
# Media tags
mt = await sor.sqlExe("SELECT COUNT(*) AS cnt FROM media_tags WHERE kb_id=${kb_id}$", {"kb_id": kb_id})
await sor.sqlExe("DELETE FROM media_tags WHERE kb_id=${kb_id}$", {"kb_id": kb_id})
deleted["media_tags"] = mt[0].cnt if mt else 0
# Delete tags
t = await sor.sqlExe(
"SELECT COUNT(*) AS cnt FROM tags WHERE kb_id=${kb_id}$",
{"kb_id": kb_id})
await sor.sqlExe(
"DELETE FROM tags WHERE kb_id=${kb_id}$",
{"kb_id": kb_id})
# Tags
t = await sor.sqlExe("SELECT COUNT(*) AS cnt FROM tags WHERE kb_id=${kb_id}$", {"kb_id": kb_id})
await sor.sqlExe("DELETE FROM tags WHERE kb_id=${kb_id}$", {"kb_id": kb_id})
deleted["tags"] = t[0].cnt if t else 0
# Delete documents
await sor.sqlExe(
"DELETE FROM documents WHERE kb_id=${kb_id}$",
{"kb_id": kb_id})
# Documents
await sor.sqlExe("DELETE FROM documents WHERE kb_id=${kb_id}$", {"kb_id": kb_id})
deleted["documents"] = len(doc_ids)
# Delete the KB itself
await sor.sqlExe(
"DELETE FROM knowledge_bases WHERE id=${id}$",
{"id": kb_id})
# KB itself
await sor.sqlExe("DELETE FROM knowledge_bases WHERE id=${id}$", {"id": kb_id})
# Delete physical files
# Physical files — paths are like /44/126/174/61/file.mp4
for d in docs:
fp = d.file_path or ''
if fp and fp.startswith("/idfile/"):
if fp and fp.startswith("/"):
try:
real_path = env.realpath(fp)
if os.path.exists(real_path):
@ -74,5 +100,6 @@ for d in docs:
except Exception:
pass
info(f'delete_kb {kb_id}: {deleted}')
return {"widgettype": "urlwidget",
"options": {"url": entire_url('/rag/knowledge_bases_list/index.ui')}}