tools: 强制 spark_submit 的 script_path 必须来自 upload_file

在 spark_submit handler 中增加 server-minted 校验:script_path 必须通过
本服务器 upload_file 的 UploadStore.Validate,否则立即返回清晰错误,避免
远程 agent 把自身本地路径转发给服务器侧 spark-submit 时出现难以定位的
FileNotFoundException。

校验对 nil UploadStore 是安全的:当 d.UploadStore 为 nil 时直接跳过,保留
不强制依赖 store 的测试或降级场景。当前 testDepsWithDataDir 始终会注入
UploadStore,所以测试走的是真实校验路径。

同步更新了 script_path 的工具描述,明确声明非 upload_file 铸造的路径会被
拒绝。这是对 7920dc9 中“path 永远在最后”规则的收紧——现在不仅位置固定,
而且必须是本机 uploads 目录下的有效上传文件。

测试调整:
- 新增 TestSparkSubmit_RejectsNonMintedPath:使用 t.TempDir() 下未通过
  upload_file 写入的文件作为 script_path,断言返回错误并包含
  "not from upload_file"。
- TestSparkSubmit_MissingRequiredField / TestSparkSubmit_BadSparkConfValue
  原来使用 "/tmp/script.py" 作为占位路径,现在会先通过 UploadStore.Save
  生成一个铸造路径,确保它们继续分别验证 master 缺失和 spark_conf 值类型
  错误,而不是被新的路径守卫拦截。

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
tao.chen
2026-07-13 19:49:26 +08:00
co-authored by Claude
parent 7920dc9597
commit c9e54a3f27
2 changed files with 57 additions and 3 deletions
+46 -2
View File
@@ -97,11 +97,16 @@ func TestSparkSubmit_StructuredCommand(t *testing.T) {
func TestSparkSubmit_MissingRequiredField(t *testing.T) {
deps, _ := testDepsWithDataDir(t)
store := deps.UploadStore
_, _, _, _, mintedPath, err := store.Save([]byte("# dummy\n"), "dummy.py")
if err != nil {
t.Fatalf("save upload: %v", err)
}
req := newToolRequest(SparkSubmitName, map[string]any{
"cluster_id": "cluster-echo",
"deploy_mode": "cluster",
"script_path": "/tmp/script.py",
"script_path": mintedPath,
"queue": "default",
"executor_memory": "2G",
"executor_cores": 1,
@@ -125,12 +130,17 @@ func TestSparkSubmit_MissingRequiredField(t *testing.T) {
func TestSparkSubmit_BadSparkConfValue(t *testing.T) {
deps, _ := testDepsWithDataDir(t)
store := deps.UploadStore
_, _, _, _, mintedPath, err := store.Save([]byte("# dummy\n"), "dummy.py")
if err != nil {
t.Fatalf("save upload: %v", err)
}
req := newToolRequest(SparkSubmitName, map[string]any{
"cluster_id": "cluster-echo",
"master": "yarn",
"deploy_mode": "cluster",
"script_path": "/tmp/script.py",
"script_path": mintedPath,
"queue": "default",
"executor_memory": "2G",
"executor_cores": 1,
@@ -152,3 +162,37 @@ func TestSparkSubmit_BadSparkConfValue(t *testing.T) {
t.Errorf("error text=%q, want mention of spark_conf", text.Text)
}
}
func TestSparkSubmit_RejectsNonMintedPath(t *testing.T) {
deps, _ := testDepsWithDataDir(t)
unminted := filepath.Join(t.TempDir(), "unminted.py")
if err := os.WriteFile(unminted, []byte("print('not from upload_file')\n"), 0o644); err != nil {
t.Fatalf("write unminted file: %v", err)
}
req := newToolRequest(SparkSubmitName, map[string]any{
"cluster_id": "cluster-echo",
"master": "yarn",
"deploy_mode": "cluster",
"script_path": unminted,
"queue": "default",
"executor_memory": "2G",
"executor_cores": 1,
"num_executors": 4,
})
res, err := deps.SparkSubmitHandler(context.Background(), req)
if err != nil {
t.Fatalf("handler error: %v", err)
}
if !res.IsError {
t.Fatalf("expected error result, got: %v", res.Content)
}
text, ok := mcp.AsTextContent(res.Content[0])
if !ok {
t.Fatalf("content is not text: %T", res.Content[0])
}
if !strings.Contains(text.Text, "not from upload_file") {
t.Errorf("error text=%q, want mention of not from upload_file", text.Text)
}
}