Files
HomeAgent/internal/plugins/integration_test.go
JianFeeeee 125bf57cfa fix(test): 测试不再抢生产端口 :8080(internal/plugins 加载内置 webui 所致)
部署过程中反复出现「8080 被 plugins.test 占用」导致生产 WebUI 起不来。
追到底:internal/plugins 的集成测试会 pluginReg.Load(全部内置插件),
其中 webui 默认监听 :8080 —— **正是生产实例的端口**。

## 为什么这个 bug 特别难查

它不是测试失败,而是**测试与生产静默抢端口**:先到者胜,另一个 bind 失败。

  - 跑测试的人看到「测试随机失败」(其实是生产先占了)
  - 用生产的人看到「WebUI 随机死掉」(其实是测试先占了)
  - 两边现象互不相干,且各自单独重跑往往都过

叠加 webui 已有的「bind 失败必须显式报错」修复后,症状从「静默死亡」
变成「随机报错」,这反而让归属更容易看错 —— 我一开始也是先怀疑自己的
部署脚本,直到采样 /proc/<pid>/cwd 才确认是测试进程。

## 修法

集成测试在 Load 之前用既有的 SetListenOverride 把地址指到
127.0.0.1:0(内核分配空闲端口),并在 cleanup 还原。
测试因此拿到真实可用的 HTTP 服务,且与任何固定端口实例完全隔离。

- webui 新增 ListenOverride() 读取当前值,供调用方保存/还原
  (只有 setter 时无法在不破坏调用方状态的前提下做临时覆盖)。

## 验证

- 修复前:跑 ./internal/plugins/ 期间 8080 持续归 plugins.test(109/200 采样)
- 修复后:35 次采样全程 8080 归 homed,测试同时全绿
- 变异验证:移除 override 后立刻复现抢端口,判据有效

另记两个测试卫生问题(同一根源,已顺手清理泄漏进程):
测试会启动**真实插件进程**(/home/newqqagent/plugins/*/plugin.bin)。
kill 测试进程后这些子进程会残留。已全部清理,生产 23 插件恢复正常。
2026-09-26 13:49:59 +08:00

629 lines
17 KiB
Go
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

package plugins
import (
"encoding/json"
"os"
"path/filepath"
"testing"
"time"
agentAPI "gitcode.com/JianFeeeee/HomeAgent/internal/agent/api"
agentCore "gitcode.com/JianFeeeee/HomeAgent/internal/agent/core"
agentIO "gitcode.com/JianFeeeee/HomeAgent/internal/agent/io"
internalConfig "gitcode.com/JianFeeeee/HomeAgent/internal/config"
"gitcode.com/JianFeeeee/HomeAgent/internal/knowledge"
luaVM "gitcode.com/JianFeeeee/HomeAgent/internal/lua"
"gitcode.com/JianFeeeee/HomeAgent/internal/memory"
doc "gitcode.com/JianFeeeee/HomeAgent/internal/memory/document"
"gitcode.com/JianFeeeee/HomeAgent/internal/plugin"
cli "gitcode.com/JianFeeeee/HomeAgent/internal/plugins/cli"
"gitcode.com/JianFeeeee/HomeAgent/internal/plugins/webui"
sdk "gitcode.com/JianFeeeee/HomeAgent/internal/sdk"
)
type testPluginEnv struct {
tmpDir string
stageHost *agentCore.StageHost
iom *agentIO.IOManager
pluginReg *plugin.Registry
memDB *memory.GraphDB
ks *knowledge.Store
docStore *doc.Store
cfgReg *internalConfig.ConfigRegistry
}
func setupIntegration(t *testing.T) *testPluginEnv {
return setupIntegrationWithProvider(t, nil)
}
func setupIntegrationWithProvider(t *testing.T, pm *agentAPI.ProviderManager) *testPluginEnv {
t.Helper()
// 测试绝不能绑生产端口。这里加载的是**全部内置插件**(含 webui),
// 而 webui 默认监听 :8080 —— 那正是生产实例的端口。
// 不覆盖的后果不是测试失败,而是**测试静默地跟生产抢端口**:
// 先到者胜,另一个 bind 失败。webui 现在会在 bind 失败时明确报错,
// 于是表现为「测试随机失败」+「生产 WebUI 随机死掉」,且两边看起来
// 互不相干,极难定位(实际踩到过)。
//
// 用 SetListenOverride 指到 127.0.0.1:0(内核分配空闲端口):
// 测试拿到真实可用的 HTTP 服务,且与任何固定端口实例互不干扰。
if prev := webui.ListenOverride(); prev == "" {
webui.SetListenOverride("127.0.0.1:0")
t.Cleanup(func() { webui.SetListenOverride(prev) })
}
tmpDir, err := os.MkdirTemp("", "hc_integration_*")
if err != nil {
t.Fatal(err)
}
stageHost := agentCore.NewStageHost()
iom := agentIO.NewIOManager()
pluginReg := plugin.NewRegistry()
pluginReg.SetIOManager(iom)
pluginReg.SetMemory(nil)
pluginReg.SetToolRegistrar(func(name string, def sdk.ToolDef, handler sdk.ToolHandler) error {
return stageHost.RegisterTool(name, def, handler)
})
pluginReg.SetStageRegistrar(func(stage sdk.Stage, handler sdk.StageHandler) {
stageHost.RegisterStage(stage, handler)
})
pluginReg.SetAPIRegistrar(func(name string) error {
return nil
})
memDB, err := memory.NewGraphDB(filepath.Join(tmpDir, "test.db"))
if err != nil {
t.Fatal(err)
}
pluginReg.SetMemory(memDB)
ks := knowledge.NewStore(filepath.Join(tmpDir, "knowledge"))
if err := ks.Start(); err != nil {
t.Fatal(err)
}
docStore := doc.NewStore(filepath.Join(tmpDir, "documents"), memory.TokenizeWords)
if err := docStore.Start(); err != nil {
t.Fatal(err)
}
pluginReg.SetStageHost(stageHost)
pluginReg.SetProviderManager(pm)
pluginReg.SetKnowledge(ks)
pluginReg.SetDocStore(docStore)
cli.DefaultSocket = filepath.Join(tmpDir, "cli.sock")
// 经 ConfigRegistry 装配内核路径配置(clawhubadapter/pluginmgr 等经 SDK settings 读取)
cfgReg := internalConfig.NewConfigRegistry("")
cfgReg.SeedDefaults(tmpDir)
// ★ 预置临时端口,避免测试之间(以及本机生产实例)抢固定默认端口。
//
// 为什么必须在 Load 之前预置:插件表由 RegisterDef 在插件 Start 时创建,
// 此时才能 Set;而端口冲突发生在 Start 内部(net.Listen 失败即 HTTP 服务
// 静默不启动,或测试二进制被信号打断)。SetPluginConfig 会先建表再写,
// 正好填补这个时序缺口。
//
// 用 :0 让 OS 分配空闲端口——固定端口在「并行跑测试」或「本机有 homed
// 常驻」时必然周期性失败(实测:干净树连跑 20 轮也复现 2 次)。
for _, kv := range []struct{ plugin, key string }{
{"pluginmgr", "http_addr"},
{"remotedevice", "listen_addr"},
} {
if err := cfgReg.SetPluginConfig(kv.plugin, kv.key, "127.0.0.1:0"); err != nil {
t.Fatalf("预置 %s.%s 临时端口: %v", kv.plugin, kv.key, err)
}
}
// webui 的监听地址走独立旁路(SetListenOverride 优先级高于 settings,
// 因为历史上内核在插件表建立前写 settings 会失败)。
webui.SetListenOverride("127.0.0.1:0")
t.Cleanup(func() { webui.SetListenOverride("") })
pluginReg.SetConfigRegistry(cfgReg)
plgDir := filepath.Join(tmpDir, "plugins")
os.MkdirAll(plgDir, 0755)
if err := pluginReg.Load(plgDir); err != nil {
t.Fatal(err)
}
return &testPluginEnv{
tmpDir: tmpDir,
stageHost: stageHost,
iom: iom,
pluginReg: pluginReg,
memDB: memDB,
ks: ks,
docStore: docStore,
cfgReg: cfgReg,
}
}
func (e *testPluginEnv) cleanup() {
e.pluginReg.StopAll()
e.memDB.Close()
e.ks.Stop()
e.docStore.Stop()
os.RemoveAll(e.tmpDir)
}
// ---------------------------------------------------------------------------
// Registration
// ---------------------------------------------------------------------------
func TestIntegrationAllPluginsRegister(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
expectedTools := []string{
"cmd_run",
"terminal_create", "terminal_write", "terminal_read",
"terminal_resize", "terminal_close", "terminal_list",
"healthcheck", "healthcheck_plugins", "healthcheck_tools", "healthcheck_memory", "healthcheck_perf",
"timer_set",
}
defs := env.stageHost.GetToolDefs()
registered := make(map[string]bool)
for _, def := range defs {
registered[def.Name] = true
}
var missing []string
for _, tool := range expectedTools {
if !registered[tool] {
missing = append(missing, tool)
}
}
if len(missing) > 0 {
all := make([]string, 0, len(defs))
for _, d := range defs {
all = append(all, d.Name)
}
t.Fatalf("missing tools: %v\nall registered: %v", missing, all)
}
t.Logf("all %d expected tools registered (total: %d)", len(expectedTools), len(defs))
}
// ---------------------------------------------------------------------------
// cmd_run
// ---------------------------------------------------------------------------
func TestIntegrationCmdRun(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
result, err := env.stageHost.ExecuteTool("cmd_run", map[string]interface{}{
"command": "echo hello_world",
})
if err != nil {
t.Fatal(err)
}
data, _ := json.Marshal(result)
var resp map[string]interface{}
json.Unmarshal(data, &resp)
if resp["exit_code"] != 0.0 {
t.Fatalf("expected exit_code 0, got %v", resp["exit_code"])
}
stdout, ok := resp["stdout"].(string)
if !ok || stdout != "hello_world" {
t.Fatalf("expected stdout 'hello_world', got %q", stdout)
}
t.Logf("cmd_run OK: exit_code=0 stdout=%q", stdout)
}
func TestIntegrationCmdRunWithWorkdir(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
result, err := env.stageHost.ExecuteTool("cmd_run", map[string]interface{}{
"command": "pwd",
"workdir": "/tmp",
})
if err != nil {
t.Fatal(err)
}
data, _ := json.Marshal(result)
var resp map[string]interface{}
json.Unmarshal(data, &resp)
if resp["exit_code"] != 0.0 {
t.Fatalf("expected exit_code 0, got %v", resp["exit_code"])
}
stdout := resp["stdout"].(string)
if stdout != "/tmp" {
t.Fatalf("expected stdout '/tmp', got %q", stdout)
}
t.Logf("cmd_run workdir OK: stdout=%q", stdout)
}
func TestIntegrationCmdRunInvalidTimeout(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
result, err := env.stageHost.ExecuteTool("cmd_run", map[string]interface{}{
"command": "echo ok",
"timeout": "not-a-duration",
})
if err != nil {
t.Fatal(err)
}
data, _ := json.Marshal(result)
var resp map[string]interface{}
json.Unmarshal(data, &resp)
errMsg, ok := resp["error"].(string)
if !ok || errMsg == "" {
t.Fatalf("expected error for invalid timeout, got %v", resp)
}
t.Logf("cmd_run invalid_timeout OK: error=%q", errMsg)
}
func TestIntegrationCmdRunStderr(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
result, err := env.stageHost.ExecuteTool("cmd_run", map[string]interface{}{
"command": "sh -c \"echo stderr_test >&2\"",
})
if err != nil {
t.Fatal(err)
}
data, _ := json.Marshal(result)
var resp map[string]interface{}
json.Unmarshal(data, &resp)
if resp["exit_code"] != 0.0 {
t.Fatalf("expected exit_code 0, got %v", resp["exit_code"])
}
stderr := resp["stderr"].(string)
if stderr != "stderr_test" {
t.Fatalf("expected stderr 'stderr_test', got %q", stderr)
}
t.Logf("cmd_run stderr OK: stderr=%q", stderr)
}
// ---------------------------------------------------------------------------
// Terminal (PTY)
// ---------------------------------------------------------------------------
func TestIntegrationPtyCreateListClose(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
// Create
result, err := env.stageHost.ExecuteTool("terminal_create", map[string]interface{}{
"shell": "/bin/sh",
"name": "hci_test_shell",
})
if err != nil {
t.Fatal(err)
}
data, _ := json.Marshal(result)
var resp map[string]interface{}
json.Unmarshal(data, &resp)
id, ok := resp["id"].(string)
if !ok || id == "" {
status := resp["status"]
detail := resp["detail"]
if status == "error" {
t.Skipf("PTY not available: %v", detail)
}
t.Fatalf("expected non-empty terminal id, got id=%q status=%v detail=%v", id, status, detail)
}
t.Logf("terminal_create OK: id=%s", id)
// List
result2, err := env.stageHost.ExecuteTool("terminal_list", map[string]interface{}{})
if err != nil {
t.Fatal(err)
}
data2, _ := json.Marshal(result2)
var resp2 map[string]interface{}
json.Unmarshal(data2, &resp2)
terminals, ok := resp2["terminals"].([]interface{})
if !ok {
t.Fatalf("expected terminals array, got %T", resp2["terminals"])
}
if len(terminals) < 1 {
t.Fatal("expected at least 1 terminal")
}
t.Logf("terminal_list OK: %d terminals", len(terminals))
// Close
_, err = env.stageHost.ExecuteTool("terminal_close", map[string]interface{}{
"id": id,
})
if err != nil {
t.Fatal(err)
}
t.Logf("terminal_close OK: id=%s", id)
}
func TestIntegrationPtyInteractive(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
createResult, err := env.stageHost.ExecuteTool("terminal_create", map[string]interface{}{
"shell": "/bin/sh",
"name": "hci_interactive",
})
if err != nil {
t.Fatal(err)
}
cdata, _ := json.Marshal(createResult)
var cresp map[string]interface{}
json.Unmarshal(cdata, &cresp)
id, ok := cresp["id"].(string)
if !ok || id == "" {
status := cresp["status"]
if status == "error" {
t.Skipf("PTY not available: %v", cresp["detail"])
}
t.Fatalf("expected terminal id, got %v", cresp)
}
// Write a command
_, err = env.stageHost.ExecuteTool("terminal_write", map[string]interface{}{
"id": id,
"input": "echo pty_works\n",
})
if err != nil {
t.Fatal(err)
}
time.Sleep(500 * time.Millisecond)
// Read output
readResult, err := env.stageHost.ExecuteTool("terminal_read", map[string]interface{}{
"id": id,
})
if err != nil {
t.Fatal(err)
}
rdata, _ := json.Marshal(readResult)
var rresp map[string]interface{}
json.Unmarshal(rdata, &rresp)
output, ok := rresp["output"].(string)
if !ok || output == "" {
t.Fatalf("expected output, got output=%q response=%v", output, rresp)
}
t.Logf("terminal_write+read OK: output=%q", output)
// Close
env.stageHost.ExecuteTool("terminal_close", map[string]interface{}{
"id": id,
})
}
func TestIntegrationPtyResize(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
createResult, err := env.stageHost.ExecuteTool("terminal_create", map[string]interface{}{
"shell": "/bin/sh",
"name": "hci_resize_test",
})
if err != nil {
t.Fatal(err)
}
cdata, _ := json.Marshal(createResult)
var cresp map[string]interface{}
json.Unmarshal(cdata, &cresp)
id, ok := cresp["id"].(string)
if !ok || id == "" {
status := cresp["status"]
if status == "error" {
t.Skipf("PTY not available: %v", cresp["detail"])
}
t.Fatalf("expected terminal id, got %v", cresp)
}
_, err = env.stageHost.ExecuteTool("terminal_resize", map[string]interface{}{
"id": id,
"rows": 80.0,
"cols": 200.0,
})
if err != nil {
t.Fatal(err)
}
t.Logf("terminal_resize OK: id=%s rows=80 cols=200", id)
env.stageHost.ExecuteTool("terminal_close", map[string]interface{}{
"id": id,
})
}
// ---------------------------------------------------------------------------
// Healthcheck
// ---------------------------------------------------------------------------
func TestIntegrationHealthcheck(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
result, err := env.stageHost.ExecuteTool("healthcheck", map[string]interface{}{})
if err != nil {
t.Fatal(err)
}
data, _ := json.Marshal(result)
var resp map[string]interface{}
json.Unmarshal(data, &resp)
if resp["status"] != "ok" {
t.Fatalf("expected status=ok, got %v", resp["status"])
}
checks := resp["checks"].([]interface{})
if len(checks) == 0 {
t.Fatal("expected non-empty checks array")
}
t.Logf("healthcheck OK: %d checks, status=%v", len(checks), resp["status"])
}
func TestIntegrationHealthcheckPlugins(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
result, err := env.stageHost.ExecuteTool("healthcheck_plugins", map[string]interface{}{})
if err != nil {
t.Fatal(err)
}
data, _ := json.Marshal(result)
var resp map[string]interface{}
json.Unmarshal(data, &resp)
if resp["status"] != "ok" {
t.Fatalf("expected status=ok, got %v", resp["status"])
}
t.Logf("healthcheck_plugins OK: status=%v", resp["status"])
}
func TestIntegrationHealthcheckToolsList(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
result, err := env.stageHost.ExecuteTool("healthcheck_tools", map[string]interface{}{})
if err != nil {
t.Fatal(err)
}
data, _ := json.Marshal(result)
var resp map[string]interface{}
json.Unmarshal(data, &resp)
if resp["status"] != "ok" {
t.Fatalf("expected status=ok, got %v", resp["status"])
}
t.Logf("healthcheck_tools OK: status=%v", resp["status"])
}
func TestIntegrationHealthcheckMemory(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
result, err := env.stageHost.ExecuteTool("healthcheck_memory", map[string]interface{}{})
if err != nil {
t.Fatal(err)
}
data, _ := json.Marshal(result)
var resp map[string]interface{}
json.Unmarshal(data, &resp)
if resp["status"] != "ok" {
t.Fatalf("expected status=ok, got %v", resp["status"])
}
t.Logf("healthcheck_memory OK: status=%v", resp["status"])
}
func TestIntegrationToolNotFound(t *testing.T) {
env := setupIntegration(t)
defer env.cleanup()
_, err := env.stageHost.ExecuteTool("nonexistent_tool_xyz", nil)
if err == nil {
t.Fatal("expected error for nonexistent tool")
}
t.Logf("tool_not_found OK: err=%v", err)
}
func TestIntegrationLLMDrivenDiscoveryWithRealKey(t *testing.T) {
apiKey := os.Getenv("DEEPSEEK_API_KEY")
if apiKey == "" {
t.Skip("DEEPSEEK_API_KEY not set")
}
// Lua VM + DeepSeek Provider
tmpVM := t.TempDir()
vm := luaVM.NewVM(tmpVM + "/adapters")
if err := vm.Start(); err != nil {
t.Fatal(err)
}
defer vm.Stop()
pm := agentAPI.NewProviderManager()
pm.Register("deepseek", agentAPI.NewLuaAdaptedProvider(agentAPI.BaseConfig{
Model: "deepseek-v4-flash",
BaseURL: "https://api.deepseek.com",
APIKey: apiKey,
}, vm, "deepseek", "deepseek"))
// Setup — 加载所有真实内置插件
env := setupIntegrationWithProvider(t, pm)
defer env.cleanup()
// 列出已加载的真实工具
defs := env.stageHost.GetToolDefs()
t.Logf("共有 %d 个已注册的真实工具:", len(defs))
for _, d := range defs {
t.Logf(" - %s", d.Name)
}
// 调用 healthcheck 进行全面检查(含 LLM 驱动阶段)
result, err := env.stageHost.ExecuteTool("healthcheck", map[string]interface{}{})
if err != nil {
t.Fatal(err)
}
data, _ := json.Marshal(result)
var resp map[string]interface{}
json.Unmarshal(data, &resp)
t.Logf("===== Full Healthcheck Result =====")
t.Logf("status: %v", resp["status"])
t.Logf("summary: %v", resp["summary"])
t.Logf("total: %v", resp["total"])
t.Logf("passed: %v", resp["passed"])
t.Logf("failed: %v", resp["failed"])
checks := resp["checks"].([]interface{})
for _, c := range checks {
cr := c.(map[string]interface{})
prefix := "✅"
if cr["status"] == "fail" {
prefix = "❌"
}
t.Logf(" %s %s: %s %s", prefix, cr["name"], cr["status"], cr["detail"])
}
if resp["status"] != "ok" {
t.Fatalf("expected status=ok, got %v", resp["status"])
}
// 验证 LLM 发现阶段的存在
foundDiscovery := false
for _, c := range checks {
cr := c.(map[string]interface{})
if cr["name"] == "llm_discovery" {
foundDiscovery = true
if cr["status"] != "ok" {
t.Fatalf("LLM discovery failed: status=%s detail=%s", cr["status"], cr["detail"])
}
break
}
}
if !foundDiscovery {
t.Fatal("expected llm_discovery check in results")
}
t.Logf("✅ LLM-driven tool discovery test PASSED")
}