Files
HomeAgent/internal/plugins/remotedevice/binary_test.go
JianFeeeee a02d8b00a5 feat(multimodal): 多模态工具移出内核为内置插件 + get_plugin_tools 收编为内核服务
## 背景

README 第一原则:「内核职责限定在 LLM 编排、记忆管理与知识检索;
所有 IO 能力由插件实现」,而「文件读取」正在 IO 列表里。
describe_image / ocr_image / transcribe_audio 读本地文件却住在内核进程内
——无沙箱、无能力面、无审计,正是 capability.go 那套权限梯度要防的事。

## 本次改动

### 1. 多模态工具移出内核(internal/plugins/multimodal)

- 新增内置插件,注册 describe_image / transcribe_audio / ocr_image
- 媒体来源支持 path(设备/工具回传,如 screensee 的 file)与 digest(CAS)
- 失败一律以 error 上抛,不渲染成普通文本(延续 screensee 的失败传播纪律)
- ocr_image 的描述明说「不是专用 OCR 引擎」:它只是用视觉模型读字,
  真 OCR 在 internal/nlp 的 ONNX 未接入。名字承诺大于实现会让模型误判精度。

内核侧:删除 media.go 整文件(9 函数,全部零引用)、3 个 dispatch case、
3 个工具定义。

**留在内核**:input 通路(用户直传媒体直接拼进本轮 LLM 请求)——
那是「输入即上下文」,不是工具调用。

### 2. get_plugin_tools 收编为内核完整服务

原先它只收 StageHost + IOManager 两处,而三个媒体工具、memory_*、
knowledge_*、spawn_child 等全是内核 buildToolDefs 里直接 append 的
——一条都数不到。agent 能调这些工具,却调 get_plugin_tools 看不见它们。

- 以 buildToolDefs()(发给模型的那份工具表)为单一真相源
- 新增 builtinToolOwner 显式归属表:内置工具恒为 core
- resolveToolPlugin 的下划线前缀猜测限定在「该前缀是已注册插件」时才采纳,
  避免 ocr_image→"ocr"、memory_commit→"memory" 这类幽灵归属(7 处调用点依赖它)

## 判据

新增 plugin_tools_service_test.go、multimodal_departed_test.go、
multimodal/plugin_test.go。三个结构判据经变异自证会红:

1. 提示词改回「请使用 describe_image 工具」→ 仍出现该名 ⇒ 红
   ★ 这条抓到了真问题:eventloop.go 的媒体提示词仍在叫模型调旧工具名,
   而媒体 block 其实已内联给模型,且新工具需要 path/digest(用户直传都没有)
   ⇒ 模型照做必然失败。已修:改为「已随本消息附上,请直接查看」。
2. 内核为多模态加一个 os.ReadFile → 红(比「名字没出现」更强)
3. core 反向 import 插件 → 红(迁移而非删除,且守住分层)

## 验证

- go build ./... / go vet / go test ./internal/... 全绿
- go test -race ./internal/agent/core/ ./internal/plugins/multimodal/ 无数据竞争
- 部署后:39 插件加载、0 panic/fatal、0 非空 error
- 三个工具注册归属为 plugin=multimodal(不是 inferToolPlugin 猜的 describe)
- graph.db integrity ok,blocks 2787 / relations 980 未动
- 快照 /home/newqqagent/memory/pre-multimodal-plugin-20261006-090541/
- 二进制备份 /tmp/homed.bak-mmplugin

一并包含此前已验证未提交的三处静默失效修复:
- MCP MCPCallResult.IsError 丢失(空结果伪装成功)
- DeleteEntity 谎报删除
- screensee 谎报成功 + 截图诊断(尺寸/magic/sha)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-10-06 09:07:18 +08:00

939 lines
28 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 remotedevice
import (
"bufio"
"encoding/base64"
"encoding/binary"
"encoding/json"
"math/rand"
"net"
"net/http"
"net/http/httptest"
"os"
"path/filepath"
"strings"
"sync"
"testing"
"time"
)
// ===== 测试用最小 WS 客户端(模拟 GUI 设备桥)=====
type testWSClient struct {
conn net.Conn
rw *bufio.ReadWriter
}
func dialTestWS(t *testing.T, url, token string) *testWSClient {
t.Helper()
req := "GET /api/v1/device/ws?token=" + token + " HTTP/1.1\r\n" +
"Host: test\r\nUpgrade: websocket\r\nConnection: Upgrade\r\n" +
"Sec-WebSocket-Key: dGhlIHNhbXBsZSBub25jZQ==\r\nSec-WebSocket-Version: 13\r\n\r\n"
conn, err := net.Dial("tcp", strings.TrimPrefix(url, "http://"))
if err != nil {
t.Fatalf("dial: %v", err)
}
if _, err := conn.Write([]byte(req)); err != nil {
t.Fatalf("write upgrade: %v", err)
}
br := bufio.NewReader(conn)
resp, err := readLine(br)
if err != nil {
t.Fatalf("read upgrade resp: %v", err)
}
if !strings.Contains(resp, "101") {
t.Fatalf("expected 101, got %s", resp)
}
for {
line, err := readLine(br)
if err != nil {
t.Fatalf("read headers: %v", err)
}
if line == "\r\n" || line == "" {
break
}
}
return &testWSClient{conn: conn, rw: &bufio.ReadWriter{Reader: br, Writer: bufio.NewWriter(conn)}}
}
func readLine(br *bufio.Reader) (string, error) {
var sb strings.Builder
for {
b, err := br.ReadByte()
if err != nil {
return sb.String(), err
}
sb.WriteByte(b)
if b == '\n' {
return sb.String(), nil
}
}
}
// sendText 发送客户端文本帧(带掩码,RFC6455 要求客户端帧必须掩码)
func (c *testWSClient) sendText(payload []byte) {
c.sendFrame(0x1, payload)
}
func (c *testWSClient) sendBinary(payload []byte) {
c.sendFrame(0x2, payload)
}
func (c *testWSClient) sendFrame(opcode byte, payload []byte) {
maskKey := make([]byte, 4)
rand.Read(maskKey)
masked := make([]byte, len(payload))
for i := range payload {
masked[i] = payload[i] ^ maskKey[i%4]
}
var hdr []byte
hdr = append(hdr, 0x80|opcode)
n := len(payload)
switch {
case n < 126:
hdr = append(hdr, 0x80|byte(n))
case n <= 0xffff:
hdr = append(hdr, 0x80|126)
ext := make([]byte, 2)
binary.BigEndian.PutUint16(ext, uint16(n))
hdr = append(hdr, ext...)
default:
hdr = append(hdr, 0x80|127)
ext := make([]byte, 8)
binary.BigEndian.PutUint64(ext, uint64(n))
hdr = append(hdr, ext...)
}
c.rw.Write(hdr)
c.rw.Write(maskKey)
c.rw.Write(masked)
c.rw.Flush()
}
// readMsg 读一帧(跳过 pong),返回 opcode 与 payload
func (c *testWSClient) readMsg() (byte, []byte, error) {
for {
payload, isClose, opcode, err := readFrame(c.rw.Reader)
if err != nil || isClose {
return 0, nil, err
}
if opcode == 0xa {
continue
}
return opcode, payload, nil
}
}
func (c *testWSClient) close() { c.conn.Close() }
func (c *testWSClient) bindDevice(t *testing.T, deviceID, token string) {
t.Helper()
c.sendText(mustJSON(map[string]interface{}{
"op": "bind", "device_id": deviceID, "token": token,
}))
op, payload, err := c.readMsg()
if err != nil {
t.Fatalf("read bind_ack: %v", err)
}
if op != 0x1 {
t.Fatalf("expected bind_ack text frame, got %x", op)
}
var ack map[string]interface{}
if err := json.Unmarshal(payload, &ack); err != nil {
t.Fatalf("decode bind_ack: %v", err)
}
if ack["op"] != "bind_ack" || ack["ok"] != true {
t.Fatalf("bind rejected: %v", ack)
}
}
func (c *testWSClient) readHelloAck(t *testing.T) string {
t.Helper()
op, payload, err := c.readMsg()
if err != nil {
t.Fatalf("read hello_ack: %v", err)
}
if op != 0x1 {
t.Fatalf("expected hello_ack text frame, got %x", op)
}
var ack map[string]interface{}
if err := json.Unmarshal(payload, &ack); err != nil {
t.Fatalf("decode hello_ack: %v", err)
}
deviceID, _ := ack["device"].(string)
if ack["op"] != "hello_ack" || deviceID == "" {
t.Fatalf("expected hello_ack with device id, got %v", ack)
}
return deviceID
}
func (c *testWSClient) readHelloAckAndBind(t *testing.T, token string) {
t.Helper()
deviceID := c.readHelloAck(t)
c.bindDevice(t, deviceID, token)
}
// ===== 端到端:hello/bind/cmd + 二进制分块回传(录像协议)=====
func TestWSBinaryChunkUpload(t *testing.T) {
reg := NewRegistry()
token := "test-token-123"
reg.SetAcceptToken(func(provided string) bool { return provided == token })
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
url := srv.URL
cli := dialTestWS(t, url, token)
defer cli.close()
// hello 后必须完成 bind,设备才会注册并开始处理数据。
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"gui-test","name":"测试机","kind":"computer","caps":["cmd"]}}`))
cli.readHelloAckAndBind(t, token)
// 模拟设备收到 cmd 后以二进制分块回传(cmd_data_start → 0x2×N → cmd_data_end)
videoData := make([]byte, 20000) // 跨多个 8KB 块
for i := range videoData {
videoData[i] = byte(i % 251)
}
go func() {
time.Sleep(100 * time.Millisecond)
cli.sendText(mustJSON(map[string]interface{}{
"op": "cmd_data_start", "req_id": "req-video-1",
"kind": "camera_video", "mime": "video/mp4",
"total": len(videoData), "chunk_size": 8192,
}))
const chunk = 8192
for off := 0; off < len(videoData); off += chunk {
end := off + chunk
if end > len(videoData) {
end = len(videoData)
}
cli.sendBinary(videoData[off:end])
}
cli.sendText(mustJSON(map[string]interface{}{
"op": "cmd_data_end", "req_id": "req-video-1", "status": "ok", "total": len(videoData),
}))
}()
// 服务端等待聚合结果(AwaitResult 由 deliverResult 唤醒)
res, err := reg.AwaitResult("req-video-1", 5*time.Second)
if err != nil {
t.Fatalf("await aggregated result: %v", err)
}
if res["status"] != "ok" {
t.Fatalf("expected status ok, got %v", res["status"])
}
if got, _ := res["size"].(int); got != len(videoData) {
t.Fatalf("size mismatch: got %v want %d", res["size"], len(videoData))
}
// 校验 base64 数据完整性
stored, ok := reg.GetResult("req-video-1")
if !ok {
t.Fatal("result not persisted")
}
b64, _ := stored["data_base64"].(string)
if len(b64) == 0 {
t.Fatal("data_base64 empty")
}
decoded, err := base64.StdEncoding.DecodeString(b64)
if err != nil {
t.Fatalf("decode base64: %v", err)
}
if string(decoded) != string(videoData) {
t.Fatal("decoded data mismatch")
}
}
// ===== 媒体落盘:SetMediaDir 后 cmd_result 返回 file 路径而非 base64 内联 =====
func TestWSBinaryMediaToFile(t *testing.T) {
reg := NewRegistry()
token := "test-token-123"
reg.SetAcceptToken(func(provided string) bool { return provided == token })
mediaDir := t.TempDir()
reg.SetMediaDir(mediaDir)
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
cli := dialTestWS(t, srv.URL, token)
defer cli.close()
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"gui-media","name":"媒体机","kind":"computer","caps":["cmd"]}}`))
cli.readHelloAckAndBind(t, token)
videoData := make([]byte, 30000)
for i := range videoData {
videoData[i] = byte(i % 253)
}
go func() {
time.Sleep(50 * time.Millisecond)
cli.sendText(mustJSON(map[string]interface{}{
"op": "cmd_data_start", "req_id": "req-file-1",
"kind": "camera_video", "mime": "video/mp4",
"total": len(videoData),
}))
const chunk = 8192
for off := 0; off < len(videoData); off += chunk {
end := off + chunk
if end > len(videoData) {
end = len(videoData)
}
cli.sendBinary(videoData[off:end])
}
cli.sendText(mustJSON(map[string]interface{}{
"op": "cmd_data_end", "req_id": "req-file-1", "status": "ok",
}))
}()
res, err := reg.AwaitResult("req-file-1", 5*time.Second)
if err != nil {
t.Fatalf("await result: %v", err)
}
// 落盘模式:file 字段存在且内容一致;不应再有 data_base64
fp, ok := res["file"].(string)
if !ok || fp == "" {
t.Fatalf("expected file path in result, got %v", res)
}
if _, hasB64 := res["data_base64"]; hasB64 {
t.Fatal("data_base64 should be absent in file mode")
}
if want := filepath.Join(mediaDir, "req-file-1.mp4"); fp != want {
t.Fatalf("file path = %s, want %s", fp, want)
}
got, err := os.ReadFile(fp)
if err != nil {
t.Fatalf("read media file: %v", err)
}
if string(got) != string(videoData) {
t.Fatal("media file content mismatch")
}
}
// ===== 端到端:PushData 下发音频(网关→设备 cmd_speech 协议)=====
func TestWSPushDataAudio(t *testing.T) {
reg := NewRegistry()
token := "test-token-456"
reg.SetAcceptToken(func(provided string) bool { return provided == token })
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
cli := dialTestWS(t, srv.URL, token)
defer cli.close()
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"audio-dev","name":"音频机","kind":"speaker"}}`))
cli.readHelloAckAndBind(t, token)
audioData := []byte("RIFF....fake-wav-audio-data-for-testing....")
// 异步下发音频
errCh := make(chan error, 1)
go func() {
errCh <- reg.PushData("audio-dev", "req-speech-1", "speech", "audio/wav", audioData)
}()
// 设备侧按协议读取:start 文本帧 → N 个二进制帧 → end 文本帧
var start map[string]interface{}
var chunks [][]byte
var end map[string]interface{}
deadline := time.After(5 * time.Second)
for end == nil {
select {
case <-deadline:
t.Fatal("timeout reading speech protocol frames")
default:
}
op, payload, err := cli.readMsg()
if err != nil {
t.Fatalf("read frame: %v", err)
}
switch op {
case 0x1:
var msg map[string]interface{}
json.Unmarshal(payload, &msg)
switch msg["op"] {
case "cmd_speech_start":
start = msg
case "cmd_speech_end":
end = msg
}
case 0x2:
chunks = append(chunks, payload)
}
}
if err := <-errCh; err != nil {
t.Fatalf("PushData error: %v", err)
}
if start == nil || start["op"] != "cmd_speech_start" {
t.Fatal("missing cmd_speech_start")
}
if start["mime"] != "audio/wav" || start["kind"] != "speech" {
t.Fatalf("unexpected start fields: %v", start)
}
if int(start["total"].(float64)) != len(audioData) {
t.Fatalf("total mismatch: %v", start["total"])
}
if end["req_id"] != "req-speech-1" {
t.Fatalf("unexpected end: %v", end)
}
var got []byte
for _, c := range chunks {
got = append(got, c...)
}
if string(got) != string(audioData) {
t.Fatalf("audio data mismatch: got %d bytes want %d", len(got), len(audioData))
}
}
// ===== PushData 对离线设备报错 =====
func TestPushDataOfflineDevice(t *testing.T) {
reg := NewRegistry()
err := reg.PushData("no-such-device", "req-x", "speech", "audio/wav", []byte{1})
if err == nil || !strings.Contains(err.Error(), "not online") {
t.Fatalf("expected not online error, got %v", err)
}
}
// ===== screensee:截屏回传 + 视觉描述回调 =====
func TestScreenseeEndToEnd(t *testing.T) {
reg := NewRegistry()
token := "test-token-see"
reg.SetAcceptToken(func(provided string) bool { return provided == token })
dev := &devicectlDevice{reg: reg}
var gotDataURL string
dev.SetSeeHandler(func(dataURL string, provider string) (string, error) {
gotDataURL = dataURL
return "屏幕上显示的是测试画面", nil
})
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
cli := dialTestWS(t, srv.URL, token)
defer cli.close()
// 设备 hello + bind(bind 需 token 才能被授权流程识别,这里直接手动授权)
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"see-dev","name":"屏幕机","kind":"computer","caps":["cmd"]}}`))
cli.readHelloAckAndBind(t, token)
// 设备侧循环收命令并回执(模拟 GUI screensee 实现)
go func() {
for {
op, payload, err := cli.readMsg()
if err != nil {
return
}
if op != 0x1 {
continue
}
var msg map[string]interface{}
if json.Unmarshal(payload, &msg) != nil {
continue
}
if msg["op"] == "cmd" && msg["command"] == "screensee" {
reqID, _ := msg["req_id"].(string)
fakeJPEG := []byte{0xFF, 0xD8, 0xFF, 0xE0, 0x00, 0x10} // JPEG magic
b64 := base64.StdEncoding.EncodeToString(fakeJPEG)
cli.sendText(mustJSON(map[string]interface{}{
"op": "cmd_result", "req_id": reqID, "device_id": "see-dev",
"status": "ok", "output": "data:image/jpeg;base64," + b64,
}))
}
}
}()
// agent 调用 screensee
res, err := dev.Execute("screensee", map[string]interface{}{"device_id": "see-dev"})
if err != nil {
t.Fatalf("screensee: %v", err)
}
m := res.(map[string]interface{})
if m["description"] != "屏幕上显示的是测试画面" {
t.Fatalf("unexpected description: %v", m["description"])
}
if !strings.HasPrefix(gotDataURL, "data:image/jpeg;base64,") {
t.Fatalf("handler received bad dataURL: %s", gotDataURL)
}
// 客户端鉴权模式:服务端不拦截,总是转发(设备端自行决定是否执行)。
// see-dev 未声明 screensee 之外的问题,此处仅验证服务端不再因授权状态报错。
if _, err := dev.Execute("screensee", map[string]interface{}{"device_id": "see-dev"}); err != nil {
t.Fatalf("server should forward regardless of authorization, got: %v", err)
}
}
// ===== computeruse:鼠标/键盘控制命令下发 =====
func TestComputeruseEndToEnd(t *testing.T) {
reg := NewRegistry()
token := "test-token-cu"
reg.SetAcceptToken(func(provided string) bool { return provided == token })
dev := &devicectlDevice{reg: reg}
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
cli := dialTestWS(t, srv.URL, token)
defer cli.close()
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"cu-dev","name":"操控机","kind":"computer","caps":["cmd","computeruse"]}}`))
cli.readHelloAckAndBind(t, token)
// 设备侧收 computeruse 命令并回执
var receivedCmd string
done := make(chan struct{})
go func() {
for {
op, payload, err := cli.readMsg()
if err != nil {
return
}
if op != 0x1 {
continue
}
var msg map[string]interface{}
if json.Unmarshal(payload, &msg) != nil {
continue
}
if msg["op"] == "cmd" && msg["cmd_type"] == "homeagent" {
receivedCmd, _ = msg["command"].(string)
reqID, _ := msg["req_id"].(string)
if strings.HasPrefix(receivedCmd, "computeruse ") {
cli.sendText(mustJSON(map[string]interface{}{
"op": "cmd_result", "req_id": reqID, "device_id": "cu-dev",
"status": "ok", "output": "clicked at (3009,450)",
}))
close(done)
return
}
}
}
}()
// agent 调用 computeruse(结构化参数)
res, err := dev.Execute("computeruse", map[string]interface{}{
"device_id": "cu-dev", "action": "click", "x": float64(3009), "y": float64(450),
})
if err != nil {
t.Fatalf("computeruse: %v", err)
}
select {
case <-done:
case <-time.After(3 * time.Second):
t.Fatal("device did not receive command")
}
// 验证下发的命令是合法 JSON 参数格式
payloadJSON := strings.TrimPrefix(receivedCmd, "computeruse ")
var params map[string]interface{}
if err := json.Unmarshal([]byte(payloadJSON), &params); err != nil {
t.Fatalf("command payload not valid JSON: %v (%s)", err, payloadJSON)
}
if params["action"] != "click" || params["x"] != float64(3009) || params["y"] != float64(450) {
t.Fatalf("unexpected params: %v", params)
}
if m := res.(map[string]interface{}); m["status"] != "ok" {
t.Fatalf("expected ok result: %v", m)
}
// 缺坐标应报错
if _, err := dev.Execute("computeruse", map[string]interface{}{"device_id": "cu-dev", "action": "click"}); err == nil {
t.Fatal("click without x/y should error")
}
// 未知 action 应报错
if _, err := dev.Execute("computeruse", map[string]interface{}{"device_id": "cu-dev", "action": "fly"}); err == nil {
t.Fatal("unknown action should error")
}
// type 需要 text
if _, err := dev.Execute("computeruse", map[string]interface{}{"device_id": "cu-dev", "action": "type"}); err == nil {
t.Fatal("type without text should error")
}
}
// ===== clipboardsee / clipboardsue:剪切板读写 =====
func TestClipboardEndToEnd(t *testing.T) {
reg := NewRegistry()
token := "test-token-clip"
reg.SetAcceptToken(func(provided string) bool { return provided == token })
dev := &devicectlDevice{reg: reg}
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
cli := dialTestWS(t, srv.URL, token)
defer cli.close()
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"clip-dev","name":"剪贴板机","kind":"computer","caps":["cmd"]}}`))
cli.readHelloAckAndBind(t, token)
// 设备侧响应剪贴板命令
go func() {
for {
op, payload, err := cli.readMsg()
if err != nil {
return
}
if op != 0x1 {
continue
}
var msg map[string]interface{}
if json.Unmarshal(payload, &msg) != nil {
continue
}
if msg["op"] != "cmd" || msg["cmd_type"] != "homeagent" {
continue
}
cmd, _ := msg["command"].(string)
reqID, _ := msg["req_id"].(string)
switch {
case cmd == "clipboardsee":
cli.sendText(mustJSON(map[string]interface{}{
"op": "cmd_result", "req_id": reqID, "device_id": "clip-dev",
"status": "ok", "output": "https://example.com/copied-link",
}))
case strings.HasPrefix(cmd, "clipboardsue "):
written := strings.TrimPrefix(cmd, "clipboardsue ")
cli.sendText(mustJSON(map[string]interface{}{
"op": "cmd_result", "req_id": reqID, "device_id": "clip-dev",
"status": "ok", "output": "clipboard set: " + written,
}))
}
}
}()
t.Run("clipboardsee_returns_content", func(t *testing.T) {
res, err := dev.Execute("clipboardsee", map[string]interface{}{"device_id": "clip-dev"})
if err != nil {
t.Fatalf("clipboardsee: %v", err)
}
m := res.(map[string]interface{})
if m["content"] != "https://example.com/copied-link" {
t.Fatalf("unexpected content: %v", m["content"])
}
if m["empty"] == true {
t.Fatal("content should not be empty")
}
})
t.Run("clipboardsue_writes_text", func(t *testing.T) {
long := strings.Repeat("你好", 100) // 200 runes,验证 preview 截断
res, err := dev.Execute("clipboardsue", map[string]interface{}{"device_id": "clip-dev", "text": long})
if err != nil {
t.Fatalf("clipboardsue: %v", err)
}
m := res.(map[string]interface{})
if m["written"] != len(long) {
t.Fatalf("written mismatch: %v", m["written"])
}
preview, _ := m["preview"].(string)
if !strings.HasSuffix(preview, "...") || len([]rune(preview)) > 64 {
t.Fatalf("preview should be truncated: %q", preview)
}
})
t.Run("clipboardsue_requires_text", func(t *testing.T) {
if _, err := dev.Execute("clipboardsue", map[string]interface{}{"device_id": "clip-dev"}); err == nil {
t.Fatal("missing text should error")
}
})
t.Run("server_forwards_regardless_of_authorization", func(t *testing.T) {
// 客户端鉴权模式:服务端不再拦截未授权设备,由设备端自行拒绝。
// 这里验证服务端能正常查询设备(不因授权状态报错)。
if _, ok := reg.Get("clip-dev"); !ok {
t.Fatal("device should be registered")
}
})
}
// ===== 能力矩阵:caps 声明 → 工具可用性 =====
func TestCapabilityMatrix(t *testing.T) {
cases := []struct {
name string
caps []string
tool string
expected bool
}{
{"摄像头只声明camera不能screensee", []string{"camera"}, "screensee", false},
{"摄像头只声明camera可以camerasue", []string{"camera"}, "camerasue", true},
{"屏幕设备支持screensue+screensee", []string{"screen"}, "screensee", true},
{"clipboard能力含读写", []string{"clipboard"}, "clipboardsue", true},
{"精确声明computeruse", []string{"computeruse"}, "computeruse", true},
{"历史cmd视为全能力", []string{"status", "cmdrun", "deviceinfo"}, "screensee", true},
{"无任何已知能力视为全兼容", []string{}, "computeruse", true},
{"混合:有已知能力则严格匹配", []string{"camera", "screen"}, "computeruse", false},
}
for _, tc := range cases {
t.Run(tc.name, func(t *testing.T) {
if got := deviceSupportsTool(tc.caps, tc.tool); got != tc.expected {
t.Fatalf("deviceSupportsTool(%v, %s) = %v, want %v", tc.caps, tc.tool, got, tc.expected)
}
})
}
// 端到端:声明 camera 的设备调 screensee 应被拒绝
reg := NewRegistry()
token := "test-cap-token"
reg.SetAcceptToken(func(provided string) bool { return provided == token })
dev := &devicectlDevice{reg: reg}
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
cli := dialTestWS(t, srv.URL, token)
defer cli.close()
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"cam-only","name":"纯摄像头","kind":"camera","caps":["camera"]}}`))
cli.readHelloAckAndBind(t, token)
if _, err := dev.Execute("screensee", map[string]interface{}{"device_id": "cam-only"}); err == nil {
t.Fatal("camera-only device should not support screensee")
} else if !strings.Contains(err.Error(), "未声明 screensee 能力") {
t.Fatalf("unexpected error: %v", err)
}
}
// ===== 设备主动上报事件 → 事件回调 =====
func TestDeviceEventReport(t *testing.T) {
reg := NewRegistry()
token := "test-evt-token"
reg.SetAcceptToken(func(provided string) bool { return provided == token })
var events []map[string]interface{}
var evtMu sync.Mutex
reg.SetEventHandler(func(deviceID string, msg map[string]interface{}) {
evtMu.Lock()
events = append(events, msg)
evtMu.Unlock()
})
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
cli := dialTestWS(t, srv.URL, token)
defer cli.close()
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"cam-watch","name":"监控摄像头","kind":"camera","caps":["camera"]}}`))
cli.readHelloAckAndBind(t, token)
// 设备主动上报:识别到未知人员驻留
cli.sendText(mustJSON(map[string]interface{}{
"op": "event", "device_id": "cam-watch",
"type": "unknown_person_detected",
"detail": "后门区域检测到陌生面孔,驻留超过30秒",
}))
// 不带 device_id 时应回退到当前连接的设备
cli.sendText(mustJSON(map[string]interface{}{
"op": "event",
"type": "motion",
}))
deadline := time.After(3 * time.Second)
for {
evtMu.Lock()
n := len(events)
evtMu.Unlock()
if n >= 2 {
break
}
select {
case <-deadline:
t.Fatalf("expected 2 events, got %d", n)
default:
time.Sleep(20 * time.Millisecond)
}
}
evtMu.Lock()
defer evtMu.Unlock()
if events[0]["type"] != "unknown_person_detected" {
t.Fatalf("unexpected first event: %v", events[0])
}
if events[1]["type"] != "motion" {
t.Fatalf("unexpected second event: %v", events[1])
}
}
func TestWSDoesNotExposeDeviceBeforeBind(t *testing.T) {
reg := NewRegistry()
token := "prebind-token"
reg.SetAcceptToken(func(provided string) bool { return provided == token })
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
cli := dialTestWS(t, srv.URL, token)
defer cli.close()
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"prebind-dev","name":"待绑定设备","kind":"computer","caps":["cmd"]}}`))
deviceID := cli.readHelloAck(t)
if deviceID != "prebind-dev" {
t.Fatalf("unexpected device id: %s", deviceID)
}
if _, ok := reg.Get(deviceID); ok {
t.Fatal("device must not be registered before bind")
}
if reg.Online(deviceID) {
t.Fatal("device must not be online before bind")
}
if err := reg.PushJSON(deviceID, map[string]interface{}{"op": "cmd"}); err == nil {
t.Fatal("command push must fail before bind")
}
cli.sendText(mustJSON(map[string]interface{}{
"op": "cmd_result", "req_id": "prebind-result", "status": "ok",
}))
time.Sleep(20 * time.Millisecond)
if _, ok := reg.GetResult("prebind-result"); ok {
t.Fatal("result must be ignored before bind")
}
}
func TestWSRejectedBindDoesNotRegister(t *testing.T) {
reg := NewRegistry()
reg.SetAcceptToken(func(provided string) bool { return provided == "expected-token" })
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
cli := dialTestWS(t, srv.URL, "")
defer cli.close()
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"rejected-dev","name":"拒绝设备","kind":"computer"}}`))
deviceID := cli.readHelloAck(t)
cli.sendText(mustJSON(map[string]interface{}{
"op": "bind", "device_id": deviceID, "token": "wrong-token",
}))
op, payload, err := cli.readMsg()
if err != nil {
t.Fatalf("read rejected bind_ack: %v", err)
}
if op != 0x1 {
t.Fatalf("expected rejected bind_ack text frame, got %x", op)
}
var ack map[string]interface{}
if err := json.Unmarshal(payload, &ack); err != nil {
t.Fatalf("decode rejected bind_ack: %v", err)
}
if ack["op"] != "bind_ack" || ack["ok"] != false {
t.Fatalf("expected rejected bind_ack, got %v", ack)
}
if _, ok := reg.Get(deviceID); ok {
t.Fatal("rejected device must not be registered")
}
if reg.Online(deviceID) {
t.Fatal("rejected device must not be online")
}
}
func TestWSHandshakeAuthorizationAllowsUnrelatedBindToken(t *testing.T) {
reg := NewRegistry()
transportToken := "transport-token"
reg.SetAcceptToken(func(provided string) bool { return provided == transportToken })
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
cli := dialTestWS(t, srv.URL, transportToken)
defer cli.close()
cli.sendText([]byte(`{"op":"hello","device":{"device_id":"transport-dev","name":"代理设备","kind":"computer"}}`))
deviceID := cli.readHelloAck(t)
cli.bindDevice(t, deviceID, "unrelated-body-token")
if !reg.Online(deviceID) {
t.Fatal("handshake-authorized device should be online after bind")
}
}
func TestAwaitResultReturnsResultDeliveredBeforeWaiter(t *testing.T) {
reg := NewRegistry()
want := map[string]interface{}{"status": "ok", "value": "early"}
reg.deliverResult("early-result", want)
got, err := reg.AwaitResult("early-result", 50*time.Millisecond)
if err != nil {
t.Fatalf("await early result: %v", err)
}
if got["status"] != want["status"] || got["value"] != want["value"] {
t.Fatalf("unexpected early result: %v", got)
}
}
// ===== ping 帧处理:不得因未 bind 而断连 =====
// ★ 未 bind(或已 bind)时收到 ping,服务端必须回 pong,**不得关闭连接**。
//
// 真实 bug(生产日志实证):原实现用 wsWriteLocked(curID, writePong),而
// conns[curID] 只在 bind 成功后才写入 ⇒ 握手后、bind 前到来的 ping 找不到
// 写锁入口,函数返回错误,读循环直接 return 关连接。
//
// 后果:客户端每 30s ping 一次,只要有一次落在未 bind 窗口就断连;生产日志里
// 同一设备 20 秒内多次 "ws connected" 且 online/offline 反复交替,正是这个。
//
// 判据直打 bug 点:只握手、**不发 hello/bind**,发 ping,要求收到 pong。
// readMsg 会跳过 pong,所以这里直接读原始帧。
func TestPingBeforeBindDoesNotDropConnection(t *testing.T) {
reg := NewRegistry()
token := "tok-ping-bind"
reg.SetAcceptToken(func(s string) bool { return s == token })
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
c := dialTestWS(t, srv.URL, token)
defer c.close()
// 只握手,不发 hello/bind —— 模拟「尚未绑定完成」的窗口
c.sendFrame(0x9, nil) // ping
// 必须收到 pong;EOF/错误说明服务端在 ping 路径上断了连接
c.conn.SetReadDeadline(time.Now().Add(3 * time.Second))
_, _, opcode, err := readFrame(c.rw.Reader)
if err != nil {
t.Fatalf("未 bind 时 ping 导致连接不可用(服务端 bug): %v", err)
}
if opcode != 0xA {
t.Fatalf("期望 pong(0xA),收到 opcode=%#x", opcode)
}
}
// 已 bind 的设备发 ping 同样必须得到 pong,且连接与在线状态都保持。
func TestPingAfterBindGetsPong(t *testing.T) {
reg := NewRegistry()
token := "tok-ping-bound"
reg.SetAcceptToken(func(s string) bool { return s == token })
srv := httptest.NewServer(http.HandlerFunc(reg.ServeWS))
defer srv.Close()
c := dialTestWS(t, srv.URL, token)
defer c.close()
c.sendText(mustJSON(map[string]interface{}{
"op": "hello",
"device": map[string]interface{}{
"device_id": "d-ping", "name": "d", "kind": "computer", "caps": []string{"status"},
},
}))
c.readHelloAck(t)
c.bindDevice(t, "d-ping", token)
if !reg.Online("d-ping") {
t.Fatal("bind 后设备应在线")
}
c.sendFrame(0x9, nil) // ping
c.conn.SetReadDeadline(time.Now().Add(3 * time.Second))
_, _, opcode, err := readFrame(c.rw.Reader)
if err != nil {
t.Fatalf("已 bind 设备 ping 失败: %v", err)
}
if opcode != 0xA {
t.Fatalf("期望 pong(0xA),收到 %#x", opcode)
}
if !reg.Online("d-ping") {
t.Error("ping 之后设备不应掉线")
}
}