71f0a1abdb
- 所有Go模块路径从 github.com/yourname/cyrene-ai 迁移到 git.yeij.top/AskaEth/Cyrene - 5个Go Dockerfile添加 GOPROXY=https://goproxy.cn,direct 解决国内构建问题 - ai-core go.mod 添加 pkg/plugins replace 指令 - Caddyfile 简化为 http:// 通配 + handle 保留 /api 前缀 - ethend Dockerfile 适配 (npm install + 仅 COPY package.json) - ethend 新增 RUNNING_IN_DOCKER 环境变量,健康检查改用Docker服务名 - ethend 数据库状态检查支持Docker hostname (postgres/redis/qdrant/minio) - process-manager 新增 CONTAINER_SVC_MAP + Docker模式自动检测 - 统一 docker-compose.dev.db.yml 卷名 (pg_data/redis_data/qdrant_data/minio_data) - docker-compose.yml ethend服务挂载docker.sock + 端口变量化 - 清理 .env 统一后的残留文件与提示信息 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
159 lines
4.1 KiB
Go
159 lines
4.1 KiB
Go
package handler
|
|
|
|
import (
|
|
"encoding/json"
|
|
"net/http"
|
|
"sync"
|
|
"time"
|
|
|
|
"github.com/gorilla/websocket"
|
|
|
|
"git.yeij.top/AskaEth/Cyrene/pkg/logger"
|
|
"git.yeij.top/AskaEth/Cyrene/voice-service/internal/service"
|
|
)
|
|
|
|
var upgrader = websocket.Upgrader{
|
|
ReadBufferSize: 4096,
|
|
WriteBufferSize: 4096,
|
|
CheckOrigin: func(r *http.Request) bool {
|
|
return true
|
|
},
|
|
}
|
|
|
|
// StreamingSTTHandler 处理实时语音识别 WebSocket 连接。
|
|
// 客户端通过 WebSocket 流式发送音频二进制帧,服务端通过一条持久的
|
|
// DashScope WebSocket 连接转发音频并持续返回识别结果。
|
|
type StreamingSTTHandler struct {
|
|
svc *service.STTService
|
|
}
|
|
|
|
// NewStreamingSTTHandler 创建流式 STT 处理器。
|
|
func NewStreamingSTTHandler(svc *service.STTService) *StreamingSTTHandler {
|
|
return &StreamingSTTHandler{svc: svc}
|
|
}
|
|
|
|
// HandleStreamingSTT 处理 WebSocket 升级和实时 STT 会话。
|
|
// GET /api/v1/stt/stream
|
|
func (h *StreamingSTTHandler) HandleStreamingSTT(w http.ResponseWriter, r *http.Request) {
|
|
if !h.svc.IsAvailable() {
|
|
writeJSON(w, http.StatusServiceUnavailable, map[string]interface{}{
|
|
"error": "STT 引擎不可用",
|
|
})
|
|
return
|
|
}
|
|
|
|
language := r.URL.Query().Get("language")
|
|
if language == "" {
|
|
language = "zh"
|
|
}
|
|
format := r.URL.Query().Get("format")
|
|
if format == "" {
|
|
format = "pcm"
|
|
}
|
|
|
|
conn, err := upgrader.Upgrade(w, r, nil)
|
|
if err != nil {
|
|
logger.Printf("[stream-stt] WebSocket 升级失败: %v", err)
|
|
return
|
|
}
|
|
defer conn.Close()
|
|
|
|
logger.Printf("[stream-stt] 客户端已连接, format=%s, language=%s", format, language)
|
|
|
|
// 创建持久的 DashScope 流式会话
|
|
session, err := h.svc.StartStreaming(format, language)
|
|
if err != nil {
|
|
logger.Printf("[stream-stt] 创建 DashScope 会话失败: %v", err)
|
|
conn.WriteJSON(map[string]interface{}{
|
|
"type": "error",
|
|
"error": "启动语音识别失败: " + err.Error(),
|
|
})
|
|
return
|
|
}
|
|
defer session.Close()
|
|
|
|
var mu sync.Mutex
|
|
conn.SetReadDeadline(time.Now().Add(300 * time.Second)) // 5 分钟超时
|
|
|
|
// goroutine: 读取 DashScope 结果并推送到客户端
|
|
resultDone := make(chan struct{})
|
|
go func() {
|
|
defer close(resultDone)
|
|
for result := range session.Results() {
|
|
mu.Lock()
|
|
if result.Error != "" {
|
|
logger.Printf("[stream-stt] DashScope 错误: %s", result.Error)
|
|
conn.WriteJSON(map[string]interface{}{
|
|
"type": "error",
|
|
"error": result.Error,
|
|
})
|
|
} else if result.Text != "" {
|
|
conn.WriteJSON(map[string]interface{}{
|
|
"type": "result",
|
|
"text": result.Text,
|
|
"isFinal": result.IsFinal,
|
|
})
|
|
}
|
|
mu.Unlock()
|
|
}
|
|
}()
|
|
|
|
// 主循环: 读取客户端音频帧
|
|
for {
|
|
msgType, data, err := conn.ReadMessage()
|
|
if err != nil {
|
|
if websocket.IsUnexpectedCloseError(err, websocket.CloseGoingAway, websocket.CloseNormalClosure) {
|
|
logger.Printf("[stream-stt] 读取错误: %v", err)
|
|
}
|
|
break
|
|
}
|
|
|
|
// 文本控制消息
|
|
if msgType == websocket.TextMessage {
|
|
var ctrl map[string]interface{}
|
|
if json.Unmarshal(data, &ctrl) == nil {
|
|
action, _ := ctrl["action"].(string)
|
|
if action == "stop" {
|
|
logger.Printf("[stream-stt] 客户端请求停止")
|
|
mu.Lock()
|
|
conn.WriteJSON(map[string]interface{}{
|
|
"type": "done",
|
|
"action": "stop",
|
|
})
|
|
mu.Unlock()
|
|
break
|
|
}
|
|
// 支持动态切换语言
|
|
if lang, ok := ctrl["language"].(string); ok && lang != "" {
|
|
language = lang
|
|
logger.Printf("[stream-stt] 切换语言: %s", lang)
|
|
}
|
|
}
|
|
continue
|
|
}
|
|
|
|
// 二进制音频帧: 发送到 DashScope
|
|
if msgType == websocket.BinaryMessage && len(data) > 0 {
|
|
if err := session.SendAudio(data); err != nil {
|
|
logger.Printf("[stream-stt] 发送音频帧失败: %v", err)
|
|
mu.Lock()
|
|
conn.WriteJSON(map[string]interface{}{
|
|
"type": "error",
|
|
"error": "发送音频失败: " + err.Error(),
|
|
})
|
|
mu.Unlock()
|
|
break
|
|
}
|
|
}
|
|
}
|
|
|
|
// 等待结果推送完成
|
|
<-resultDone
|
|
logger.Printf("[stream-stt] 会话结束")
|
|
}
|
|
|
|
// RegisterStreamingRoutes 注册流式 STT 路由。
|
|
func (h *StreamingSTTHandler) RegisterStreamingRoutes(mux *http.ServeMux) {
|
|
mux.HandleFunc("/api/v1/stt/stream", h.HandleStreamingSTT)
|
|
}
|