feat: 初始化 AI Gateway 项目

实现支持 OpenAI 和 Anthropic 双协议的统一大模型 API 网关 MVP 版本，包含：
- OpenAI 和 Anthropic 协议代理
- 供应商和模型管理
- 用量统计
- 前端配置界面

backend/internal/handler/openai_handler.go

@@ -0,0 +1,167 @@
+package handler
+
+import (
+	"bufio"
+	"fmt"
+	"net/http"
+
+	"github.com/gin-gonic/gin"
+
+	"nex/backend/internal/config"
+	"nex/backend/internal/protocol/openai"
+	"nex/backend/internal/provider"
+	"nex/backend/internal/router"
+)
+
+// OpenAIHandler OpenAI 协议处理器
+type OpenAIHandler struct {
+	client *provider.Client
+	router *router.Router
+}
+
+// NewOpenAIHandler 创建 OpenAI 处理器
+func NewOpenAIHandler() *OpenAIHandler {
+	return &OpenAIHandler{
+		client: provider.NewClient(),
+		router: router.NewRouter(),
+	}
+}
+
+// HandleChatCompletions 处理 Chat Completions 请求
+func (h *OpenAIHandler) HandleChatCompletions(c *gin.Context) {
+	// 解析请求
+	var req openai.ChatCompletionRequest
+	if err := c.ShouldBindJSON(&req); err != nil {
+		c.JSON(http.StatusBadRequest, openai.ErrorResponse{
+			Error: openai.ErrorDetail{
+				Message: "无效的请求格式: " + err.Error(),
+				Type:    "invalid_request_error",
+			},
+		})
+		return
+	}
+
+	// 路由到供应商
+	routeResult, err := h.router.Route(req.Model)
+	if err != nil {
+		h.handleError(c, err)
+		return
+	}
+
+	// 根据是否流式选择处理方式
+	if req.Stream {
+		h.handleStreamRequest(c, &req, routeResult)
+	} else {
+		h.handleNonStreamRequest(c, &req, routeResult)
+	}
+}
+
+// handleNonStreamRequest 处理非流式请求
+func (h *OpenAIHandler) handleNonStreamRequest(c *gin.Context, req *openai.ChatCompletionRequest, routeResult *router.RouteResult) {
+	// 发送请求到供应商
+	resp, err := h.client.SendRequest(c.Request.Context(), req, routeResult.Provider.APIKey, routeResult.Provider.BaseURL)
+	if err != nil {
+		c.JSON(http.StatusInternalServerError, openai.ErrorResponse{
+			Error: openai.ErrorDetail{
+				Message: "供应商请求失败: " + err.Error(),
+				Type:    "api_error",
+			},
+		})
+		return
+	}
+
+	// 记录统计
+	go func() {
+		_ = config.RecordRequest(routeResult.Provider.ID, req.Model)
+	}()
+
+	// 返回响应
+	c.JSON(http.StatusOK, resp)
+}
+
+// handleStreamRequest 处理流式请求
+func (h *OpenAIHandler) handleStreamRequest(c *gin.Context, req *openai.ChatCompletionRequest, routeResult *router.RouteResult) {
+	// 发送流式请求到供应商
+	eventChan, err := h.client.SendStreamRequest(c.Request.Context(), req, routeResult.Provider.APIKey, routeResult.Provider.BaseURL)
+	if err != nil {
+		c.JSON(http.StatusInternalServerError, openai.ErrorResponse{
+			Error: openai.ErrorDetail{
+				Message: "供应商请求失败: " + err.Error(),
+				Type:    "api_error",
+			},
+		})
+		return
+	}
+
+	// 设置 SSE 响应头
+	c.Header("Content-Type", "text/event-stream")
+	c.Header("Cache-Control", "no-cache")
+	c.Header("Connection", "keep-alive")
+
+	// 创建流写入器
+	writer := bufio.NewWriter(c.Writer)
+
+	// 流式转发事件
+	for event := range eventChan {
+		if event.Error != nil {
+			// 流错误，记录日志
+			fmt.Printf("流错误: %v\n", event.Error)
+			break
+		}
+
+		if event.Done {
+			// 流结束
+			writer.WriteString("data: [DONE]\n\n")
+			writer.Flush()
+			break
+		}
+
+		// 写入事件数据
+		writer.WriteString("data: ")
+		writer.Write(event.Data)
+		writer.WriteString("\n\n")
+		writer.Flush()
+	}
+
+	// 记录统计
+	go func() {
+		_ = config.RecordRequest(routeResult.Provider.ID, req.Model)
+	}()
+}
+
+// handleError 处理路由错误
+func (h *OpenAIHandler) handleError(c *gin.Context, err error) {
+	switch err {
+	case router.ErrModelNotFound:
+		c.JSON(http.StatusNotFound, openai.ErrorResponse{
+			Error: openai.ErrorDetail{
+				Message: "模型未找到",
+				Type:    "invalid_request_error",
+				Code:    "model_not_found",
+			},
+		})
+	case router.ErrModelDisabled:
+		c.JSON(http.StatusNotFound, openai.ErrorResponse{
+			Error: openai.ErrorDetail{
+				Message: "模型已禁用",
+				Type:    "invalid_request_error",
+				Code:    "model_disabled",
+			},
+		})
+	case router.ErrProviderDisabled:
+		c.JSON(http.StatusNotFound, openai.ErrorResponse{
+			Error: openai.ErrorDetail{
+				Message: "供应商已禁用",
+				Type:    "invalid_request_error",
+				Code:    "provider_disabled",
+			},
+		})
+	default:
+		c.JSON(http.StatusInternalServerError, openai.ErrorResponse{
+			Error: openai.ErrorDetail{
+				Message: "内部错误: " + err.Error(),
+				Type:    "internal_error",
+			},
+		})
+	}
+}