From 0c63c3b44daf6803e238b42a3f84ac1573f089ad Mon Sep 17 00:00:00 2001
From: "DESKTOP-72TV0V4\\caoxiaozhu" <leocaoxiaozhu@163.com>
Date: Sat, 7 Mar 2026 17:18:30 +0800
Subject: [PATCH] =?UTF-8?q?feat:=20=E6=94=AF=E6=8C=81=20Embedding=20?=
 =?UTF-8?q?=E6=A8=A1=E5=9E=8B=E8=BF=9E=E6=8E=A5=E6=B5=8B=E8=AF=95?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- 添加 model_type 字段区分 chat 和 embedding 模型
- 根据 model_type 使用不同的 API 端点和请求格式
- embedding 模型使用 /v1/embeddings 端点

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 server/internal/model/model_info.go      |  1 +
 server/internal/service/model_service.go | 64 +++++++++++++++---------
 2 files changed, 40 insertions(+), 25 deletions(-)

diff --git a/server/internal/model/model_info.go b/server/internal/model/model_info.go
index 38da606..b1666bf 100644
--- a/server/internal/model/model_info.go
+++ b/server/internal/model/model_info.go
@@ -58,6 +58,7 @@ type UpdateModelRequest struct {
 type TestModelRequest struct {
 	Provider    string `json:"provider" binding:"required"`
 	Model       string `json:"model" binding:"required"`
+	ModelType   string `json:"model_type" binding:"required"`
 	APIKey      string `json:"api_key" binding:"required"`
 	BaseURL     string `json:"base_url" binding:"required"`
 	APIEndpoint string `json:"api_endpoint"`
diff --git a/server/internal/service/model_service.go b/server/internal/service/model_service.go
index 7410caf..99a1360 100644
--- a/server/internal/service/model_service.go
+++ b/server/internal/service/model_service.go
@@ -114,8 +114,6 @@ func (s *ModelService) Delete(id string) error {
 
 // TestConnection 测试模型连接
 func (s *ModelService) TestConnection(req model.TestModelRequest) (*model.TestModelResponse, error) {
-	log.Printf("[TestConnection] 开始测试连接: provider=%s, model=%s, base_url=%s, api_endpoint=%s", req.Provider, req.Model, req.BaseURL, req.APIEndpoint)
-
 	// 构建请求 URL
 	baseURL := req.BaseURL
 	// 去掉 base_url 末尾的斜杠
@@ -126,31 +124,49 @@ func (s *ModelService) TestConnection(req model.TestModelRequest) (*model.TestMo
 		apiEndpoint := strings.TrimLeft(req.APIEndpoint, "/")
 		baseURL = baseURL + "/" + apiEndpoint
 	} else {
-		// 默认端点 - 根据不同 provider 设置
-		switch req.Provider {
-		case "OpenAI":
-			baseURL = baseURL + "/v1/chat/completions"
-		case "Ollama":
-			baseURL = baseURL + "/api/chat"
-		case "ali", "Ali", "aliyun", "Aliyun":
-			// 阿里云 DashScope 兼容 OpenAI 格式，需要添加 /chat/completions
-			// base_url 格式: https://dashscope.aliyuncs.com/compatible-mode/v1
-			baseURL = baseURL + "/chat/completions"
+		// 根据 model_type 确定端点
+		switch req.ModelType {
+		case "embedding":
+			// embedding 模型使用 /v1/embeddings
+			switch req.Provider {
+			case "Ollama":
+				baseURL = baseURL + "/api/embeddings"
+			default:
+				baseURL = baseURL + "/v1/embeddings"
+			}
 		default:
-			// 默认使用 OpenAI 兼容格式
-			baseURL = baseURL + "/v1/chat/completions"
+			// chat 模型使用 /chat/completions
+			switch req.Provider {
+			case "OpenAI":
+				baseURL = baseURL + "/v1/chat/completions"
+			case "Ollama":
+				baseURL = baseURL + "/api/chat"
+			case "ali", "Ali", "aliyun", "Aliyun":
+				// 阿里云 DashScope 兼容 OpenAI 格式
+				baseURL = baseURL + "/chat/completions"
+			default:
+				// 默认使用 OpenAI 兼容格式
+				baseURL = baseURL + "/v1/chat/completions"
+			}
 		}
 	}
 
-	log.Printf("[TestConnection] 请求 URL: %s", baseURL)
-
-	// 构建请求体
-	requestBody := map[string]interface{}{
-		"model": req.Model,
-		"messages": []map[string]string{
-			{"role": "user", "content": "Hello"},
-		},
-		"max_tokens": 10,
+	// 构建请求体 - 根据 model_type 使用不同的格式
+	var requestBody map[string]interface{}
+	if req.ModelType == "embedding" {
+		requestBody = map[string]interface{}{
+			"model":  req.Model,
+			"input":  "Hello",
+			"format": "float",
+		}
+	} else {
+		requestBody = map[string]interface{}{
+			"model": req.Model,
+			"messages": []map[string]string{
+				{"role": "user", "content": "Hello"},
+			},
+			"max_tokens": 10,
+		}
 	}
 
 	body, err := json.Marshal(requestBody)
@@ -191,8 +207,6 @@ func (s *ModelService) TestConnection(req model.TestModelRequest) (*model.TestMo
 		return &model.TestModelResponse{Success: false, Message: err.Error()}, nil
 	}
 
-	log.Printf("[TestConnection] 响应状态码: %d, 响应体: %s", resp.StatusCode, string(respBody))
-
 	if resp.StatusCode >= 200 && resp.StatusCode < 300 {
 		return &model.TestModelResponse{Success: true, Message: "Connection successful"}, nil
 	}