Skip to content

Commit d6785c7

Browse files
BlueSkyXNBlueSkyXN
committed
0.1.2
1 parent 35e20e7 commit d6785c7

2 files changed

Lines changed: 67 additions & 34 deletions

File tree

‎.github/workflows/go-build.yml‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -54,7 +54,7 @@ jobs:
5454
with:
5555
name: api-pool-${{ matrix.os }}-${{ matrix.arch }}-${{ steps.set-version.outputs.version }}
5656
path: api-pool-${{ matrix.os }}-${{ matrix.arch }}${{ matrix.os == 'windows' && '.exe' || '' }}
57-
retention-days: 7
57+
retention-days: 90
5858

5959
build-wo2api:
6060
name: Build wo2api
@@ -100,7 +100,7 @@ jobs:
100100
with:
101101
name: wo2api-${{ matrix.os }}-${{ matrix.arch }}-${{ steps.set-version.outputs.version }}
102102
path: wo2api-${{ matrix.os }}-${{ matrix.arch }}${{ matrix.os == 'windows' && '.exe' || '' }}
103-
retention-days: 7
103+
retention-days: 90
104104

105105
build-hunyuan2api:
106106
name: Build hunyuan2api
@@ -146,4 +146,4 @@ jobs:
146146
with:
147147
name: hunyuan2api-${{ matrix.os }}-${{ matrix.arch }}-${{ steps.set-version.outputs.version }}
148148
path: hunyuan2api-${{ matrix.os }}-${{ matrix.arch }}${{ matrix.os == 'windows' && '.exe' || '' }}
149-
retention-days: 7
149+
retention-days: 90

‎hunyuan2api.go‎

Lines changed: 64 additions & 31 deletions
Original file line numberDiff line numberDiff line change
@@ -29,10 +29,16 @@ type Config struct {
2929
MaxRetries int // 最大重试次数
3030
Timeout int // 请求超时时间(秒)
3131
VerifySSL bool // 是否验证SSL证书
32-
ModelName string // 模型名称
32+
ModelName string // 默认模型名称
3333
BearerToken string // Bearer Token (默认提供公开Token)
3434
}
3535

36+
// 支持的模型列表
37+
var SupportedModels = []string{
38+
"hunyuan-t1-latest",
39+
"hunyuan-turbos-latest",
40+
}
41+
3642
// 腾讯混元 API 目标URL
3743
const (
3844
TargetURL = "https://llm.hunyuan.tencent.com/aide/api/v2/triton_image/demo_text_chat/"
@@ -57,7 +63,7 @@ func parseFlags() *Config {
5763
flag.IntVar(&cfg.MaxRetries, "max-retries", 3, "Maximum number of retries for failed requests")
5864
flag.IntVar(&cfg.Timeout, "timeout", 300, "Request timeout in seconds")
5965
flag.BoolVar(&cfg.VerifySSL, "verify-ssl", true, "Verify SSL certificates")
60-
flag.StringVar(&cfg.ModelName, "model", "hunyuan-t1-latest", "Hunyuan model name")
66+
flag.StringVar(&cfg.ModelName, "model", "hunyuan-t1-latest", "Default Hunyuan model name")
6167
flag.StringVar(&cfg.BearerToken, "token", "7auGXNATFSKl7dF", "Bearer token for Hunyuan API")
6268
flag.Parse()
6369

@@ -234,8 +240,8 @@ func main() {
234240
// 初始化日志
235241
initLogger(appConfig.LogLevel)
236242

237-
logInfo("启动服务: TargetURL=%s, Address=%s, Port=%s, Version=%s, LogLevel=%s, BearerToken=***",
238-
TargetURL, appConfig.Address, appConfig.Port, Version, appConfig.LogLevel)
243+
logInfo("启动服务: TargetURL=%s, Address=%s, Port=%s, Version=%s, LogLevel=%s, 支持模型=%v, BearerToken=***",
244+
TargetURL, appConfig.Address, appConfig.Port, Version, appConfig.LogLevel, SupportedModels)
239245

240246
// 配置更高的并发处理能力
241247
http.DefaultTransport.(*http.Transport).MaxIdleConnsPerHost = 100
@@ -415,6 +421,16 @@ func generateQueryID() string {
415421
return fmt.Sprintf("%s%d", getRandomString(8), time.Now().UnixNano())
416422
}
417423

424+
// 判断模型是否在支持列表中
425+
func isModelSupported(modelName string) bool {
426+
for _, supportedModel := range SupportedModels {
427+
if modelName == supportedModel {
428+
return true
429+
}
430+
}
431+
return false
432+
}
433+
418434
// 处理模型列表请求
419435
func handleModelsRequest(w http.ResponseWriter, r *http.Request) {
420436
logInfo("处理模型列表请求")
@@ -423,21 +439,25 @@ func handleModelsRequest(w http.ResponseWriter, r *http.Request) {
423439
w.Header().Set("Content-Type", "application/json")
424440
w.WriteHeader(http.StatusOK)
425441

442+
// 构建模型数据
443+
modelData := make([]map[string]interface{}, 0, len(SupportedModels))
444+
for _, model := range SupportedModels {
445+
modelData = append(modelData, map[string]interface{}{
446+
"id": model,
447+
"object": "model",
448+
"created": time.Now().Unix(),
449+
"owned_by": "TencentCloud",
450+
"capabilities": map[string]interface{}{
451+
"chat": true,
452+
"completions": true,
453+
"reasoning": true,
454+
},
455+
})
456+
}
457+
426458
modelsList := map[string]interface{}{
427459
"object": "list",
428-
"data": []map[string]interface{}{
429-
{
430-
"id": "hunyuan-t1-latest",
431-
"object": "model",
432-
"created": time.Now().Unix(),
433-
"owned_by": "TencentCloud",
434-
"capabilities": map[string]interface{}{
435-
"chat": true,
436-
"completions": true,
437-
"reasoning": true,
438-
},
439-
},
440-
},
460+
"data": modelData,
441461
}
442462

443463
json.NewEncoder(w).Encode(modelsList)
@@ -468,10 +488,23 @@ func handleChatCompletionRequest(w http.ResponseWriter, r *http.Request) {
468488
// 是否使用流式处理
469489
isStream := apiReq.Stream
470490

491+
// 确定使用的模型
492+
modelName := appConfig.ModelName
493+
if apiReq.Model != "" {
494+
// 检查请求的模型是否是我们支持的
495+
if isModelSupported(apiReq.Model) {
496+
modelName = apiReq.Model
497+
} else {
498+
logWarn("[reqID:%s] 请求的模型 %s 不支持，使用默认模型 %s", reqID, apiReq.Model, modelName)
499+
}
500+
}
501+
502+
logInfo("[reqID:%s] 使用模型: %s", reqID, modelName)
503+
471504
// 创建混元API请求
472505
hunyuanReq := HunyuanRequest{
473506
Stream: true, // 混元API总是使用流式响应
474-
Model: appConfig.ModelName,
507+
Model: modelName,
475508
QueryID: generateQueryID(),
476509
Messages: apiReq.Messages,
477510
StreamModeration: true,
@@ -593,7 +626,7 @@ func handleStreamingRequest(w http.ResponseWriter, r *http.Request, hunyuanReq H
593626
}
594627

595628
// 发送角色块
596-
roleChunk := createRoleChunk(respID, createdTime)
629+
roleChunk := createRoleChunk(respID, createdTime, hunyuanReq.Model)
597630
w.Write([]byte("data: " + string(roleChunk) + "\n\n"))
598631
flusher.Flush()
599632

@@ -643,14 +676,14 @@ func handleStreamingRequest(w http.ResponseWriter, r *http.Request, hunyuanReq H
643676
for _, choice := range hunyuanResp.Choices {
644677
if choice.Delta.Content != "" {
645678
// 发送内容块
646-
contentChunk := createContentChunk(respID, createdTime, choice.Delta.Content)
679+
contentChunk := createContentChunk(respID, createdTime, hunyuanReq.Model, choice.Delta.Content)
647680
w.Write([]byte("data: " + string(contentChunk) + "\n\n"))
648681
flusher.Flush()
649682
}
650683

651684
if choice.Delta.ReasoningContent != "" {
652685
// 发送推理内容块
653-
reasoningChunk := createReasoningChunk(respID, createdTime, choice.Delta.ReasoningContent)
686+
reasoningChunk := createReasoningChunk(respID, createdTime, hunyuanReq.Model, choice.Delta.ReasoningContent)
654687
w.Write([]byte("data: " + string(reasoningChunk) + "\n\n"))
655688
flusher.Flush()
656689
}
@@ -659,7 +692,7 @@ func handleStreamingRequest(w http.ResponseWriter, r *http.Request, hunyuanReq H
659692
if choice.FinishReason != nil {
660693
finishReason := *choice.FinishReason
661694
if finishReason != "" {
662-
doneChunk := createDoneChunk(respID, createdTime, finishReason)
695+
doneChunk := createDoneChunk(respID, createdTime, hunyuanReq.Model, finishReason)
663696
w.Write([]byte("data: " + string(doneChunk) + "\n\n"))
664697
flusher.Flush()
665698
}
@@ -670,7 +703,7 @@ func handleStreamingRequest(w http.ResponseWriter, r *http.Request, hunyuanReq H
670703

671704
// 发送结束信号（如果没有正常结束）
672705
finishReason := "stop"
673-
doneChunk := createDoneChunk(respID, createdTime, finishReason)
706+
doneChunk := createDoneChunk(respID, createdTime, hunyuanReq.Model, finishReason)
674707
w.Write([]byte("data: " + string(doneChunk) + "\n\n"))
675708
w.Write([]byte("data: [DONE]\n\n"))
676709
flusher.Flush()
@@ -830,12 +863,12 @@ func extractFullContentFromStream(bodyBytes []byte, reqID string) (string, strin
830863
}
831864

832865
// 创建角色块
833-
func createRoleChunk(id string, created int64) []byte {
866+
func createRoleChunk(id string, created int64, model string) []byte {
834867
chunk := StreamChunk{
835868
ID: id,
836869
Object: "chat.completion.chunk",
837870
Created: created,
838-
Model: appConfig.ModelName,
871+
Model: model,
839872
Choices: []struct {
840873
Index int `json:"index"`
841874
FinishReason *string `json:"finish_reason,omitempty"`
@@ -863,12 +896,12 @@ func createRoleChunk(id string, created int64) []byte {
863896
}
864897

865898
// 创建内容块
866-
func createContentChunk(id string, created int64, content string) []byte {
899+
func createContentChunk(id string, created int64, model string, content string) []byte {
867900
chunk := StreamChunk{
868901
ID: id,
869902
Object: "chat.completion.chunk",
870903
Created: created,
871-
Model: appConfig.ModelName,
904+
Model: model,
872905
Choices: []struct {
873906
Index int `json:"index"`
874907
FinishReason *string `json:"finish_reason,omitempty"`
@@ -896,12 +929,12 @@ func createContentChunk(id string, created int64, content string) []byte {
896929
}
897930

898931
// 创建推理内容块
899-
func createReasoningChunk(id string, created int64, reasoningContent string) []byte {
932+
func createReasoningChunk(id string, created int64, model string, reasoningContent string) []byte {
900933
chunk := StreamChunk{
901934
ID: id,
902935
Object: "chat.completion.chunk",
903936
Created: created,
904-
Model: appConfig.ModelName,
937+
Model: model,
905938
Choices: []struct {
906939
Index int `json:"index"`
907940
FinishReason *string `json:"finish_reason,omitempty"`
@@ -929,13 +962,13 @@ func createReasoningChunk(id string, created int64, reasoningContent string) []b
929962
}
930963

931964
// 创建完成块
932-
func createDoneChunk(id string, created int64, reason string) []byte {
965+
func createDoneChunk(id string, created int64, model string, reason string) []byte {
933966
finishReason := reason
934967
chunk := StreamChunk{
935968
ID: id,
936969
Object: "chat.completion.chunk",
937970
Created: created,
938-
Model: appConfig.ModelName,
971+
Model: model,
939972
Choices: []struct {
940973
Index int `json:"index"`
941974
FinishReason *string `json:"finish_reason,omitempty"`

0 commit comments

Comments
 (0)