oadmin/new-api
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: 修复重试后请求结构混乱，修复rerank端点无法使用

Browse Source
This commit is contained in:
CaIon
2025-08-23 13:12:15 +08:00
parent e581422810
commit 4f23e53002
20 changed files with 273 additions and 106 deletions
Download Patch File Download Diff File Expand all files Collapse all files

35
relay/claude_handler.go
View File

@@ -21,13 +21,18 @@ func ClaudeHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *typ
info.InitChannelMeta(c)
textRequest, ok := info.Request.(*dto.ClaudeRequest)
claudeReq, ok := info.Request.(*dto.ClaudeRequest)
if !ok {
common.FatalLog(fmt.Sprintf("invalid request type, expected *dto.ClaudeRequest, got %T", info.Request))
}
err := helper.ModelMappedHelper(c, info, textRequest)
request, err := common.DeepCopy(claudeReq)
if err != nil {
return types.NewError(fmt.Errorf("failed to copy request to ClaudeRequest: %w", err), types.ErrorCodeInvalidRequest, types.ErrOptionWithSkipRetry())
}
err = helper.ModelMappedHelper(c, info, request)
if err != nil {
return types.NewError(err, types.ErrorCodeChannelModelMappedError, types.ErrOptionWithSkipRetry())
}
@@ -38,30 +43,30 @@ func ClaudeHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *typ
}
adaptor.Init(info)
if textRequest.MaxTokens == 0 {
textRequest.MaxTokens = uint(model_setting.GetClaudeSettings().GetDefaultMaxTokens(textRequest.Model))
if request.MaxTokens == 0 {
request.MaxTokens = uint(model_setting.GetClaudeSettings().GetDefaultMaxTokens(request.Model))
}
if model_setting.GetClaudeSettings().ThinkingAdapterEnabled &&
strings.HasSuffix(textRequest.Model, "-thinking") {
if textRequest.Thinking == nil {
strings.HasSuffix(request.Model, "-thinking") {
if request.Thinking == nil {
// 因为BudgetTokens 必须大于1024
if textRequest.MaxTokens < 1280 {
textRequest.MaxTokens = 1280
if request.MaxTokens < 1280 {
request.MaxTokens = 1280
}
// BudgetTokens 为 max_tokens 的 80%
textRequest.Thinking = &dto.Thinking{
request.Thinking = &dto.Thinking{
Type: "enabled",
BudgetTokens: common.GetPointer[int](int(float64(textRequest.MaxTokens) * model_setting.GetClaudeSettings().ThinkingAdapterBudgetTokensPercentage)),
BudgetTokens: common.GetPointer[int](int(float64(request.MaxTokens) * model_setting.GetClaudeSettings().ThinkingAdapterBudgetTokensPercentage)),
}
// TODO: 临时处理
// https://docs.anthropic.com/en/docs/build-with-claude/extended-thinking#important-considerations-when-using-extended-thinking
textRequest.TopP = 0
textRequest.Temperature = common.GetPointer[float64](1.0)
request.TopP = 0
request.Temperature = common.GetPointer[float64](1.0)
}
textRequest.Model = strings.TrimSuffix(textRequest.Model, "-thinking")
info.UpstreamModelName = textRequest.Model
request.Model = strings.TrimSuffix(request.Model, "-thinking")
info.UpstreamModelName = request.Model
}
var requestBody io.Reader
@@ -72,7 +77,7 @@ func ClaudeHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *typ
}
requestBody = bytes.NewBuffer(body)
} else {
convertedRequest, err := adaptor.ConvertClaudeRequest(c, info, textRequest)
convertedRequest, err := adaptor.ConvertClaudeRequest(c, info, request)
if err != nil {
return types.NewError(err, types.ErrorCodeConvertRequestFailed, types.ErrOptionWithSkipRetry())
}