github/new-api
1
0
Fork 0
mirror of https://github.com/QuantumNous/new-api.git synced 2026-03-30 02:05:21 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: openai 格式请求 claude 没计费 create cache token

Browse Source
This commit is contained in:
Xyfacai
2025-09-10 15:30:23 +08:00
parent 3f9698bb47
commit fcdfd027cd
10 changed files with 31 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
relay/compatible_handler.go
View File

@@ -158,7 +158,7 @@ func TextHelper(c *gin.Context, info *relaycommon.RelayInfo) (newAPIError *types
httpResp = resp.(*http.Response)
info.IsStream = info.IsStream || strings.HasPrefix(httpResp.Header.Get("Content-Type"), "text/event-stream")
if httpResp.StatusCode != http.StatusOK {
newApiErr := service.RelayErrorHandler(httpResp, false)
newApiErr := service.RelayErrorHandler(c.Request.Context(), httpResp, false)
// reset status code 重置状态码
service.ResetStatusCode(newApiErr, statusCodeMappingStr)
return newApiErr
@@ -195,6 +195,8 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage
imageTokens := usage.PromptTokensDetails.ImageTokens
audioTokens := usage.PromptTokensDetails.AudioTokens
completionTokens := usage.CompletionTokens
cachedCreationTokens := usage.PromptTokensDetails.CachedCreationTokens
modelName := relayInfo.OriginModelName
tokenName := ctx.GetString("token_name")
@@ -204,6 +206,7 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage
modelRatio := relayInfo.PriceData.ModelRatio
groupRatio := relayInfo.PriceData.GroupRatioInfo.GroupRatio
modelPrice := relayInfo.PriceData.ModelPrice
cachedCreationRatio := relayInfo.PriceData.CacheCreationRatio
// Convert values to decimal for precise calculation
dPromptTokens := decimal.NewFromInt(int64(promptTokens))
@@ -211,12 +214,14 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage
dImageTokens := decimal.NewFromInt(int64(imageTokens))
dAudioTokens := decimal.NewFromInt(int64(audioTokens))
dCompletionTokens := decimal.NewFromInt(int64(completionTokens))
dCachedCreationTokens := decimal.NewFromInt(int64(cachedCreationTokens))
dCompletionRatio := decimal.NewFromFloat(completionRatio)
dCacheRatio := decimal.NewFromFloat(cacheRatio)
dImageRatio := decimal.NewFromFloat(imageRatio)
dModelRatio := decimal.NewFromFloat(modelRatio)
dGroupRatio := decimal.NewFromFloat(groupRatio)
dModelPrice := decimal.NewFromFloat(modelPrice)
dCachedCreationRatio := decimal.NewFromFloat(cachedCreationRatio)
dQuotaPerUnit := decimal.NewFromFloat(common.QuotaPerUnit)
ratio := dModelRatio.Mul(dGroupRatio)
@@ -284,6 +289,11 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage
baseTokens = baseTokens.Sub(dCacheTokens)
cachedTokensWithRatio = dCacheTokens.Mul(dCacheRatio)
}
var dCachedCreationTokensWithRatio decimal.Decimal
if !dCachedCreationTokens.IsZero() {
baseTokens = baseTokens.Sub(dCachedCreationTokens)
dCachedCreationTokensWithRatio = dCachedCreationTokens.Mul(dCachedCreationRatio)
}
// 减去 image tokens
var imageTokensWithRatio decimal.Decimal
@@ -302,7 +312,9 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage
extraContent += fmt.Sprintf("Audio Input 花费 %s", audioInputQuota.String())
}
}
promptQuota := baseTokens.Add(cachedTokensWithRatio).Add(imageTokensWithRatio)
promptQuota := baseTokens.Add(cachedTokensWithRatio).
Add(imageTokensWithRatio).
Add(dCachedCreationTokensWithRatio)
completionQuota := dCompletionTokens.Mul(dCompletionRatio)
@@ -395,6 +407,10 @@ func postConsumeQuota(ctx *gin.Context, relayInfo *relaycommon.RelayInfo, usage
other["image_ratio"] = imageRatio
other["image_output"] = imageTokens
}
if cachedCreationTokens != 0 {
other["cache_creation_tokens"] = cachedCreationTokens
other["cache_creation_ratio"] = cachedCreationRatio
}
if !dWebSearchQuota.IsZero() {
if relayInfo.ResponsesUsageInfo != nil {
if webSearchTool, exists := relayInfo.ResponsesUsageInfo.BuiltInTools[dto.BuildInToolWebSearchPreview]; exists {