fix: Claude 流式断流时不再整份覆盖 usage,保留 cache 计费字段

HandleStreamFinalResponse 在 !Done 时调用 ResponseText2Usage 整份覆盖
claudeInfo.Usage,导致 message_start 已获取的 CacheReadInputTokens、
CacheCreationInputTokens 等字段丢失,prompt 退化为占位值 1。

修复:
- 只补缺失的 CompletionTokens/PromptTokens,保留已有 cache 数据
- PromptTokens 兜底改用 info.GetEstimatePromptTokens()(与其他渠道对齐)

Fixes #4127
This commit is contained in:
zuiho 2026-04-07 17:41:08 +08:00
parent 960bf9c49e
commit f7cdc727df

View File

@ -809,7 +809,15 @@ func HandleStreamFinalResponse(c *gin.Context, info *relaycommon.RelayInfo, clau
if common.DebugEnabled {
common.SysLog("claude response usage is not complete, maybe upstream error")
}
claudeInfo.Usage = service.ResponseText2Usage(c, claudeInfo.ResponseText.String(), info.UpstreamModelName, claudeInfo.Usage.PromptTokens)
// 只补缺失字段,不整份覆盖——保留 message_start 已拿到的 cache 字段
fallback := service.ResponseText2Usage(c, claudeInfo.ResponseText.String(), info.UpstreamModelName, info.GetEstimatePromptTokens())
if claudeInfo.Usage.CompletionTokens == 0 {
claudeInfo.Usage.CompletionTokens = fallback.CompletionTokens
}
if claudeInfo.Usage.PromptTokens == 0 {
claudeInfo.Usage.PromptTokens = fallback.PromptTokens
}
claudeInfo.Usage.TotalTokens = claudeInfo.Usage.PromptTokens + claudeInfo.Usage.CompletionTokens
}
if claudeInfo.Usage != nil {
claudeInfo.Usage.UsageSemantic = "anthropic"