fix: responses 流 cache token 未计费

relay/channel/openai/relay_responses.go

@@ -37,10 +37,14 @@ func OaiResponsesHandler(c *gin.Context, info *relaycommon.RelayInfo, resp *http
 
 
 	// compute usage
 	// compute usage
 	usage := dto.Usage{}
 	usage := dto.Usage{}
+	if responsesResponse.Usage != nil {
 		usage.PromptTokens = responsesResponse.Usage.InputTokens
 		usage.PromptTokens = responsesResponse.Usage.InputTokens
 		usage.CompletionTokens = responsesResponse.Usage.OutputTokens
 		usage.CompletionTokens = responsesResponse.Usage.OutputTokens
 		usage.TotalTokens = responsesResponse.Usage.TotalTokens
 		usage.TotalTokens = responsesResponse.Usage.TotalTokens
+		if responsesResponse.Usage.InputTokensDetails != nil {
 			usage.PromptTokensDetails.CachedTokens = responsesResponse.Usage.InputTokensDetails.CachedTokens
 			usage.PromptTokensDetails.CachedTokens = responsesResponse.Usage.InputTokensDetails.CachedTokens
+		}
+	}
 	// 解析 Tools 用量
 	// 解析 Tools 用量
 	for _, tool := range responsesResponse.Tools {
 	for _, tool := range responsesResponse.Tools {
 		info.ResponsesUsageInfo.BuiltInTools[common.Interface2String(tool["type"])].CallCount++
 		info.ResponsesUsageInfo.BuiltInTools[common.Interface2String(tool["type"])].CallCount++
@@ -65,9 +69,14 @@ func OaiResponsesStreamHandler(c *gin.Context, info *relaycommon.RelayInfo, resp
 			sendResponsesStreamData(c, streamResponse, data)
 			sendResponsesStreamData(c, streamResponse, data)
 			switch streamResponse.Type {
 			switch streamResponse.Type {
 			case "response.completed":
 			case "response.completed":
+				if streamResponse.Response.Usage != nil {
 					usage.PromptTokens = streamResponse.Response.Usage.InputTokens
 					usage.PromptTokens = streamResponse.Response.Usage.InputTokens
 					usage.CompletionTokens = streamResponse.Response.Usage.OutputTokens
 					usage.CompletionTokens = streamResponse.Response.Usage.OutputTokens
 					usage.TotalTokens = streamResponse.Response.Usage.TotalTokens
 					usage.TotalTokens = streamResponse.Response.Usage.TotalTokens
+					if streamResponse.Response.Usage.InputTokensDetails != nil {
+						usage.PromptTokensDetails.CachedTokens = streamResponse.Response.Usage.InputTokensDetails.CachedTokens
+					}
+				}
 			case "response.output_text.delta":
 			case "response.output_text.delta":
 				// 处理输出文本
 				// 处理输出文本
 				responseTextBuilder.WriteString(streamResponse.Delta)
 				responseTextBuilder.WriteString(streamResponse.Delta)