CLIProxyAPIPlus/sdk/api/handlers/openai/openai_responses_handlers.go

// Package openai provides HTTP handlers for OpenAIResponses API endpoints.
// This package implements the OpenAIResponses-compatible API interface, including model listing
// and chat completion functionality. It supports both streaming and non-streaming responses,
// and manages a pool of clients to interact with backend services.
// The handlers translate OpenAIResponses API requests to the appropriate backend format and
// convert responses back to OpenAIResponses-compatible format.
package openai

import (
	"bytes"
	"context"
	"fmt"
	"net/http"

	"github.com/gin-gonic/gin"
	. "github.com/router-for-me/CLIProxyAPI/v6/internal/constant"
	"github.com/router-for-me/CLIProxyAPI/v6/internal/interfaces"
	"github.com/router-for-me/CLIProxyAPI/v6/internal/registry"
	responsesconverter "github.com/router-for-me/CLIProxyAPI/v6/internal/translator/openai/openai/responses"
	"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers"
	"github.com/tidwall/gjson"
)

// OpenAIResponsesAPIHandler contains the handlers for OpenAIResponses API endpoints.
// It holds a pool of clients to interact with the backend service.
type OpenAIResponsesAPIHandler struct {
	*handlers.BaseAPIHandler
}

// NewOpenAIResponsesAPIHandler creates a new OpenAIResponses API handlers instance.
// It takes an BaseAPIHandler instance as input and returns an OpenAIResponsesAPIHandler.
//
// Parameters:
//   - apiHandlers: The base API handlers instance
//
// Returns:
//   - *OpenAIResponsesAPIHandler: A new OpenAIResponses API handlers instance
func NewOpenAIResponsesAPIHandler(apiHandlers *handlers.BaseAPIHandler) *OpenAIResponsesAPIHandler {
	return &OpenAIResponsesAPIHandler{
		BaseAPIHandler: apiHandlers,
	}
}

// HandlerType returns the identifier for this handler implementation.
func (h *OpenAIResponsesAPIHandler) HandlerType() string {
	return OpenaiResponse
}

// Models returns the OpenAIResponses-compatible model metadata supported by this handler.
func (h *OpenAIResponsesAPIHandler) Models() []map[string]any {
	// Get dynamic models from the global registry
	modelRegistry := registry.GetGlobalRegistry()
	return modelRegistry.GetAvailableModels("openai")
}

// OpenAIResponsesModels handles the /v1/models endpoint.
// It returns a list of available AI models with their capabilities
// and specifications in OpenAIResponses-compatible format.
func (h *OpenAIResponsesAPIHandler) OpenAIResponsesModels(c *gin.Context) {
	c.JSON(http.StatusOK, gin.H{
		"object": "list",
		"data":   h.Models(),
	})
}

// Responses handles the /v1/responses endpoint.
// It determines whether the request is for a streaming or non-streaming response
// and calls the appropriate handler based on the model provider.
//
// Parameters:
//   - c: The Gin context containing the HTTP request and response
func (h *OpenAIResponsesAPIHandler) Responses(c *gin.Context) {
	rawJSON, err := c.GetRawData()
	// If data retrieval fails, return a 400 Bad Request error.
	if err != nil {
		c.JSON(http.StatusBadRequest, handlers.ErrorResponse{
			Error: handlers.ErrorDetail{
				Message: fmt.Sprintf("Invalid request: %v", err),
				Type:    "invalid_request_error",
			},
		})
		return
	}

	// Check if the client requested a streaming response.
	streamResult := gjson.GetBytes(rawJSON, "stream")
	stream := streamResult.Type == gjson.True

	modelName := gjson.GetBytes(rawJSON, "model").String()
	if overrideEndpoint, ok := resolveEndpointOverride(modelName, openAIResponsesEndpoint); ok && overrideEndpoint == openAIChatEndpoint {
		chatJSON := responsesconverter.ConvertOpenAIResponsesRequestToOpenAIChatCompletions(modelName, rawJSON, stream)
		stream = gjson.GetBytes(chatJSON, "stream").Bool()
		if stream {
			h.handleStreamingResponseViaChat(c, rawJSON, chatJSON)
		} else {
			h.handleNonStreamingResponseViaChat(c, rawJSON, chatJSON)
		}
		return
	}

	if stream {
		h.handleStreamingResponse(c, rawJSON)
	} else {
		h.handleNonStreamingResponse(c, rawJSON)
	}

}

// handleNonStreamingResponse handles non-streaming chat completion responses
// for Gemini models. It selects a client from the pool, sends the request, and
// aggregates the response before sending it back to the client in OpenAIResponses format.
//
// Parameters:
//   - c: The Gin context containing the HTTP request and response
//   - rawJSON: The raw JSON bytes of the OpenAIResponses-compatible request
func (h *OpenAIResponsesAPIHandler) handleNonStreamingResponse(c *gin.Context, rawJSON []byte) {
	c.Header("Content-Type", "application/json")

	modelName := gjson.GetBytes(rawJSON, "model").String()
	cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
	stopKeepAlive := h.StartNonStreamingKeepAlive(c, cliCtx)

	resp, errMsg := h.ExecuteWithAuthManager(cliCtx, h.HandlerType(), modelName, rawJSON, "")
	stopKeepAlive()
	if errMsg != nil {
		h.WriteErrorResponse(c, errMsg)
		cliCancel(errMsg.Error)
		return
	}
	_, _ = c.Writer.Write(resp)
	cliCancel()
}

func (h *OpenAIResponsesAPIHandler) handleNonStreamingResponseViaChat(c *gin.Context, originalResponsesJSON, chatJSON []byte) {
	c.Header("Content-Type", "application/json")

	modelName := gjson.GetBytes(chatJSON, "model").String()
	cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
	resp, errMsg := h.ExecuteWithAuthManager(cliCtx, OpenAI, modelName, chatJSON, "")
	if errMsg != nil {
		h.WriteErrorResponse(c, errMsg)
		cliCancel(errMsg.Error)
		return
	}
	var param any
	converted := responsesconverter.ConvertOpenAIChatCompletionsResponseToOpenAIResponsesNonStream(cliCtx, modelName, originalResponsesJSON, originalResponsesJSON, resp, &param)
	if converted == "" {
		h.WriteErrorResponse(c, &interfaces.ErrorMessage{
			StatusCode: http.StatusInternalServerError,
			Error:      fmt.Errorf("failed to convert chat completion response to responses format"),
		})
		cliCancel(fmt.Errorf("response conversion failed"))
		return
	}
	_, _ = c.Writer.Write([]byte(converted))
	cliCancel()
}

// handleStreamingResponse handles streaming responses for Gemini models.
// It establishes a streaming connection with the backend service and forwards
// the response chunks to the client in real-time using Server-Sent Events.
//
// Parameters:
//   - c: The Gin context containing the HTTP request and response
//   - rawJSON: The raw JSON bytes of the OpenAIResponses-compatible request
func (h *OpenAIResponsesAPIHandler) handleStreamingResponse(c *gin.Context, rawJSON []byte) {
	// Get the http.Flusher interface to manually flush the response.
	flusher, ok := c.Writer.(http.Flusher)
	if !ok {
		c.JSON(http.StatusInternalServerError, handlers.ErrorResponse{
			Error: handlers.ErrorDetail{
				Message: "Streaming not supported",
				Type:    "server_error",
			},
		})
		return
	}

	// New core execution path
	modelName := gjson.GetBytes(rawJSON, "model").String()
	cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
	dataChan, errChan := h.ExecuteStreamWithAuthManager(cliCtx, h.HandlerType(), modelName, rawJSON, "")

	setSSEHeaders := func() {
		c.Header("Content-Type", "text/event-stream")
		c.Header("Cache-Control", "no-cache")
		c.Header("Connection", "keep-alive")
		c.Header("Access-Control-Allow-Origin", "*")
	}

	// Peek at the first chunk
	for {
		select {
		case <-c.Request.Context().Done():
			cliCancel(c.Request.Context().Err())
			return
		case errMsg, ok := <-errChan:
			if !ok {
				// Err channel closed cleanly; wait for data channel.
				errChan = nil
				continue
			}
			// Upstream failed immediately. Return proper error status and JSON.
			h.WriteErrorResponse(c, errMsg)
			if errMsg != nil {
				cliCancel(errMsg.Error)
			} else {
				cliCancel(nil)
			}
			return
		case chunk, ok := <-dataChan:
			if !ok {
				// Stream closed without data? Send headers and done.
				setSSEHeaders()
				_, _ = c.Writer.Write([]byte("\n"))
				flusher.Flush()
				cliCancel(nil)
				return
			}

			// Success! Set headers.
			setSSEHeaders()

			// Write first chunk logic (matching forwardResponsesStream)
			if bytes.HasPrefix(chunk, []byte("event:")) {
				_, _ = c.Writer.Write([]byte("\n"))
			}
			_, _ = c.Writer.Write(chunk)
			_, _ = c.Writer.Write([]byte("\n"))
			flusher.Flush()

			// Continue
			h.forwardResponsesStream(c, flusher, func(err error) { cliCancel(err) }, dataChan, errChan)
			return
		}
	}
}

func (h *OpenAIResponsesAPIHandler) handleStreamingResponseViaChat(c *gin.Context, originalResponsesJSON, chatJSON []byte) {
	flusher, ok := c.Writer.(http.Flusher)
	if !ok {
		c.JSON(http.StatusInternalServerError, handlers.ErrorResponse{
			Error: handlers.ErrorDetail{
				Message: "Streaming not supported",
				Type:    "server_error",
			},
		})
		return
	}

	modelName := gjson.GetBytes(chatJSON, "model").String()
	cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
	dataChan, errChan := h.ExecuteStreamWithAuthManager(cliCtx, OpenAI, modelName, chatJSON, "")
	var param any

	setSSEHeaders := func() {
		c.Header("Content-Type", "text/event-stream")
		c.Header("Cache-Control", "no-cache")
		c.Header("Connection", "keep-alive")
		c.Header("Access-Control-Allow-Origin", "*")
	}

	for {
		select {
		case <-c.Request.Context().Done():
			cliCancel(c.Request.Context().Err())
			return
		case errMsg, ok := <-errChan:
			if !ok {
				errChan = nil
				continue
			}
			h.WriteErrorResponse(c, errMsg)
			if errMsg != nil {
				cliCancel(errMsg.Error)
			} else {
				cliCancel(nil)
			}
			return
		case chunk, ok := <-dataChan:
			if !ok {
				setSSEHeaders()
				_, _ = c.Writer.Write([]byte("\n"))
				flusher.Flush()
				cliCancel(nil)
				return
			}

			setSSEHeaders()
			writeChatAsResponsesChunk(c, cliCtx, modelName, originalResponsesJSON, chunk, &param)
			flusher.Flush()

			h.forwardChatAsResponsesStream(c, flusher, func(err error) { cliCancel(err) }, dataChan, errChan, cliCtx, modelName, originalResponsesJSON, &param)
			return
		}
	}
}

func writeChatAsResponsesChunk(c *gin.Context, ctx context.Context, modelName string, originalResponsesJSON, chunk []byte, param *any) {
	outputs := responsesconverter.ConvertOpenAIChatCompletionsResponseToOpenAIResponses(ctx, modelName, originalResponsesJSON, originalResponsesJSON, chunk, param)
	for _, out := range outputs {
		if out == "" {
			continue
		}
		if bytes.HasPrefix([]byte(out), []byte("event:")) {
			_, _ = c.Writer.Write([]byte("\n"))
		}
		_, _ = c.Writer.Write([]byte(out))
		_, _ = c.Writer.Write([]byte("\n"))
	}
}

func (h *OpenAIResponsesAPIHandler) forwardChatAsResponsesStream(c *gin.Context, flusher http.Flusher, cancel func(error), data <-chan []byte, errs <-chan *interfaces.ErrorMessage, ctx context.Context, modelName string, originalResponsesJSON []byte, param *any) {
	h.ForwardStream(c, flusher, cancel, data, errs, handlers.StreamForwardOptions{
		WriteChunk: func(chunk []byte) {
			outputs := responsesconverter.ConvertOpenAIChatCompletionsResponseToOpenAIResponses(ctx, modelName, originalResponsesJSON, originalResponsesJSON, chunk, param)
			for _, out := range outputs {
				if out == "" {
					continue
				}
				if bytes.HasPrefix([]byte(out), []byte("event:")) {
					_, _ = c.Writer.Write([]byte("\n"))
				}
				_, _ = c.Writer.Write([]byte(out))
				_, _ = c.Writer.Write([]byte("\n"))
			}
		},
		WriteTerminalError: func(errMsg *interfaces.ErrorMessage) {
			if errMsg == nil {
				return
			}
			status := http.StatusInternalServerError
			if errMsg.StatusCode > 0 {
				status = errMsg.StatusCode
			}
			errText := http.StatusText(status)
			if errMsg.Error != nil && errMsg.Error.Error() != "" {
				errText = errMsg.Error.Error()
			}
			body := handlers.BuildErrorResponseBody(status, errText)
			_, _ = fmt.Fprintf(c.Writer, "\nevent: error\ndata: %s\n\n", string(body))
		},
		WriteDone: func() {
			_, _ = c.Writer.Write([]byte("\n"))
		},
	})
}

func (h *OpenAIResponsesAPIHandler) forwardResponsesStream(c *gin.Context, flusher http.Flusher, cancel func(error), data <-chan []byte, errs <-chan *interfaces.ErrorMessage) {
	h.ForwardStream(c, flusher, cancel, data, errs, handlers.StreamForwardOptions{
		WriteChunk: func(chunk []byte) {
			if bytes.HasPrefix(chunk, []byte("event:")) {
				_, _ = c.Writer.Write([]byte("\n"))
			}
			_, _ = c.Writer.Write(chunk)
			_, _ = c.Writer.Write([]byte("\n"))
		},
		WriteTerminalError: func(errMsg *interfaces.ErrorMessage) {
			if errMsg == nil {
				return
			}
			status := http.StatusInternalServerError
			if errMsg.StatusCode > 0 {
				status = errMsg.StatusCode
			}
			errText := http.StatusText(status)
			if errMsg.Error != nil && errMsg.Error.Error() != "" {
				errText = errMsg.Error.Error()
			}
			body := handlers.BuildErrorResponseBody(status, errText)
			_, _ = fmt.Fprintf(c.Writer, "\nevent: error\ndata: %s\n\n", string(body))
		},
		WriteDone: func() {
			_, _ = c.Writer.Write([]byte("\n"))
		},
	})
}