Files
CLIProxyAPI/sdk/api/handlers/claude/code_handlers.go
Ben Vargas 9ad0f3f91e feat: Add Amp CLI integration with comprehensive documentation
Add full Amp CLI support to enable routing AI model requests through the proxy
while maintaining Amp-specific features like thread management, user info, and
telemetry. Includes complete documentation and pull bot configuration.

Features:
- Modular architecture with RouteModule interface for clean integration
- Reverse proxy for Amp management routes (thread/user/meta/ads/telemetry)
- Provider-specific route aliases (/api/provider/{provider}/*)
- Secret management with precedence: config > env > file
- 5-minute secret caching to reduce file I/O
- Automatic gzip decompression for responses
- Proper connection cleanup to prevent leaks
- Localhost-only restriction for management routes (configurable)
- CORS protection for management endpoints

Documentation:
- Complete setup guide (USING_WITH_FACTORY_AND_AMP.md)
- OAuth setup for OpenAI (ChatGPT Plus/Pro) and Anthropic (Claude Pro/Max)
- Factory CLI config examples with all model variants
- Amp CLI/IDE configuration examples
- tmux setup for remote server deployment
- Screenshots and diagrams

Configuration:
- Pull bot disabled for this repo (manual rebase workflow)
- Config fields: AmpUpstreamURL, AmpUpstreamAPIKey, AmpRestrictManagementToLocalhost
- Compatible with upstream DisableCooling and other features

Technical details:
- internal/api/modules/amp/: Complete Amp routing module
- sdk/api/httpx/: HTTP utilities for gzip/transport
- 94.6% test coverage with 34 comprehensive test cases
- Clean integration minimizes merge conflict risk

Security:
- Management routes restricted to localhost by default
- Configurable via amp-restrict-management-to-localhost
- Prevents drive-by browser attacks on user data

This provides a production-ready foundation for Amp CLI integration while
maintaining clean separation from upstream code for easy rebasing.

Amp-Thread-ID: https://ampcode.com/threads/T-9e2befc5-f969-41c6-890c-5b779d58cf18
2025-11-19 18:23:17 -07:00

311 lines
10 KiB
Go

// Package claude provides HTTP handlers for Claude API code-related functionality.
// This package implements Claude-compatible streaming chat completions with sophisticated
// client rotation and quota management systems to ensure high availability and optimal
// resource utilization across multiple backend clients. It handles request translation
// between Claude API format and the underlying Gemini backend, providing seamless
// API compatibility while maintaining robust error handling and connection management.
package claude
import (
"bufio"
"bytes"
"compress/gzip"
"context"
"encoding/json"
"fmt"
"io"
"net/http"
"time"
"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"
"github.com/router-for-me/CLIProxyAPI/v6/sdk/api/handlers"
log "github.com/sirupsen/logrus"
"github.com/tidwall/gjson"
)
// ClaudeCodeAPIHandler contains the handlers for Claude API endpoints.
// It holds a pool of clients to interact with the backend service.
type ClaudeCodeAPIHandler struct {
*handlers.BaseAPIHandler
}
// NewClaudeCodeAPIHandler creates a new Claude API handlers instance.
// It takes an BaseAPIHandler instance as input and returns a ClaudeCodeAPIHandler.
//
// Parameters:
// - apiHandlers: The base API handler instance.
//
// Returns:
// - *ClaudeCodeAPIHandler: A new Claude code API handler instance.
func NewClaudeCodeAPIHandler(apiHandlers *handlers.BaseAPIHandler) *ClaudeCodeAPIHandler {
return &ClaudeCodeAPIHandler{
BaseAPIHandler: apiHandlers,
}
}
// HandlerType returns the identifier for this handler implementation.
func (h *ClaudeCodeAPIHandler) HandlerType() string {
return Claude
}
// Models returns a list of models supported by this handler.
func (h *ClaudeCodeAPIHandler) Models() []map[string]any {
// Get dynamic models from the global registry
modelRegistry := registry.GetGlobalRegistry()
return modelRegistry.GetAvailableModels("claude")
}
// ClaudeMessages handles Claude-compatible streaming chat completions.
// This function implements a sophisticated client rotation and quota management system
// to ensure high availability and optimal resource utilization across multiple backend clients.
//
// Parameters:
// - c: The Gin context for the request.
func (h *ClaudeCodeAPIHandler) ClaudeMessages(c *gin.Context) {
// Extract raw JSON data from the incoming request
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")
if !streamResult.Exists() || streamResult.Type == gjson.False {
h.handleNonStreamingResponse(c, rawJSON)
} else {
h.handleStreamingResponse(c, rawJSON)
}
}
// ClaudeMessages handles Claude-compatible streaming chat completions.
// This function implements a sophisticated client rotation and quota management system
// to ensure high availability and optimal resource utilization across multiple backend clients.
//
// Parameters:
// - c: The Gin context for the request.
func (h *ClaudeCodeAPIHandler) ClaudeCountTokens(c *gin.Context) {
// Extract raw JSON data from the incoming request
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
}
c.Header("Content-Type", "application/json")
alt := h.GetAlt(c)
cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
modelName := gjson.GetBytes(rawJSON, "model").String()
resp, errMsg := h.ExecuteCountWithAuthManager(cliCtx, h.HandlerType(), modelName, rawJSON, alt)
if errMsg != nil {
h.WriteErrorResponse(c, errMsg)
cliCancel(errMsg.Error)
return
}
_, _ = c.Writer.Write(resp)
cliCancel()
}
// ClaudeModels handles the Claude models listing endpoint.
// It returns a JSON response containing available Claude models and their specifications.
//
// Parameters:
// - c: The Gin context for the request.
func (h *ClaudeCodeAPIHandler) ClaudeModels(c *gin.Context) {
c.JSON(http.StatusOK, gin.H{
"data": h.Models(),
})
}
// handleNonStreamingResponse handles non-streaming content generation requests for Claude models.
// This function processes the request synchronously and returns the complete generated
// response in a single API call. It supports various generation parameters and
// response formats.
//
// Parameters:
// - c: The Gin context for the request
// - modelName: The name of the Gemini model to use for content generation
// - rawJSON: The raw JSON request body containing generation parameters and content
func (h *ClaudeCodeAPIHandler) handleNonStreamingResponse(c *gin.Context, rawJSON []byte) {
c.Header("Content-Type", "application/json")
alt := h.GetAlt(c)
cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
modelName := gjson.GetBytes(rawJSON, "model").String()
resp, errMsg := h.ExecuteWithAuthManager(cliCtx, h.HandlerType(), modelName, rawJSON, alt)
if errMsg != nil {
h.WriteErrorResponse(c, errMsg)
cliCancel(errMsg.Error)
return
}
// Decompress gzipped responses - Claude API sometimes returns gzip without Content-Encoding header
// This fixes title generation and other non-streaming responses that arrive compressed
if len(resp) >= 2 && resp[0] == 0x1f && resp[1] == 0x8b {
gzReader, err := gzip.NewReader(bytes.NewReader(resp))
if err != nil {
log.Warnf("failed to decompress gzipped Claude response: %v", err)
} else {
defer gzReader.Close()
if decompressed, err := io.ReadAll(gzReader); err != nil {
log.Warnf("failed to read decompressed Claude response: %v", err)
} else {
resp = decompressed
}
}
}
_, _ = c.Writer.Write(resp)
cliCancel()
}
// handleStreamingResponse streams Claude-compatible responses backed by Gemini.
// It sets up SSE, selects a backend client with rotation/quota logic,
// forwards chunks, and translates them to Claude CLI format.
//
// Parameters:
// - c: The Gin context for the request.
// - rawJSON: The raw JSON request body.
func (h *ClaudeCodeAPIHandler) handleStreamingResponse(c *gin.Context, rawJSON []byte) {
// Set up Server-Sent Events (SSE) headers for streaming response
// These headers are essential for maintaining a persistent connection
// and enabling real-time streaming of chat completions
c.Header("Content-Type", "text/event-stream")
c.Header("Cache-Control", "no-cache")
c.Header("Connection", "keep-alive")
c.Header("Access-Control-Allow-Origin", "*")
// Get the http.Flusher interface to manually flush the response.
// This is crucial for streaming as it allows immediate sending of data chunks
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(rawJSON, "model").String()
// Create a cancellable context for the backend client request
// This allows proper cleanup and cancellation of ongoing requests
cliCtx, cliCancel := h.GetContextWithCancel(h, c, context.Background())
dataChan, errChan := h.ExecuteStreamWithAuthManager(cliCtx, h.HandlerType(), modelName, rawJSON, "")
h.forwardClaudeStream(c, flusher, func(err error) { cliCancel(err) }, dataChan, errChan)
return
}
func (h *ClaudeCodeAPIHandler) forwardClaudeStream(c *gin.Context, flusher http.Flusher, cancel func(error), data <-chan []byte, errs <-chan *interfaces.ErrorMessage) {
// v6.1: Intelligent Buffered Streamer strategy
// Enhanced buffering with larger buffer size (16KB) and longer flush interval (120ms).
// Smart flush only when buffer is sufficiently filled (≥50%), dramatically reducing
// flush frequency from ~12.5Hz to ~5-8Hz while maintaining low latency.
writer := bufio.NewWriterSize(c.Writer, 16*1024) // 4KB → 16KB
ticker := time.NewTicker(120 * time.Millisecond) // 80ms → 120ms
defer ticker.Stop()
var chunkIdx int
for {
select {
case <-c.Request.Context().Done():
// Context cancelled, flush any remaining data before exit
_ = writer.Flush()
cancel(c.Request.Context().Err())
return
case <-ticker.C:
// Smart flush: only flush when buffer has sufficient data (≥50% full)
// This reduces flush frequency while ensuring data flows naturally
buffered := writer.Buffered()
if buffered >= 8*1024 { // At least 8KB (50% of 16KB buffer)
if err := writer.Flush(); err != nil {
// Error flushing, cancel and return
cancel(err)
return
}
flusher.Flush() // Also flush the underlying http.ResponseWriter
}
case chunk, ok := <-data:
if !ok {
// Stream ended, flush remaining data
_ = writer.Flush()
cancel(nil)
return
}
// Forward the complete SSE event block directly (already formatted by the translator).
// The translator returns a complete SSE-compliant event block, including event:, data:, and separators.
// The handler just needs to forward it without reassembly.
if len(chunk) > 0 {
_, _ = writer.Write(chunk)
}
chunkIdx++
case errMsg, ok := <-errs:
if !ok {
continue
}
if errMsg != nil {
// An error occurred: emit as a proper SSE error event
errorBytes, _ := json.Marshal(h.toClaudeError(errMsg))
_, _ = writer.WriteString("event: error\n")
_, _ = writer.WriteString("data: ")
_, _ = writer.Write(errorBytes)
_, _ = writer.WriteString("\n\n")
_ = writer.Flush()
}
var execErr error
if errMsg != nil {
execErr = errMsg.Error
}
cancel(execErr)
return
}
}
}
type claudeErrorDetail struct {
Type string `json:"type"`
Message string `json:"message"`
}
type claudeErrorResponse struct {
Type string `json:"type"`
Error claudeErrorDetail `json:"error"`
}
func (h *ClaudeCodeAPIHandler) toClaudeError(msg *interfaces.ErrorMessage) claudeErrorResponse {
return claudeErrorResponse{
Type: "error",
Error: claudeErrorDetail{
Type: "api_error",
Message: msg.Error.Error(),
},
}
}