Merge branch 'feat/proxy-api-server' into refactor/api-gateway

src/main/apiServer/routes/chat.ts

@@ -3,7 +3,7 @@ import express from 'express'
 
 import { loggerService } from '../../services/LoggerService'
 import type { ExtendedChatCompletionCreateParams } from '../adapters'
-import { generateMessage, streamToResponse } from '../services/ProxyStreamService'
+import { processMessage } from '../services/ProxyStreamService'
 import { validateModelId } from '../utils'
 
 const logger = loggerService.withContext('ApiServerChatRoutes')
@@ -205,38 +205,15 @@ router.post('/completions', async (req: Request, res: Response) => {
 
     const provider = modelValidation.provider!
     const modelId = modelValidation.modelId!
-    const isStreaming = !!request.stream
 
-    if (isStreaming) {
-      try {
-        await streamToResponse({
-          response: res,
-          provider,
-          modelId,
-          params: request,
-          inputFormat: 'openai',
-          outputFormat: 'openai'
-        })
-      } catch (streamError) {
-        logger.error('Stream error', { error: streamError })
-        // If headers weren't sent yet, return JSON error
-        if (!res.headersSent) {
-          const { status, body } = mapChatCompletionError(streamError)
-          return res.status(status).json(body)
-        }
-        // Otherwise the error is already handled by streamToResponse
-      }
-      return
-    }
-
-    const response = await generateMessage({
+    return processMessage({
+      response: res,
       provider,
       modelId,
       params: request,
       inputFormat: 'openai',
       outputFormat: 'openai'
     })
-    return res.json(response)
   } catch (error: unknown) {
     const { status, body } = mapChatCompletionError(error)
     return res.status(status).json(body)

src/main/apiServer/routes/messages.ts

@@ -8,7 +8,7 @@ import express from 'express'
 import { approximateTokenSize } from 'tokenx'
 
 import { messagesService } from '../services/messages'
-import { generateMessage, streamToResponse } from '../services/ProxyStreamService'
+import { processMessage } from '../services/ProxyStreamService'
 import { getProviderById, isModelAnthropicCompatible, validateModelId } from '../utils'
 
 /**
@@ -321,29 +321,19 @@ async function handleUnifiedProcessing({
       providerId: provider.id
     })
 
-    if (request.stream) {
-      await streamToResponse({
-        response: res,
-        provider,
-        modelId: actualModelId,
-        params: request,
-        middlewares,
-        onError: (error) => {
-          logger.error('Stream error', error as Error)
-        },
-        onComplete: () => {
-          logger.debug('Stream completed')
-        }
-      })
-    } else {
-      const response = await generateMessage({
-        provider,
-        modelId: actualModelId,
-        params: request,
-        middlewares
-      })
-      res.json(response)
-    }
+    await processMessage({
+      response: res,
+      provider,
+      modelId: actualModelId,
+      params: request,
+      middlewares,
+      onError: (error) => {
+        logger.error('Message error', error as Error)
+      },
+      onComplete: () => {
+        logger.debug('Message completed')
+      }
+    })
   } catch (error: any) {
     const { statusCode, errorResponse } = messagesService.transformError(error)
     res.status(statusCode).json(errorResponse)

src/main/apiServer/services/ProxyStreamService.ts

@@ -33,6 +33,8 @@ import type { Response } from 'express'
 
 import type { InputFormat, InputParamsMap, IStreamAdapter } from '../adapters'
 import { MessageConverterFactory, type OutputFormat, StreamAdapterFactory } from '../adapters'
+import { LONG_POLL_TIMEOUT_MS } from '../config/timeouts'
+import { createStreamAbortController } from '../utils/createStreamAbortController'
 import { googleReasoningCache, openRouterReasoningCache } from './reasoning-cache'
 
 const logger = loggerService.withContext('ProxyStreamService')
@@ -42,10 +44,6 @@ initializeSharedProviders({
   error: (message, error) => logger.error(message, error)
 })
 
-// ============================================================================
-// Configuration Interfaces
-// ============================================================================
-
 /**
  * Middleware type alias
  */
@@ -57,9 +55,9 @@ type LanguageModelMiddleware = LanguageModelV2Middleware
 type InputParams = InputParamsMap[InputFormat]
 
 /**
- * Configuration for streaming message requests
+ * Configuration for message requests (both streaming and non-streaming)
  */
-export interface StreamConfig {
+export interface MessageConfig {
   response: Response
   provider: Provider
   modelId: string
@@ -72,19 +70,6 @@ export interface StreamConfig {
   plugins?: AiPlugin[]
 }
 
-/**
- * Configuration for non-streaming message generation
- */
-export interface GenerateConfig {
-  provider: Provider
-  modelId: string
-  params: InputParams
-  inputFormat?: InputFormat
-  outputFormat?: OutputFormat
-  middlewares?: LanguageModelMiddleware[]
-  plugins?: AiPlugin[]
-}
-
 /**
  * Internal configuration for stream execution
  */
@@ -96,6 +81,7 @@ interface ExecuteStreamConfig {
   outputFormat: OutputFormat
   middlewares?: LanguageModelMiddleware[]
   plugins?: AiPlugin[]
+  abortSignal?: AbortSignal
 }
 
 // ============================================================================
@@ -248,7 +234,7 @@ async function executeStream(config: ExecuteStreamConfig): Promise<{
   adapter: IStreamAdapter
   outputStream: ReadableStream<unknown>
 }> {
-  const { provider, modelId, params, inputFormat, outputFormat, middlewares = [], plugins = [] } = config
+  const { provider, modelId, params, inputFormat, outputFormat, middlewares = [], plugins = [], abortSignal } = config
 
   // Convert provider config to AI SDK config
   let sdkConfig = providerToAiSdkConfig(provider, modelId)
@@ -291,7 +277,8 @@ async function executeStream(config: ExecuteStreamConfig): Promise<{
     stopWhen: stepCountIs(100),
     headers: defaultAppHeaders(),
     tools,
-    providerOptions
+    providerOptions,
+    abortSignal
   })
 
   // Transform stream using adapter
@@ -300,27 +287,14 @@ async function executeStream(config: ExecuteStreamConfig): Promise<{
   return { adapter, outputStream }
 }
 
-// ============================================================================
-// Public API
-// ============================================================================
-
 /**
- * Stream a message request and write to HTTP response
+ * Process a message request - handles both streaming and non-streaming
  *
- * Uses TransformStream-based adapters for efficient streaming.
- *
- * @example
- * ```typescript
- * await streamToResponse({
- *   response: res,
- *   provider,
- *   modelId: 'claude-3-opus',
- *   params: messageCreateParams,
- *   outputFormat: 'anthropic'
- * })
- * ```
+ * Automatically detects streaming mode from params.stream:
+ * - stream=true: SSE streaming response
+ * - stream=false: JSON response
  */
-export async function streamToResponse(config: StreamConfig): Promise<void> {
+export async function processMessage(config: MessageConfig): Promise<void> {
   const {
     response,
     provider,
@@ -334,7 +308,9 @@ export async function streamToResponse(config: StreamConfig): Promise<void> {
     plugins = []
   } = config
 
-  logger.info('Starting proxy stream', {
+  const isStreaming = 'stream' in params && params.stream === true
+
+  logger.info(`Starting ${isStreaming ? 'streaming' : 'non-streaming'} message`, {
     providerId: provider.id,
     providerType: provider.type,
     modelId,
@@ -344,90 +320,21 @@ export async function streamToResponse(config: StreamConfig): Promise<void> {
     pluginCount: plugins.length
   })
 
-  try {
-    // Set SSE headers
-    response.setHeader('Content-Type', 'text/event-stream')
-    response.setHeader('Cache-Control', 'no-cache')
-    response.setHeader('Connection', 'keep-alive')
-    response.setHeader('X-Accel-Buffering', 'no')
+  // Create abort controller with timeout
+  const streamController = createStreamAbortController({ timeoutMs: LONG_POLL_TIMEOUT_MS })
+  const { abortController, dispose } = streamController
 
-    const { outputStream } = await executeStream({
-      provider,
-      modelId,
-      params,
-      inputFormat,
-      outputFormat,
-      middlewares,
-      plugins
-    })
-
-    // Get formatter for the output format
-    const formatter = StreamAdapterFactory.getFormatter(outputFormat)
-
-    // Stream events to response
-    const reader = outputStream.getReader()
-    try {
-      while (true) {
-        const { done, value } = await reader.read()
-        if (done) break
-        response.write(formatter.formatEvent(value))
-      }
-    } finally {
-      reader.releaseLock()
-    }
-
-    // Send done marker and end response
-    response.write(formatter.formatDone())
-    response.end()
-
-    logger.info('Proxy stream completed', { providerId: provider.id, modelId })
-    onComplete?.()
-  } catch (error) {
-    logger.error('Error in proxy stream', error as Error, { providerId: provider.id, modelId })
-    onError?.(error)
-    throw error
+  const handleDisconnect = () => {
+    if (abortController.signal.aborted) return
+    logger.info('Client disconnected, aborting', { providerId: provider.id, modelId })
+    abortController.abort('Client disconnected')
   }
-}
 
-/**
- * Generate a non-streaming message response
- *
- * Uses simulateStreamingMiddleware to reuse the same streaming logic.
- *
- * @example
- * ```typescript
- * const message = await generateMessage({
- *   provider,
- *   modelId: 'claude-3-opus',
- *   params: messageCreateParams,
- *   outputFormat: 'anthropic'
- * })
- * ```
- */
-export async function generateMessage(config: GenerateConfig): Promise<unknown> {
-  const {
-    provider,
-    modelId,
-    params,
-    inputFormat = 'anthropic',
-    outputFormat = 'anthropic',
-    middlewares = [],
-    plugins = []
-  } = config
-
-  logger.info('Starting message generation', {
-    providerId: provider.id,
-    providerType: provider.type,
-    modelId,
-    inputFormat,
-    outputFormat,
-    middlewareCount: middlewares.length,
-    pluginCount: plugins.length
-  })
+  response.on('close', handleDisconnect)
 
   try {
-    // Add simulateStreamingMiddleware to reuse streaming logic
-    const allMiddlewares = [simulateStreamingMiddleware(), ...middlewares]
+    // For non-streaming, add simulateStreamingMiddleware
+    const allMiddlewares = isStreaming ? middlewares : [simulateStreamingMiddleware(), ...middlewares]
 
     const { adapter, outputStream } = await executeStream({
       provider,
@@ -436,30 +343,60 @@ export async function generateMessage(config: GenerateConfig): Promise<unknown>
       inputFormat,
       outputFormat,
       middlewares: allMiddlewares,
-      plugins
+      plugins,
+      abortSignal: abortController.signal
     })
 
-    // Consume the stream to populate adapter state
-    const reader = outputStream.getReader()
-    while (true) {
-      const { done } = await reader.read()
-      if (done) break
+    if (isStreaming) {
+      // Streaming: Set SSE headers and stream events
+      response.setHeader('Content-Type', 'text/event-stream')
+      response.setHeader('Cache-Control', 'no-cache')
+      response.setHeader('Connection', 'keep-alive')
+      response.setHeader('X-Accel-Buffering', 'no')
+
+      const formatter = StreamAdapterFactory.getFormatter(outputFormat)
+      const reader = outputStream.getReader()
+
+      try {
+        while (true) {
+          const { done, value } = await reader.read()
+          if (done) break
+          if (response.writableEnded) break
+          response.write(formatter.formatEvent(value))
+        }
+      } finally {
+        reader.releaseLock()
+      }
+
+      if (!response.writableEnded) {
+        response.write(formatter.formatDone())
+        response.end()
+      }
+    } else {
+      // Non-streaming: Consume stream and return JSON
+      const reader = outputStream.getReader()
+      while (true) {
+        const { done } = await reader.read()
+        if (done) break
+      }
+      reader.releaseLock()
+
+      const finalResponse = adapter.buildNonStreamingResponse()
+      response.json(finalResponse)
     }
-    reader.releaseLock()
 
-    // Build final response from adapter
-    const finalResponse = adapter.buildNonStreamingResponse()
-
-    logger.info('Message generation completed', { providerId: provider.id, modelId })
-
-    return finalResponse
+    logger.info('Message completed', { providerId: provider.id, modelId, streaming: isStreaming })
+    onComplete?.()
   } catch (error) {
-    logger.error('Error in message generation', error as Error, { providerId: provider.id, modelId })
+    logger.error('Error in message processing', error as Error, { providerId: provider.id, modelId })
+    onError?.(error)
     throw error
+  } finally {
+    response.off('close', handleDisconnect)
+    dispose()
   }
 }
 
 export default {
-  streamToResponse,
-  generateMessage
+  processMessage
 }

src/main/apiServer/services/models.ts

@@ -15,11 +15,6 @@ export class ModelsService {
 
       const providers = await getAvailableProviders()
 
-      // Note: When providerType === 'anthropic', we now return ALL available models
-      // because the API Server's unified adapter (AiSdkToAnthropicSSE) can convert
-      // any provider's response to Anthropic SSE format. This enables Claude Code Agent
-      // to work with OpenAI, Gemini, and other providers transparently.
-
       const models = await listAllAvailableModels(providers)
       // Use Map to deduplicate models by their full ID (provider:model_id)
       const uniqueModels = new Map<string, ApiModel>()