feat: 支持 openai provider 通过 OPENAI_WIRE_API 切换到 responses API

Author: vpen66

src/services/api/openai/__tests__/responses.test.ts

@@ -0,0 +1,242 @@
+import { afterEach, describe, expect, test } from 'bun:test'
+import type { ResponseStreamEvent } from 'openai/resources/responses/responses.mjs'
+import {
+  adaptResponsesStreamToAnthropic,
+  buildOpenAIResponsesRequestBody,
+  resolveOpenAIWireAPI,
+} from '../responses.js'
+
+const originalWireAPI = process.env.OPENAI_WIRE_API
+
+afterEach(() => {
+  if (originalWireAPI === undefined) {
+    delete process.env.OPENAI_WIRE_API
+  } else {
+    process.env.OPENAI_WIRE_API = originalWireAPI
+  }
+})
+
+async function collectAdaptedEvents(events: ResponseStreamEvent[]) {
+  async function* stream() {
+    for (const event of events) {
+      yield event
+    }
+  }
+
+  const result = []
+  for await (const event of adaptResponsesStreamToAnthropic(
+    stream() as any,
+    'test-model',
+  )) {
+    result.push(event)
+  }
+  return result
+}
+
+describe('resolveOpenAIWireAPI', () => {
+  test('defaults to chat completions', () => {
+    delete process.env.OPENAI_WIRE_API
+    expect(resolveOpenAIWireAPI()).toBe('chat_completions')
+  })
+
+  test('accepts responses env override', () => {
+    process.env.OPENAI_WIRE_API = 'responses'
+    expect(resolveOpenAIWireAPI()).toBe('responses')
+  })
+})
+
+describe('buildOpenAIResponsesRequestBody', () => {
+  test('converts messages, tools, and tool choice', () => {
+    const body = buildOpenAIResponsesRequestBody({
+      model: 'gpt-test',
+      messages: [
+        {
+          type: 'user',
+          message: { content: 'hello' },
+        },
+        {
+          type: 'assistant',
+          message: {
+            content: [
+              {
+                type: 'tool_use',
+                id: 'toolu_123',
+                name: 'bash',
+                input: { command: 'ls' },
+              },
+            ],
+          },
+        },
+        {
+          type: 'user',
+          message: {
+            content: [
+              {
+                type: 'tool_result',
+                tool_use_id: 'toolu_123',
+                content: 'ok',
+              },
+              {
+                type: 'text',
+                text: 'next',
+              },
+            ],
+          },
+        },
+      ] as any,
+      systemPrompt: ['system prompt'] as any,
+      tools: [
+        {
+          type: 'custom',
+          name: 'bash',
+          description: 'Run shell commands',
+          input_schema: {
+            type: 'object',
+            properties: {
+              command: { const: 'ls' },
+            },
+          },
+          strict: true,
+        },
+      ] as any,
+      toolChoice: { type: 'tool', name: 'bash' },
+      enableThinking: false,
+      maxTokens: 4096,
+      temperatureOverride: 0.2,
+    })
+
+    expect(body.instructions).toBe('system prompt')
+    expect(body.max_output_tokens).toBe(4096)
+    expect(body.tool_choice).toEqual({ type: 'function', name: 'bash' })
+    expect(body.tools).toEqual([
+      {
+        type: 'function',
+        name: 'bash',
+        description: 'Run shell commands',
+        parameters: {
+          type: 'object',
+          properties: {
+            command: { enum: ['ls'] },
+          },
+        },
+        strict: true,
+      },
+    ])
+    expect(body.input).toEqual([
+      {
+        type: 'message',
+        role: 'user',
+        content: [{ type: 'input_text', text: 'hello' }],
+      },
+      {
+        type: 'function_call',
+        call_id: 'toolu_123',
+        name: 'bash',
+        arguments: '{"command":"ls"}',
+      },
+      {
+        type: 'function_call_output',
+        call_id: 'toolu_123',
+        output: 'ok',
+      },
+      {
+        type: 'message',
+        role: 'user',
+        content: [{ type: 'input_text', text: 'next' }],
+      },
+    ])
+  })
+})
+
+describe('adaptResponsesStreamToAnthropic', () => {
+  test('maps streamed function calls and terminal usage', async () => {
+    const events = await collectAdaptedEvents([
+      {
+        type: 'response.created',
+        sequence_number: 1,
+        response: {
+          id: 'resp_1',
+          object: 'response',
+          created_at: 0,
+          model: 'test-model',
+          output: [],
+          output_text: '',
+          tools: [],
+          tool_choice: 'auto',
+          parallel_tool_calls: false,
+          temperature: null,
+          top_p: null,
+          error: null,
+          incomplete_details: null,
+          instructions: null,
+          metadata: null,
+          usage: null,
+        },
+      } as any,
+      {
+        type: 'response.output_item.added',
+        sequence_number: 2,
+        output_index: 0,
+        item: {
+          type: 'function_call',
+          id: 'fc_1',
+          call_id: 'toolu_123',
+          name: 'bash',
+          arguments: '',
+          status: 'in_progress',
+        },
+      } as any,
+      {
+        type: 'response.function_call_arguments.delta',
+        sequence_number: 3,
+        output_index: 0,
+        item_id: 'fc_1',
+        delta: '{"command":"ls"}',
+      } as any,
+      {
+        type: 'response.completed',
+        sequence_number: 4,
+        response: {
+          usage: {
+            input_tokens: 11,
+            output_tokens: 7,
+            total_tokens: 18,
+            input_tokens_details: { cached_tokens: 2 },
+            output_tokens_details: { reasoning_tokens: 0 },
+          },
+        },
+      } as any,
+    ])
+
+    expect(events).toEqual([
+      expect.objectContaining({ type: 'message_start' }),
+      expect.objectContaining({
+        type: 'content_block_start',
+        content_block: expect.objectContaining({
+          type: 'tool_use',
+          id: 'toolu_123',
+          name: 'bash',
+        }),
+      }),
+      expect.objectContaining({
+        type: 'content_block_delta',
+        delta: {
+          type: 'input_json_delta',
+          partial_json: '{"command":"ls"}',
+        },
+      }),
+      expect.objectContaining({ type: 'content_block_stop' }),
+      expect.objectContaining({
+        type: 'message_delta',
+        delta: { stop_reason: 'tool_use', stop_sequence: null },
+        usage: {
+          input_tokens: 11,
+          output_tokens: 7,
+          cache_creation_input_tokens: 0,
+          cache_read_input_tokens: 2,
+        },
+      }),
+      expect.objectContaining({ type: 'message_stop' }),
+    ])
+  })
+})

src/services/api/openai/index.ts

@@ -24,6 +24,11 @@ import { logForDebugging } from '../../../utils/debug.js'
 import { addToTotalSessionCost } from '../../../cost-tracker.js'
 import { calculateUSDCost } from '../../../utils/modelCost.js'
 import { isOpenAIThinkingEnabled, resolveOpenAIMaxTokens, buildOpenAIRequestBody } from './requestBody.js'
+import {
+  adaptResponsesStreamToAnthropic,
+  buildOpenAIResponsesRequestBody,
+  resolveOpenAIWireAPI,
+} from './responses.js'
 import { recordLLMObservation } from '../../../services/langfuse/tracing.js'
 import { convertMessagesToLangfuse, convertOutputToLangfuse, convertToolsToLangfuse } from '../../../services/langfuse/convert.js'
 export { isOpenAIThinkingEnabled, resolveOpenAIMaxTokens, buildOpenAIRequestBody }
@@ -223,29 +228,45 @@ export async function* queryModelOpenAI(
       source: options.querySource,
     })
 
+    const wireAPI = resolveOpenAIWireAPI()
     logForDebugging(
-      `[OpenAI] Calling model=${openaiModel}, messages=${openaiMessages.length}, tools=${openaiTools.length}, thinking=${enableThinking}`,
+      `[OpenAI] Calling model=${openaiModel}, wire_api=${wireAPI}, messages=${openaiMessages.length}, tools=${openaiTools.length}, thinking=${enableThinking}`,
     )
 
     // 12. Call OpenAI API with streaming
-    const requestBody = buildOpenAIRequestBody({
-      model: openaiModel,
-      messages: openaiMessages,
-      tools: openaiTools,
-      toolChoice: openaiToolChoice,
-      enableThinking,
-      maxTokens,
-      temperatureOverride: options.temperatureOverride,
-    })
-    const stream = await client.chat.completions.create(
-      requestBody,
-      { signal },
-    )
+    let adaptedStream: AsyncIterable<any>
+    if (wireAPI === 'responses') {
+      const requestBody = buildOpenAIResponsesRequestBody({
+        model: openaiModel,
+        messages: messagesForAPI,
+        systemPrompt,
+        tools: standardTools,
+        toolChoice: options.toolChoice,
+        enableThinking,
+        maxTokens,
+        temperatureOverride: options.temperatureOverride,
+      })
+      const stream = await client.responses.create(requestBody, { signal })
+      adaptedStream = adaptResponsesStreamToAnthropic(stream, openaiModel)
+    } else {
+      const requestBody = buildOpenAIRequestBody({
+        model: openaiModel,
+        messages: openaiMessages,
+        tools: openaiTools,
+        toolChoice: openaiToolChoice,
+        enableThinking,
+        maxTokens,
+        temperatureOverride: options.temperatureOverride,
+      })
+      const stream = await client.chat.completions.create(
+        requestBody,
+        { signal },
+      )
+      adaptedStream = adaptOpenAIStreamToAnthropic(stream, openaiModel)
+    }
 
     // 12. Convert OpenAI stream to Anthropic events, then process into
     //     AssistantMessage + StreamEvent (matching the Anthropic path behavior)
-    const adaptedStream = adaptOpenAIStreamToAnthropic(stream, openaiModel)
-
     // Accumulate content blocks and usage, same as the Anthropic path in claude.ts
     const contentBlocks: Record<number, any> = {}
     const collectedMessages: AssistantMessage[] = []