fix: spoof ollama streaming responses for tool calls

src/proxy/forward.ts

@@ -2,6 +2,7 @@ import { config } from '../config';
 import { rewriteResponse } from './response-rewriter';
 import { normalizeRequest } from './request-normalizer';
 import { logger } from '../utils/logger';
+import { Readable } from 'stream';
 
 export async function forwardChatRequest(requestBody: any, authorization?: string): Promise<any> {
   const targetHost = config.targetUrl;
@@ -108,3 +109,75 @@ export async function forwardChatRequest(requestBody: any, authorization?: strin
 
   return rewrittenData;
 }
+
+export async function forwardAndSpoofOllamaStreamRequest(requestBody: any, authorization?: string): Promise<Readable> {
+  const upstreamBody = { ...requestBody, stream: false };
+  const rewrittenData = await forwardChatRequest(upstreamBody, authorization);
+
+  const sseStream = new Readable({
+    read() {}
+  });
+
+  const pushChunk = (delta: any, finishReason: string | null = null) => {
+    const chunk = {
+      id: rewrittenData.id || `chatcmpl-${Date.now()}`,
+      object: 'chat.completion.chunk',
+      created: Math.floor(Date.now() / 1000),
+      model: rewrittenData.model || upstreamBody.model,
+      choices: [
+        {
+          index: 0,
+          delta,
+          finish_reason: finishReason
+        }
+      ]
+    };
+
+    sseStream.push(`data: ${JSON.stringify(chunk)}\n\n`);
+  };
+
+  process.nextTick(() => {
+    try {
+      pushChunk({ role: 'assistant' });
+
+      const message = rewrittenData.message || {};
+
+      if (message.content) {
+        const chunkSize = 16;
+        for (let i = 0; i < message.content.length; i += chunkSize) {
+          pushChunk({ content: message.content.substring(i, i + chunkSize) });
+        }
+      }
+
+      if (message.tool_calls && message.tool_calls.length > 0) {
+        const streamToolCalls = message.tool_calls.map((tc: any, idx: number) => ({
+          index: idx,
+          id: tc.id,
+          type: tc.type,
+          function: {
+            name: tc.function.name,
+            arguments: JSON.stringify(tc.function.arguments ?? {})
+          }
+        }));
+
+        pushChunk({ tool_calls: streamToolCalls });
+      }
+
+      const finalFinishReason =
+        message.tool_calls?.length > 0
+          ? 'tool_calls'
+          : (rewrittenData.done_reason || 'stop');
+
+      pushChunk({}, finalFinishReason);
+      sseStream.push('data: [DONE]\n\n');
+      sseStream.push(null);
+    } catch (e: any) {
+      logger.error('Error generating Ollama SSE stream:', e.message);
+      sseStream.push('data: {"error":"Internal stream spoofing error"}\n\n');
+      sseStream.push('data: [DONE]\n\n');
+      sseStream.push(null);
+    }
+  });
+
+  return sseStream;
+}

src/routes/ollama.ts

@@ -1,5 +1,5 @@
 import { FastifyInstance, FastifyPluginAsync } from 'fastify';
-import { forwardChatRequest } from '../proxy/forward';
+import { forwardAndSpoofOllamaStreamRequest, forwardChatRequest } from '../proxy/forward';
 import { logger } from '../utils/logger';
 
 const ollamaRoutes: FastifyPluginAsync = async (server: FastifyInstance) => {
@@ -7,14 +7,14 @@ const ollamaRoutes: FastifyPluginAsync = async (server: FastifyInstance) => {
     try {
       const body = request.body as any;
 
-      // Currently only supporting non-streaming requests in this proxy MVP
       if (body?.stream === true) {
-        // As per requirements: return clear error or pass through without rewriting
-        // We'll return a clear error for now, because stream parsing is out of scope for MVP
-        reply.status(400).send({ 
-          error: "Streaming is not supported by this proxy MVP. Please set stream=false." 
-        });
-        return;
+        const spoofedStream = await forwardAndSpoofOllamaStreamRequest(body, request.headers.authorization);
+
+        reply.raw.setHeader('Content-Type', 'text/event-stream');
+        reply.raw.setHeader('Cache-Control', 'no-cache');
+        reply.raw.setHeader('Connection', 'keep-alive');
+
+        return reply.send(spoofedStream);
       }
 
       const response = await forwardChatRequest(body, request.headers.authorization);

test/integration.proxy.test.ts

@@ -62,20 +62,36 @@ describe('Proxy Integration Test', () => {
     });
   });
 
-  it('rejects streaming requests cleanly', async () => {
+  it('spoofs streaming responses for stream=true requests', async () => {
     const requestFixturePath = path.join(__dirname, 'fixtures', 'openclaw-like-request.json');
+    const responseFixturePath = path.join(__dirname, 'fixtures', 'ollama-xml-response.json');
     const requestJson = JSON.parse(fs.readFileSync(requestFixturePath, 'utf8'));
+    const responseJson = JSON.parse(fs.readFileSync(responseFixturePath, 'utf8'));
     requestJson.stream = true;
 
+    (global.fetch as any).mockResolvedValue({
+      ok: true,
+      text: async () => JSON.stringify(responseJson),
+      json: async () => responseJson
+    });
+
     const response = await server.inject({
       method: 'POST',
       url: '/api/chat',
       payload: requestJson
     });
 
-    expect(response.statusCode).toBe(400);
-    const body = JSON.parse(response.payload);
-    expect(body.error).toContain('Streaming is not supported');
-    expect(global.fetch).not.toHaveBeenCalled();
+    expect(response.statusCode).toBe(200);
+    expect(response.headers['content-type']).toContain('text/event-stream');
+    expect(global.fetch).toHaveBeenCalledTimes(1);
+
+    const fetchArgs = (global.fetch as any).mock.calls[0];
+    const upstreamBody = JSON.parse(fetchArgs[1].body);
+    expect(upstreamBody.stream).toBe(false);
+
+    expect(response.payload).toContain('"role":"assistant"');
+    expect(response.payload).toContain('"tool_calls"');
+    expect(response.payload).toContain('"finish_reason":"tool_calls"');
+    expect(response.payload).toContain('data: [DONE]');
   });
 });