feat: add OpenAI-compatible HTTP handler with parameter translation in server.py

- Inserted `print("AAAAA")` at the start of `http_handler` for debugging request flow
- Added `import logging` and `logging.basicConfig(level=logging.DEBUG)` before app creation
- Changed `app = web.Application()` to `app = web.Application(debug=True)` to enable debug mode
- Added blank line after `except ValueError:` block for code formatting consistency

server.py

@@ -157,6 +157,7 @@ async def websocket_handler(request):
     return ws
 
 async def http_handler(request):
+    print("AAAAA")
     request_id = str(uuid.uuid4())
     data = None
     try:
@@ -183,6 +184,75 @@ async def http_handler(request):
     await resp.write_eof()
     return resp
 
+async def openai_http_handler(request):
+    request_id = str(uuid.uuid4())
+    openai_request_data = None
+    try:
+        openai_request_data = await request.json()
+    except ValueError:
+        return web.Response(status=400)
+
+    # Translate OpenAI request to Ollama request
+    ollama_request_data = {
+        "model": openai_request_data.get("model"),
+        "messages": openai_request_data.get("messages"),
+        "stream": openai_request_data.get("stream", False),
+        "options": {}
+    }
+
+    if "temperature" in openai_request_data:
+        ollama_request_data["options"]["temperature"] = openai_request_data["temperature"]
+    if "max_tokens" in openai_request_data:
+        ollama_request_data["options"]["num_predict"] = openai_request_data["max_tokens"]
+    # Add more OpenAI to Ollama parameter mappings here as needed
+
+    resp = web.StreamResponse(headers={'Content-Type': 'text/event-stream', 'Cache-Control': 'no-cache', 'Transfer-Encoding': 'chunked'})
+    await resp.prepare(request)
+    import json
+    try:
+        completion_id = "chatcmpl-" + str(uuid.uuid4()).replace("-", "")[:24] # Generate a unique ID
+        created_time = int(asyncio.get_event_loop().time()) # Unix timestamp
+
+        async for ollama_result in server_manager.forward_to_websocket(request_id, ollama_request_data, path="/api/chat"):
+            openai_response_chunk = {
+                "id": completion_id,
+                "object": "chat.completion.chunk",
+                "created": created_time,
+                "model": ollama_result.get("model", openai_request_data.get("model")),
+                "choices": [
+                    {
+                        "index": 0,
+                        "delta": {},
+                        "logprobs": None,
+                        "finish_reason": None
+                    }
+                ]
+            }
+
+            if "message" in ollama_result and "content" in ollama_result["message"]:
+                openai_response_chunk["choices"][0]["delta"]["content"] = ollama_result["message"]["content"]
+            elif "role" in ollama_result["message"]: # First chunk might have role
+                 openai_response_chunk["choices"][0]["delta"]["role"] = ollama_result["message"]["role"]
+
+
+            if ollama_result.get("done"):
+                openai_response_chunk["choices"][0]["finish_reason"] = "stop" # Or "length", etc.
+                await resp.write(f"data: {json.dumps(openai_response_chunk)}\n\n".encode())
+                await resp.write(b"data: [DONE]\n\n")
+            else:
+                await resp.write(f"data: {json.dumps(openai_response_chunk)}\n\n".encode())
+
+    except NoServerFoundException:
+        await resp.write(f"data: {json.dumps(dict(error='No server with that model found.', available=server_manager.get_models()))}\n\n".encode())
+        await resp.write(b"data: [DONE]\n\n")
+    except Exception as e:
+        logging.error(f"Error in openai_http_handler: {e}")
+        await resp.write(f"data: {json.dumps(dict(error=str(e)))}\n\n".encode())
+        await resp.write(b"data: [DONE]\n\n")
+
+    await resp.write_eof()
+    return resp
+
 async def index_handler(request):
     index_template = pathlib.Path("index.html").read_text()
     client_py = pathlib.Path("client.py").read_text()
@@ -198,7 +268,12 @@ async def models_handler(self):
     response_json = json.dumps(server_manager.get_models(),indent=2)
     return web.Response(text=response_json,content_type="application/json")
 
-app = web.Application()
+import logging
+logging.basicConfig(
+    level=logging.DEBUG
+)
+
+app = web.Application(debug=True)
 
 app.router.add_get("/", index_handler)
 app.router.add_route('GET', '/publish', websocket_handler)
@@ -206,6 +281,7 @@ app.router.add_route('POST', '/api/chat', http_handler)
 app.router.add_route('POST', '/v1/chat', http_handler)
 app.router.add_route('POST', '/v1/completions', http_handler)
 app.router.add_route('POST', '/v1/chat/completions', http_handler)
+app.router.add_route('POST', '/v1/chat/completions', openai_http_handler)
 app.router.add_route('GET', '/models', models_handler)
 app.router.add_route('GET', '/v1/models', models_handler)
 app.router.add_route('*', '/{tail:.*}', not_found_handler)

test.py

@@ -1,52 +1,87 @@
-from ollama import Client
-client = Client(
-  host='https://ollama.molodetz.nl/',
-  headers={'x-some-header': 'some-value'}
-)
+import asyncio
+import aiohttp
+import json
+import uuid
 
-def times_two(nr_1: int) -> int:
-    return nr_1 * 2
+# Configuration for the local server
+LOCAL_SERVER_URL = "http://localhost:1984"
 
-available_functions = {
-    'times_two': times_two
-}
+async def test_openai_chat_completions():
+    print("--- Starting OpenAI Chat Completions Test ---")
+    session = aiohttp.ClientSession()
 
-messages = []
+    try:
+        # OpenAI-style request payload
+        payload = {
+            "model": "qwen2.5:3b", # Assuming this model is available on an Ollama instance connected to the server
+            "messages": [
+                {"role": "user", "content": "Tell me a short story about a brave knight."}
+            ],
+            "stream": True,
+            "temperature": 0.7,
+            "max_tokens": 50
+        }
 
+        openai_endpoint = f"{LOCAL_SERVER_URL}/v1/chat/completions"
+        print(f"Sending request to: {openai_endpoint}")
+        print(f"Payload: {json.dumps(payload, indent=2)}")
 
-def chat_stream(message):
-    if message:
-        messages.append({'role': 'user', 'content': message})
-    content = ''
-    for response in client.chat(model='qwen2.5-coder:0.5b', messages=messages, stream=True):
-        content += response.message.content
-        print(response.message.content, end='', flush=True)
-    messages.append({'role': 'assistant', 'content': content})
-    print("")
+        async with session.post(openai_endpoint, json=payload) as response:
+            print(f"Response status: {response.status}")
+            assert response.status == 200, f"Expected status 200, got {response.status}"
 
+            response_data = ""
+            async for chunk in response.content.iter_any():
+                chunk_str = chunk.decode('utf-8')
+                # print(f"Received chunk: {chunk_str.strip()}")
                 
-def chat(message, stream=False):
-    if stream:
-        return chat_stream(message)
-    if message:
-        messages.append({'role': 'user', 'content': message})
-    response = client.chat(model='qwen2.5:3b', messages=messages,
-    tools=[times_two])
-    if response.message.tool_calls:
-        for tool in response.message.tool_calls:
-            if function_to_call := available_functions.get(tool.function.name):
-                print('Calling function:', tool.function.name)
-                print('Arguments:', tool.function.arguments)
-                output = function_to_call(**tool.function.arguments)
-                print('Function output:', output)
-            else:
-                print('Function', tool.function.name, 'not found')
+                # Split by 'data: ' and process each JSON object
+                for line in chunk_str.splitlines():
+                    if line.startswith("data: "):
+                        json_str = line[len("data: "):].strip()
+                        if json_str == "[DONE]":
+                            print("Received [DONE] signal.")
+                            continue
 
-            if response.message.tool_calls:
-                messages.append(response.message)
-                messages.append({'role': 'tool', 'content': str(output), 'name': tool.function.name})
-                return chat(None)
-    return response.message.content
+                        try:
+                            data = json.loads(json_str)
+                            # Basic assertions for OpenAI streaming format
+                            assert "id" in data
+                            assert "object" in data
+                            assert data["object"] == "chat.completion.chunk"
+                            assert "choices" in data
+                            assert isinstance(data["choices"], list)
+                            assert len(data["choices"]) > 0
+                            assert "delta" in data["choices"][0]
                             
-while True:
-    chat_stream("A farmer and a sheep are standing on one side of a river. There is a boat with enough room for one human and one animal. How can the farmer get across the river with the sheep in the fewest number of trips?")
+                            if "content" in data["choices"][0]["delta"]:
+                                response_data += data["choices"][0]["delta"]["content"]
+
+                        except json.JSONDecodeError as e:
+                            print(f"JSON Decode Error: {e} in chunk: {json_str}")
+                            assert False, f"Invalid JSON received: {json_str}"
+                    elif line.strip(): # Handle potential non-data lines, though unlikely for SSE
+                        print(f"Non-data line received: {line.strip()}")
+
+            print(f"\nFull response content:\n{response_data}")
+            assert len(response_data) > 0, "No content received in OpenAI response."
+            print("--- OpenAI Chat Completions Test Passed ---")
+
+    except aiohttp.ClientError as e:
+        print(f"Client error during test: {e}")
+        assert False, f"Client error: {e}"
+    except AssertionError as e:
+        print(f"Assertion failed: {e}")
+        assert False, f"Test failed: {e}"
+    except Exception as e:
+        print(f"An unexpected error occurred during test: {e}")
+        assert False, f"Unexpected error: {e}"
+    finally:
+        await session.close()
+        print("--- OpenAI Chat Completions Test Finished ---")
+
+async def main():
+    await test_openai_chat_completions()
+
+if __name__ == '__main__':
+    asyncio.run(main())