Receiving a null value in the content field

[dussab]

Describe the issue

I'm building a client application that calls through CodeGate and out to OpenAI. If I send a prompt through my client app with the text "Codegate version" a null is retuned. If send a prompt that isn't related to CodeGate (e.g. what is my name) i get a the following response.

On the failed (null) response i see the following json:

{
 "id": "pipeline-codegate-cli-1738105991",
 "created": 1738105991,
 "model": "gpt-4o-mini",
 "object": "chat.completion",
 "system_fingerprint": null,
 "choices": [
   {
     **"text": "CodeGate version: v0.1.14",**
     "finish_reason": "stop",
     "index": 0,
     "message": {
       **"content": null,**
       "role": "assistant",
       "tool_calls": null,
       "function_call": null
     }
   }
 ],
 "usage": {
   "completion_tokens": 0,
   "prompt_tokens": 0,
   "total_tokens": 0,
   "completion_tokens_details": null,
   "prompt_tokens_details": null
 }
}

It looks like the "text": "CodeGate version: v0.1.14", is returning the codegate version number but I'm wondering if it should be in the content field which is currently -> "content": null. It's possible that I need to handle this situation on the client and that this is by design.

Here is a successful response to compare against:

{
  "id": "chatcmpl-Aup48XBfAM4msvRZAI4lwDBPpZOxj",
  "object": "chat.completion",
  "created": 1738106028,
  "model": "gpt-4o-mini-2024-07-18",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        **"content": "It seems like you're testing the functionality. How can I assist you today? If you have any questions or need help with something specific, feel free to ask!",**
        "refusal": null
      },
      "logprobs": null,
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 74,
    "completion_tokens": 33,
    "total_tokens": 107,
    "prompt_tokens_details": {
      "cached_tokens": 0,
      "audio_tokens": 0
    },
    "completion_tokens_details": {
      "reasoning_tokens": 0,
      "audio_tokens": 0,
      "accepted_prediction_tokens": 0,
      "rejected_prediction_tokens": 0
    }
  },
  "service_tier": "default",
  "system_fingerprint": "fp_72ed7ab54c"
}

Steps to Reproduce

Run sample project https://github.com/StacklokLabs/spring-ai-codegate-sample/tree/main/sample

Operating System

MacOS (Arm)

IDE and Version

Vscode

Extension and Version

NA

Provider

OpenAI

Model

gpt-4o-mini

Codegate version

v0.1.14

Logs

Here are the CodeGate logs from the response returning null:

2025-01-28 15:00:15 2025-01-28T20:00:15.649744Z [debug    ] User-Agent header received: Java-http-client/23.0.1 from 172.17.0.1 lineno=45 module=server pathname=/app/src/codegate/server.py
2025-01-28 15:00:15 2025-01-28T20:00:15.655106Z [info     ] Chat completion pipeline selected for execution. lineno=120 module=base pathname=/app/src/codegate/providers/base.py
2025-01-28 15:00:15 2025-01-28T20:00:15.658325Z [info     ] Total secrets redacted since last assistant message: 0 lineno=298 module=secrets pathname=/app/src/codegate/pipeline/secrets/secrets.py
2025-01-28 15:00:15 2025-01-28T20:00:15.684066Z [info     ] Recorded context in DB. Output chunks: 1. Alerts: 0. lineno=249 module=connection pathname=/app/src/codegate/db/connection.py
2025-01-28 15:00:15 2025-01-28T20:00:15.684869Z [info     ] 172.17.0.1:61530 - "POST /openai/v1/chat/completions HTTP/1.1" 200 lineno=473 module=h11_impl pathname=/usr/local/lib/python3.12/site-packages/uvicorn/protocols/http/h11_impl.py

Additional Context

No response

[dussab]

I've also noticed that anytime i receive "content": null, , the prompt and responses don't get logged in the CodeGate portal.

[dussab]

Might be related to #542