Wrong concatenation of textwrap.dedent string when using custom system message

[JohanBekker]

• This is actually a bug report.
• I am not getting good LLM Results
• I have tried asking for help in the community on discord or discussions and have not received a response.
• I have tried searching the documentation and have not found an answer.

What Model are you using?

• gpt-3.5-turbo
• gpt-4-turbo
• gpt-4
• Other (please specify)

Describe the bug
A clear and concise description of what the bug is.

When using a custom system message, the concatenated Instructor system message, wrapped in textwrap.dedent, has wrong indentation:

You are a helpful assistant.


        As a genius expert, your task is to understand the content and provide
        the parsed objects in json that match the following json_schema:


        {
  "properties": {
    "name": {
      "title": "Name",
      "type": "string"
    },
    "age": {
      "title": "Age",
      "type": "integer"
    }
  },
  "required": [
    "name",
    "age"
  ],
  "title": "UserInfo",
  "type": "object"
}

        Make sure to return an instance of the JSON, not the schema itself

To Reproduce

import instructor
from dotenv import load_dotenv
from langfuse.openai import OpenAI
from pydantic import BaseModel

load_dotenv()


class UserInfo(BaseModel):
    name: str
    age: int


llm = OpenAI()
client = instructor.from_openai(llm, mode=instructor.Mode.MD_JSON)

system_message = "You are a helpful assistant."
user_message = "John Doe is 30 years old."

user_info = client.chat.completions.create(
    model="gpt-4o-mini",
    max_tokens=4000,
    temperature=0,
    response_model=UserInfo,
    messages=[
        {"role": "system", "content": system_message},
        {"role": "user", "content": user_message},
    ],
)

Expected behavior
I'm not sure how much it matters for performance though but it would be nice if the prompts are all neatly formatted.

[JohanBekker]

It's actually not just with a custom system message, but always:

[
0: {
role: "system"
content: "
        As a genius expert, your task is to understand the content and provide
        the parsed objects in json that match the following json_schema:


        {
  "properties": {
    "name": {
      "title": "Name",
      "type": "string"
    },
    "age": {
      "title": "Age",
      "type": "integer"
    }
  },
  "required": [
    "name",
    "age"
  ],
  "title": "UserInfo",
  "type": "object"
}

        Make sure to return an instance of the JSON, not the schema itself
"
}
1: {
role: "user"
content: "John Doe is 30 years old.

Return the correct JSON response within a ```json codeblock. not the JSON_SCHEMA"
}
]