Pydantic and JSON schemas need upgrade

[HamzaYslmn]

Confirm this is a feature request for the Python library and not the underlying OpenAI API.

• This is a feature request for the Python library

Describe the feature or improvement you're requesting

When JSON schemas become overly complex and nested, pydantic comes to the rescue. It converts them into a more readable structure, but I provide a separate schema for each user in the API. This has made management quite challenging.

I want a step towards simplifying a global JSON struct or its transformations and being able to save it to the database.

Or at least

let us use
Schema.model_json_schema()

Additional context

If I don't know a method, please tell me what you do.

[RobertCraigie]

Sorry I'm not quite sure what you're asking, can you expand?

[HamzaYslmn]

Sorry I'm not quite sure what you're asking, can you expand?

for example, lets handle this basic situation:

{
  "name": "math_response",
  "schema": {
    "type": "object",
    "properties": {
      "steps": {
        "type": "array",
        "description": "A list of step objects that show the process of calculation.",
        "items": {
          "type": "object",
          "properties": {
            "explanation": {
              "type": "string",
              "description": "A description of the reasoning or operation performed in this step."
            },
            "output": {
              "type": "string",
              "description": "The result or output after this step."
            }
          },
          "required": [
            "explanation",
            "output"
          ],
          "additionalProperties": false
        }
      },
      "final_answer": {
        "type": "string",
        "description": "The final answer derived from all steps."
      }
    },
    "required": [
      "steps",
      "final_answer"
    ],
    "additionalProperties": false
  },
  "strict": true
}

Pydantic version

from pydantic import BaseModel
from typing import List

class Step(BaseModel):
    explanation: str
    output: str

class MathResponse(BaseModel):
    steps: List[Step]
    final_answer: str

print(MathResponse.model_json_schema())

Output Python:

{'$defs': {'Step': {'properties': {'explanation': {'title': 'Explanation', 'type': 'string'}, 'output': {'title': 'Output', 'type': 'string'}}, 'required': ['explanation', 'output'], 'title': 'Step', 'type': 'object'}}, 'properties': {'steps': {'items': {'$ref': '#/$defs/Step'}, 'title': 'Steps', 'type': 'array'}, 'final_answer': {'title': 'Final Answer', 'type': 'string'}}, 'required': ['steps', 'final_answer'], 'title': 'MathResponse', 'type': 'object'}

Since I can't save the Pydantic object to the database, I have to convert it to JSON. (I provide a separate schema for each user, that's why.)

Therefore, I have to manually prepare a schema for each user. But since these schemas have about 350 keys and are nested, the code is very difficult to read and edit.

That's why I want to write it with Pydantic and convert it to JSON. (There's already a function somewhere inside that does this; we can use it with the “parse” method.)

[HamzaYslmn]

@RobertCraigie I hope that's clear enough.

[HamzaYslmn]

#2093 (comment)

Found solution but this is not best solve.

[HamzaYslmn]

from openai.resources.responses.responses import _type_to_text_format_param

gives better schema for responses api