4. The daemon runs the request, sending back data such as logging output or tooling API events and intermediate models while doing so.
5. The daemon sends the result back. For some requests, this might be a simple success/failure result, and for others this might also include a more complex object, such as a tooling API model.

自從 Pydantic v2 發佈後，生成的 OpenAPI 比以往更精確也更正確。😎

實際上，在某些情況下，同一個 Pydantic 模型在 OpenAPI 中會同時有兩個 JSON Schema：分別用於輸入與輸出，這取決於它是否有預設值。

來看看它如何運作，以及若需要時該如何調整。

## 作為輸入與輸出的 Pydantic 模型 { #pydantic-models-for-input-and-output }

假設你有一個帶有預設值的 Pydantic 模型，如下所示：

{* ../../docs_src/separate_openapi_schemas/tutorial001_py310.py ln[1:7] hl[7] *}

### 輸入用模型 { #model-for-input }

如果你把這個模型用作輸入，如下所示：

自从发布了 **Pydantic v2**，生成的 OpenAPI 比之前更精确、更**正确**了。😎

事实上，在某些情况下，对于同一个 Pydantic 模型，OpenAPI 中会根据是否带有**默认值**，为输入和输出分别生成**两个 JSON Schema**。

我们来看看它如何工作，以及在需要时如何修改。

## 用于输入和输出的 Pydantic 模型 { #pydantic-models-for-input-and-output }

假设你有一个带有默认值的 Pydantic 模型，例如：

{* ../../docs_src/separate_openapi_schemas/tutorial001_py310.py ln[1:7] hl[7] *}

### 输入用的模型 { #model-for-input }

如果你像下面这样把该模型用作输入：

Bunun nasıl çalıştığına ve gerekirse nasıl değiştirebileceğinize bir bakalım.

## Input ve Output için Pydantic Modelleri { #pydantic-models-for-input-and-output }

Default değerleri olan bir Pydantic modeliniz olduğunu varsayalım; örneğin şöyle:

{* ../../docs_src/separate_openapi_schemas/tutorial001_py310.py ln[1:7] hl[7] *}

Розгляньмо, як це працює, і як це змінити за потреби.

## Моделі Pydantic для введення та виведення { #pydantic-models-for-input-and-output }

Припустімо, у вас є модель Pydantic зі значеннями за замовчуванням, наприклад:

{* ../../docs_src/separate_openapi_schemas/tutorial001_py310.py ln[1:7] hl[7] *}

ml_models = {}


@asynccontextmanager
async def lifespan(app: FastAPI):
    # Load the ML model
    ml_models["answer_to_everything"] = fake_answer_to_everything_ml_model
    yield
    # Clean up the ML models and release the resources
    ml_models.clear()


app = FastAPI(lifespan=lifespan)


@app.get("/predict")
async def predict(x: float):
    result = ml_models["answer_to_everything"](x)

 * in compliance with, at your election, the Elastic License 2.0 or the Server
 * Side Public License, v 1.
 */

package org.elasticsearch.gradle.internal;

/**
 * This class models the different Docker base images that are used to build Docker distributions of Elasticsearch.
 */
public enum DockerBase {
    CENTOS("centos:8", ""),

    // "latest" here is intentional, since the image name specifies "8"

import org.apache.maven.api.Service;
import org.apache.maven.api.annotations.Experimental;
import org.apache.maven.api.annotations.Nonnull;
import org.apache.maven.api.model.Model;

/**
 * Provides the super POM that all models implicitly inherit from.
 *
 * @since 4.0.0
 */
@Experimental
public interface SuperPomProvider extends Service {

    /**
     * Gets the super POM for the specified model version.
     *

import pytest
from fastapi.openapi.models import Schema, SchemaType


@pytest.mark.parametrize(
    "type_value",
    [
        "array",
        ["string", "null"],
        None,
    ],
)
def test_allowed_schema_type(
    type_value: SchemaType | list[SchemaType] | None,
) -> None:
    """Test that Schema accepts SchemaType, List[SchemaType] and None for type field."""
    schema = Schema(type=type_value)

Import `BaseSettings` from Pydantic and create a sub-class, very much like with a Pydantic model.

The same way as with Pydantic models, you declare class attributes with type annotations, and possibly default values.

You can use all the same validation features and tools you use for Pydantic models, like different data types and additional validations with `Field()`.

{* ../../docs_src/settings/tutorial001_py310.py hl[2,5:8,11] *}