```Python
q: Union[str, None] = None
```

Ou bien, en Python 3.10 et supérieur :

```Python
q: str | None = None
```

Par exemple :

{* ../../docs_src/body_multiple_params/tutorial004_an_py310.py hl[27] *}

/// info

`@lru_cache` es parte de `functools`, que es parte del paquete estándar de Python, puedes leer más sobre él en las <a href="https://docs.python.org/3/library/functools.html#functools.lru_cache" class="external-link" target="_blank">docs de Python para `@lru_cache`</a>.

## Resumen { #recap }

load(
    "//tensorflow/python/tools/api/generator:api_gen.bzl",
    "get_compat_files",
    "get_nested_compat_files",
)
load(
    "//tensorflow/python/tools/api/generator:api_init_files.bzl",
    "TENSORFLOW_API_INIT_FILES",
)
load(
    "//tensorflow/python/tools/api/generator:api_init_files_v1.bzl",
    "TENSORFLOW_API_INIT_FILES_V1",
)

Um Formulare zu verwenden, installieren Sie zuerst <a href="https://github.com/Kludex/python-multipart" class="external-link" target="_blank">`python-multipart`</a>.

Stellen Sie sicher, dass Sie eine [virtuelle Umgebung](../virtual-environments.md){.internal-link target=_blank} erstellen, sie aktivieren und es dann installieren, zum Beispiel:

```console
$ pip install python-multipart
```

///

/// note | Hinweis

這一點，再加上 Python 是 **資料科學**、機器學習，尤其是深度學習的主要語言，讓 **FastAPI** 成為資料科學/機器學習 Web API 和應用程式（以及許多其他應用程式）的絕佳選擇。

想了解如何在生產環境中實現這種平行性，請參見 [部屬](deployment/index.md){.internal-link target=_blank}。

## `async` 和 `await`

現代 Python 版本提供一種非常直觀的方式定義非同步程式碼。這使得它看起來就像正常的「順序」程式碼，並在適當的時機「等待」。

當某個操作需要等待才能回傳結果，並且支援這些新的 Python 特性時，你可以像這樣編寫程式碼：

```Python
burgers = await get_burgers(2)
```

/// note | Hinweis

FastAPI weiß, dass der Wert von `q` nicht erforderlich ist, aufgrund des definierten Defaultwertes `= None`.

Das `str | None` (Python 3.10+) oder `Union` in `Union[str, None]` (Python 3.9+) wird von FastAPI nicht verwendet, um zu bestimmen, dass der Wert nicht erforderlich ist. FastAPI weiß, dass er nicht erforderlich ist, weil er einen Defaultwert von `= None` hat.

    ],
)

# -----------------------------------------------------------------------------
# Python API target

tf_cuda_library(
    name = "python_api",
    srcs = ["python_api.cc"],
    hdrs = ["python_api.h"],
    visibility = [
        "//tensorflow:internal",
        "//tensorflow/python:__pkg__",
    ],
    deps = [
        ":c_api",
        ":c_api_internal",

## `Depends()`

Dependencies are handled mainly with the special function `Depends()` that takes a callable.

Here is the reference for it and its parameters.

You can import it directly from `fastapi`:

```python
from fastapi import Depends
```

::: fastapi.Depends

## `Security()`

For many scenarios, you can handle security (authorization, authentication, etc.) with dependencies, using `Depends()`.

When you raise an exception, as would happen with normal Python, the rest of the execution is aborted. This way you can raise these exceptions from anywhere in the code to abort a request and show the error to the client.

You can use:

* `HTTPException`
* `WebSocketException`

These exceptions can be imported directly from `fastapi`:

```python
from fastapi import HTTPException, WebSocketException
```

commands.

*Nightly binaries are available for testing using the
[tf-nightly](https://pypi.python.org/pypi/tf-nightly) and
[tf-nightly-cpu](https://pypi.python.org/pypi/tf-nightly-cpu) packages on PyPI.*

#### *Try your first TensorFlow program*

```shell
$ python
```

```python
>>> import tensorflow as tf
>>> tf.add(1, 2).numpy()
3
>>> hello = tf.constant('Hello, TensorFlow!')