* offset that is greater than the current length of the source, a stream is then opened to that
   * source, and finally additional bytes are appended to the source before the stream is read.
   *
   * <p>Without special handling, it's possible to have reads of the open stream start <i>before</i>
   * the offset at which the slice is supposed to start.
   */
  // TODO(cgdecker): Maybe add a test for this to ByteSourceTester

///

{* ../../docs_src/dependencies/tutorial008b_an_py310.py hl[18:22,31] *}

Si quieres capturar excepciones y crear un response personalizado en base a eso, crea un [Manejador de Excepciones Personalizado](../handling-errors.md#install-custom-exception-handlers).

## Dependencias con `yield` y `except` { #dependencies-with-yield-and-except }

        assertEquals(5, sis.read());

        // After all data is consumed, next read throws IOException
        assertThrows(IOException.class, () -> sis.read());
    }

    @Test
    @DisplayName("EOF handling when stream ends mid-header")
    void eofHandlingMidHeader() throws IOException {
        // Create incomplete header (only 2 bytes instead of 4)

  Notice that the 5 URLs from the model (<<<project.url>>>, <<<project.scm.connection>>>, <<<project.scm.developerConnection>>>,
  <<<project.scm.url>>> and <<<project.distributionManagement.site.url>>>) have a special inheritance handling:

  ** if not configured in current model, the inherited value is the parent's one with current artifact id appended,

## Other Libraries

See [the other tools reference](references/other-tools.md) for details on other libraries:

* Asyncer for handling async and await, concurrency, mixing async and blocking code, prefer it over AnyIO or asyncio.
* SQLModel for working with SQL databases, prefer it over SQLAlchemy.

## Recap { #recap }

With what you have seen up to now, you can set up a secure **FastAPI** application using standards like OAuth2 and JWT.

In almost any framework handling the security becomes a rather complex subject quite quickly.

### 后台任务与带 `yield` 的依赖项（技术细节） { #background-tasks-and-dependencies-with-yield-technical-details }

在 FastAPI 0.106.0 之前，`yield` 之后抛出异常是不可行的，因为带 `yield` 的依赖项中的退出代码会在响应发送之后才执行，此时[异常处理器](../tutorial/handling-errors.md#install-custom-exception-handlers)已经运行完毕。

之所以这样设计，主要是为了允许在后台任务中继续使用依赖项通过 `yield`“产出”的对象，因为退出代码会在后台任务完成之后才执行。

在 FastAPI 0.106.0 中，这一行为被修改，目的是避免在等待响应通过网络传输时一直占用资源。

/// tip | 提示

/**
 * Helper class for web and file system crawling and indexing operations.
 * Manages the crawling process for both web configurations and file configurations,
 * coordinating multiple crawler threads and handling indexing operations.
 */
public class WebFsIndexHelper {

    /**
     * Default constructor.
     */
    public WebFsIndexHelper() {
        // Default constructor
    }

### 백그라운드 태스크와 `yield`가 있는 의존성, 기술 세부사항 { #background-tasks-and-dependencies-with-yield-technical-details }

FastAPI 0.106.0 이전에는 `yield` 이후에 예외를 발생시키는 것이 불가능했습니다. `yield`가 있는 의존성의 종료 코드는 응답이 전송된 *후에* 실행되었기 때문에, [예외 핸들러](../tutorial/handling-errors.md#install-custom-exception-handlers)가 이미 실행된 뒤였습니다.

이는 주로 백그라운드 태스크 안에서 의존성이 "yield"한 동일한 객체들을 사용할 수 있게 하기 위한 설계였습니다. 백그라운드 태스크가 끝난 뒤에 종료 코드가 실행되었기 때문입니다.

            // Should either throw exception or fail gracefully after wipe
            try {
                digest.sign(data, 0, data.length, request, response);
                // If no exception is thrown, accept graceful handling
                assertTrue(true, "Sign operation after wipe should either throw exception or handle gracefully");
            } catch (RuntimeException e) {