Ici, nous utilisons [`time.perf_counter()`](https://docs.python.org/3/library/time.html#time.perf_counter) au lieu de `time.time()` car cela peut être plus précis pour ces cas d’usage. 🤓

///

## Ordre d’exécution de plusieurs middlewares { #multiple-middleware-execution-order }

///

## Ausführungsreihenfolge bei mehreren Middlewares { #multiple-middleware-execution-order }

                }
            }
        }

        return merged.toArray(new String[merged.size()]);
    }

    /**
     * Merges two string arrays, maintaining uniqueness and order.
     */
    private static String[] mergeStringArrays(final String[] array1, final String[] array2) {
        final Set<String> merged = new LinkedHashSet<>();

        if (array1 != null) {

# Header Parametre Modelleri { #header-parameter-models }

Birbiriyle ilişkili **header parametreleri**nden oluşan bir grubunuz varsa, bunları tanımlamak için bir **Pydantic model** oluşturabilirsiniz.

Bu sayede modeli **birden fazla yerde yeniden kullanabilir**, ayrıca tüm parametreler için doğrulamaları ve metadata bilgilerini tek seferde tanımlayabilirsiniz. 😎

/// note | Not

Bu özellik FastAPI `0.115.0` sürümünden beri desteklenmektedir. 🤓

///

   *
   * <p>Although all tasks are immediately executed in the thread that submitted the task, this
   * {@code ExecutorService} imposes a small locking overhead on each task submission in order to
   * implement shutdown and termination behavior.
   *
   * <p>Because of the nature of single-thread execution, the methods {@code scheduleAtFixedRate}

/// tip

這裡我們使用 [`time.perf_counter()`](https://docs.python.org/3/library/time.html#time.perf_counter) 而不是 `time.time()`，因為在這些用例中它可能更精確。🤓

///

## 多個中介軟體的執行順序 { #multiple-middleware-execution-order }

當你使用 `@app.middleware()` 裝飾器或 `app.add_middleware()` 方法加入多個中介軟體時，每個新的中介軟體都會包裹應用程式，形成一個堆疊。最後加入的中介軟體位於最外層，最先加入的位於最內層。

在請求路徑上，最外層的中介軟體最先執行。

在回應路徑上，它最後執行。

例如：

```Python

 *
 * <p>The factory maintains a map of regular expression patterns to crawler clients,
 * allowing for URL-based client selection. Clients can be added with specific patterns
 * and optionally at specific positions in the processing order.</p>
 *
 * <p>This factory is typically initialized through dependency injection and can be
 * configured with initialization parameters that are passed to all registered clients.</p>
 *
 * <p>Features:</p>
 * <ul>

     */
    public String jobLogging;

    /**
     * Whether the scheduled job is available.
     */
    public String available;

    /**
     * Sort order of the scheduled job.
     */
    public String sortOrder;

    /**
     * The user who created the scheduled job.
     */
    public String createdBy;

    /**

      h1 = fmix64(h1);
      h2 = fmix64(h2);

      h1 += h2;
      h2 += h1;

      return HashCode.fromBytesNoCopy(
          ByteBuffer.wrap(new byte[CHUNK_SIZE])
              .order(ByteOrder.LITTLE_ENDIAN)
              .putLong(h1)
              .putLong(h2)
              .array());
    }

    private static long fmix64(long k) {
      k ^= k >>> 33;
      k *= 0xff51afd7ed558ccdL;

import java.util.Arrays;
import org.jspecify.annotations.NullMarked;
import org.jspecify.annotations.Nullable;

/**
 * {@code ObjectCountLinkedHashMap} is a subclass of {@code ObjectCountHashMap} with insertion
 * iteration order, and uses arrays to store key objects and count values. Comparing to using a
 * traditional {@code LinkedHashMap} implementation which stores keys and count values as map