myData := []byte("this-is-a-checksum-data-test")
	chksm := NewChecksumFromData(ChecksumCRC32, myData)
	if chksm == nil {
		t.Fatal("NewChecksumFromData returned nil")
	}
	// Serialize the checksum to bytes
	b := chksm.AppendTo(nil, nil)
	if b == nil {
		t.Fatal("AppendTo returned nil")
	}

	// Deserialize the checksum from bytes
	chksmOut := ChecksumFromBytes(b)

          "alg":"RS256",
          "kid":"2011-04-29"}
       ]
     }`

	pubKeys := publicKeys{
		RWMutex: &sync.RWMutex{},
		pkMap:   map[string]any{},
	}
	err := pubKeys.parseAndAdd(bytes.NewBuffer([]byte(jsonkey)))
	if err != nil {
		t.Fatal("Error loading pubkeys:", err)
	}
	if len(pubKeys.pkMap) != 1 {
		t.Fatalf("Expected 1 keys, got %d", len(pubKeys.pkMap))
	}

    void testReadBytesWireFormatValidStructure() throws SMBProtocolDecodingException {
        // Given
        byte[] buffer = new byte[256];
        int offset = 10;

        // Write valid structure (16 bytes)
        SMBUtil.writeInt2(16, buffer, offset); // Structure size
        buffer[offset + 2] = (byte) 0x01; // Share type (DISK)
        buffer[offset + 3] = 0; // Reserved

それは `openapi_extra` で行えます。

{* ../../docs_src/path_operation_advanced_configuration/tutorial006_py310.py hl[19:36, 39:40] *}

この例では、Pydanticモデルを一切宣言していません。実際、リクエストボディはJSONとして <dfn title="bytes などのプレーンな形式から Python オブジェクトに変換される">パース</dfn> されず、直接 `bytes` として読み取られます。そして `magic_data_reader()` 関数が、何らかの方法でそれをパースする責務を担います。

それでも、リクエストボディに期待されるスキーマを宣言できます。

### カスタムOpenAPI content type { #custom-openapi-content-type }

      sink.flush()
    }
  }

  /** The maximum size of bytes that may be sent in a single call to [data]. */
  fun maxDataLength(): Int = maxFrameSize

  /**
   * `source.length` may be longer than the max length of the variant's data frame. Implementations
   * must send multiple frames as necessary.
   *
   * @param source the buffer to draw bytes from. May be null if byteCount is 0.

        int[][] testCases = { { 0, 50 }, // No offset, 50 bytes
                { 10, 90 }, // 10 byte offset, 90 bytes
                { 50, 50 }, // 50 byte offset, 50 bytes
                { 0, 0 }, // Empty buffer
                { 100, 100 } // Large values
        };

        for (int[] testCase : testCases) {
            // Arrange
            byte[] buffer = new byte[200];
            int bufferIndex = testCase[0];

    return asByteSource(url).asCharSource(charset);
  }

  /**
   * Reads all bytes from a URL into a byte array.
   *
   * @param url the URL to read from
   * @return a byte array containing all the bytes from the URL
   * @throws IOException if an I/O error occurs
   */
  public static byte[] toByteArray(URL url) throws IOException {
    return asByteSource(url).read();
  }

  /**

En este ejemplo, no declaramos ningún modelo Pydantic. De hecho, el request body ni siquiera es <dfn title="convertido desde algún formato plano, como bytes, a objetos de Python">parseado</dfn> como JSON, se lee directamente como `bytes`, y la función `magic_data_reader()` sería la encargada de hacer parse de él de alguna manera.

Sin embargo, podemos declarar el esquema esperado para el request body.

        writeLittleEndianInt(baos, PacSignature.KERB_CHECKSUM_HMAC_MD5);
        byte[] checksum = new byte[16];
        for (int i = 0; i < checksum.length; i++) {
            checksum[i] = (byte) i;
        }
        baos.write(checksum);
        byte[] data = baos.toByteArray();

        // Create PacSignature
        PacSignature pacSignature = new PacSignature(data);

 */
public final class NtlmChallenge implements Serializable {

    /**
     * The NTLM challenge bytes received from the server.
     */
    public byte[] challenge;
    /**
     * The domain controller address that issued the challenge.
     */
    public UniAddress dc;

    NtlmChallenge(final byte[] challenge, final UniAddress dc) {
        this.challenge = challenge;
        this.dc = dc;
    }