为什么InputStream.seek导致乱码内容？

Question

我必须部分地读取资源，然后稍后（在非常不同的时间点），我需要跳过我最初读取的字节数。 我需要粘上我读过的两个部分。

这是我的代码的一个非常简单的说明。 我正在使用单独的ByteArrayInputStream -s，因为如上所述，调用将彼此完全无关（几乎可以肯定，不使用相同的InputStream ）。

我不太清楚这里出了什么问题。 而不是获得带有值的串联String This is a big fat super long text has no meaning, but is good for the test. ，我得到了： This is a big fat super long text has no meaning, but is good for the test.aning, but is good fo 。 This is a big fat super long text has no meaning, but is good for the test.aning, but is good fo 。

public void testFoo()
{
    String s = "This is a big fat super long text has no meaning, but is good for the test.";

    ByteArrayInputStream bais1 = new ByteArrayInputStream(s.getBytes());
    ByteArrayInputStream bais2 = new ByteArrayInputStream(s.getBytes());

    ByteArrayOutputStream baos1 = new ByteArrayOutputStream();
    ByteArrayOutputStream baos2 = new ByteArrayOutputStream();

    int size = 32;
    byte[] bytes = new byte[size];

    int total = 0;
    int len;

    while ((len = bais1.read(bytes, 0, size)) != -1)
    {
        baos1.write(bytes);
        baos1.flush();

        total += len;
        if (total >= size)
        {
            // This is here just to illustrate that
            // we are reading a few bytes, and then
            // just terminating before the rest of
            // the stream has been read.
            break;
        }
    }

    bytes = new byte[size];
    bais1.close();

    System.out.println("Read " + total + " bytes.");

    // Here we are supposed to skip the number
    // of bytes that have already been read:
    bais2.skip(total);

    System.out.println("Skipped " + total + "/" + s.getBytes().length + " bytes.");

    while ((len = bais2.read(bytes, 0, size)) != -1)
    {
        baos2.write(bytes);
        baos2.flush();

        total += len;
    }

    System.out.println("Original:      " + s);
    System.out.println("Partial read1: " + new String(baos1.toByteArray()));
    System.out.println("Partial read2: " + new String(baos2.toByteArray()));

    System.out.println("Read " + total + " bytes.");
}

有人可以指出什么是错的以及如何解决它？ 我不太明白为什么在第二次读取过程中跳过字节数后，读取数据的结尾会搞砸。 请指教！

Answer 1

在第二个读取循环中，您忽略了读取的字节数，因此缓冲区中已有的字节将附加到输出流中。

尝试将其更改为：

  System.out.println("Skipped " + total + "/" + s.getBytes().length + " bytes.");

  while ((len = bais2.read(bytes, 0, size)) != -1)
  {
    baos2.write(bytes, 0, len); // <- added write offset and length
    baos2.flush();

    total += len;
  }