BufferedInputStream mark/reset issue

Question

Can someone help me figure out what I am doing wrong here? This method does nothing except produce empty lines when calling new InputStreamReader(stream, getSet(stream);

Thank you all!

private static final byte[] UTF8_BOM = new byte[] {(byte) 0xEF, (byte) 0xBB, (byte) 0xBF};

private static final byte[] UTF16LE_BOM = new byte[] {(byte) 0xFF, (byte) 0xFE};

private static final byte[] UTF16BE_BOM = new byte[] {(byte) 0xFE, (byte) 0xFF};

public static Charset getSet(final InputStream stream) throws IOException {

        final byte[] UTF8_Buffer = new byte[3];
        final byte[] UTF16LE_Buffer = new byte[2];
        final byte[] UTF16BE_Buffer = new byte[2];
        int byteValue;
        InputStream bufferedStream = new BufferedInputStream(stream);
        bufferedStream.mark(3);

        byteValue = stream.read();

        UTF16LE_Buffer[0] = (byte) byteValue;
        UTF16BE_Buffer[0] = (byte) byteValue;
        UTF8_Buffer[0] = (byte) byteValue;

        byteValue = bufferedStream.read();
        UTF16LE_Buffer[1] = (byte) byteValue;
        UTF16BE_Buffer[1] = (byte) byteValue;
        UTF8_Buffer[1] = (byte) byteValue;
        if (Arrays.equals(UTF16LE_Buffer, UTF16LE_BOM)) {
          return StandardCharsets.UTF_16LE;
        } else if (Arrays.equals(UTF16BE_Buffer, UTF16LE_BOM)) {
          return StandardCharsets.UTF_16BE;
        } else {
          byteValue = bufferedStream.read();
          UTF8_Buffer[2] = (byte) byteValue;
          if (Arrays.equals(UTF8_Buffer, UTF8_BOM)) {
            return StandardCharsets.UTF_8;
          }
        }

        bufferedStream.reset();
        return StandardCharsets.UTF_8;
      }

Sort of yes, I seem to be getting -1 on the first `bufferedStream.read()` call. Thanks to Albert! for formatted my code, I could not get everything into the code block — salvadore-jefferson, Nov 16 '15 at 22:01
I just ran it in debug again and I was wrong, I am getting values other then -1. — salvadore-jefferson, Nov 16 '15 at 22:04
can you please write the values here so that others can also see — Albert Pinto, Nov 16 '15 at 22:07
Sure, the first call returns: 224 the second returns 164 and the 3rd call returns 174 — salvadore-jefferson, Nov 16 '15 at 22:09
I should mention, I am calling this method on a text file with no BOM and the byte values: (byte) 0xE0, (byte) 0xA4, (byte) 0xAE, which is the Hindi symbol for "MA" — salvadore-jefferson, Nov 16 '15 at 22:12
Are you aware that 224 = 0xE0? 164 = 0xA4? 174 = 0xAE? What's the question here? — user207421, Nov 16 '15 at 22:20
The question is why doesn't my method return the correct `StandardCharsets`? — salvadore-jefferson, Nov 16 '15 at 22:24
Why should it? The input data doesn't match any of the prefixes you're looking for. — user207421, Nov 16 '15 at 22:27
Okay, so now we are getting somewhere. But my method returns `StandardCharsets.UTF_8;` if there is no match. At least that is what its supposed to do. — salvadore-jefferson, Nov 16 '15 at 22:30
So it's working correctly? It isn't working correctly? It's doing something else instead? What? And what on earth does 'this method does nothing except produce empty lines' mean, when it doesn't produce at lines at all? What's the question? — user207421, Nov 16 '15 at 22:33
This line produces the "MA" character: `try (final Reader reader = new InputStreamReader((stream), StandardCharsets.UTF_8))` This line does not: `try (final Reader reader = new InputStreamReader((stream), getSet(stream));` — salvadore-jefferson, Nov 16 '15 at 22:37
Thanks for the answer EJP. I will see about working this idea into the code. Your answer makes a lot of sense. — salvadore-jefferson, Nov 16 '15 at 22:55

user207421 · Answer 1 · 2015-11-16T23:48:09.273

1

The problem here is that you have already read two or three bytes from the input stream with the BufferedInputStream. Calling mark()/reset() on the BufferedInputStream allows you to reread those bytes from the BufferedInputStream, but it doesn't do anything to the underlying stream.

You should change the caller of this method to pass a BufferedInputStream instead of an InputStream, and remove the new BufferedInputStream from inside the method, so that the caller and the callee are both marking/resetting the same stream. If the caller doesn't pass an InputStream that supports mark()/reset(), calling either of those will throw an exception.

edited Nov 16 '15 at 23:48

answered Nov 16 '15 at 22:44

user207421

305,947
44
307
483

Is there any other solution? I am supposed to be programming to interfaces, or Abstract super classes in the case of InputStreams. Plus it doesn't look like BufferedInputStream supports file paths as input? – salvadore-jefferson Nov 16 '15 at 23:16
Well let me put it another way, see edit. You can do `new BufferedInputStream(new FileInputStream(...))`. – user207421 Nov 16 '15 at 23:48
This worked for the one aspect of my assignment, but it broke all my unit tests where I was calling the method with ByteArrayInputStream. I know I am on the wrong track when my existing unit test fail after refactoring. For the most part. But thanks for the tip, it did work how you said it would. – salvadore-jefferson Nov 17 '15 at 00:04
Why would it break your unit tests? `ByteArrayInputStream ` supports mark/reset. – user207421 Nov 17 '15 at 00:23
Because I changed the method signature from readText(InputStream) to readText(BufferedInputStream) So I could no longer call readText(ByteArrayInputStream) – salvadore-jefferson Nov 17 '15 at 00:26
So don't do that. I removed that part of my answer 3/4 of an hour ago. – user207421 Nov 17 '15 at 00:35

salvadore-jefferson · Answer 2 · 2015-11-17T20:57:52.800

-2

I found a solution to my issue.

I changed the method from this:

public static String readText(@Nonnull final InputStream stream) throws IOException {
     try (final Reader reader = new InputStreamReader(stream, getCharset(stream))) {....

To this:

public static String readText(@Nonnull final InputStream stream) throws   IOException {
    final BufferedInputStream bufferedStream = new BufferedInputStream(stream);
    try (final Reader reader = new InputStreamReader(bufferedStream, getCharset(bufferedStream))) {...

Thanks for all the help folks!

edited Nov 17 '15 at 20:57

answered Nov 17 '15 at 20:34

salvadore-jefferson

1
5

You didn't 'find' a solution to your issue. I provided it. And you should accept and/or upvote any answers you find useful. And this answer contains changes to code you've never posted before. Not useful, and somewhat irritating. – user207421 Nov 18 '15 at 01:08

BufferedInputStream mark/reset issue

2 Answers2