Python Struct and reading values in Java

Question

I want to export a binary format and then I read the binary in Java, But i am not able to get correct values, for example

f.write(struct.pack('<f', 21.988))

in Java I have this value: 8.962863E27

I try to send a binary and match the output to ubjson library written in java, at first I use Big-endian mark but does not work, and when I use Little-endian it works like that.

Thanks for any guides.

Edit: some part of library

public JsonValue parse(final DataInputStream din) throws IOException {
        return parse(din, din.readByte());
    }

    protected JsonValue parse(final DataInputStream din, final byte type) throws IOException {
        if (type == '[')
            return parseArray(din);
        else if (type == '{')
            return parseObject(din);
        else if (type == 'Z')
            return new JsonValue(JsonValue.ValueType.nullValue);
        else if (type == 'T')
            return new JsonValue(true);
        else if (type == 'F') 
                .....

Have you looked at the file encoding to ensure that it's reading both files in the exact same format? — Mike McMahon
– Mike McMahon, Commented Nov 12, 2013 at 17:43
Mike I edited Post with part of library, it uses default Java Streams and read bytes. — daniel
– daniel, Commented Nov 12, 2013 at 17:50
and you're opening the file in python as 'wb' correct? Which is binary mode in python (only required on windows - but should be left in for cross platform support). — Mike McMahon
– Mike McMahon, Commented Nov 12, 2013 at 17:59
thanks, but I export to Java, as @martin wrote it was problem of Big/little-endian. — daniel
– daniel, Commented Nov 12, 2013 at 18:04

Martijn Pieters · Accepted Answer · 2013-11-12 17:51:08Z

1

Your Java application is using the opposite endianess; you are writing little-endian, but Java interpreted the value as big endian:

>>> struct.unpack('>f', struct.pack('<f', 21.988))
(8.962863280123082e+27,)

Write big-endian and Java will read the values correctly:

struct.pack('>f', 21.988)

If that doesn't work there are other reasons that your output is not interpreted correctly. The UBJSON specification is quite clear about endianess, it should all be encoded big-endian.

edited Nov 12, 2013 at 17:51

answered Nov 12, 2013 at 17:44

Martijn Pieters

1.1m326 gold badges4.2k silver badges3.4k bronze badges

Sign up to request clarification or add additional context in comments.

7 Comments

Martijn Pieters Over a year ago

@daniel: Then study the source code I linked you to some more. It certainly uses big-endian packing and unpacking.

daniel Over a year ago

@martin I studied before, in 3 days researching and tests.

Martijn Pieters Over a year ago

I didn't say you had not. But you'll need to compare your output with that of that library. Switching to little-endianess is not going to solve your problem.

daniel Over a year ago

Thanks @martin, you answer was currect, but I do not know other part is still in little-endian but float number in big-endian, and it works, f.write(struct.pack('<s', b"vert")) f.write(struct.pack('<c', b"d")) f.write(struct.pack('>f', 21.988))

Martijn Pieters Over a year ago

Characters are just one byte long, so there is no difference between endianess. struct.pack('<s', b"vert") == struct.pack('>s', b"vert") is True.

|

Collectives™ on Stack Overflow

Python Struct and reading values in Java

1 Answer 1

7 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

7 Comments

Your Answer

Sign up or log in

Post as a guest

Related