Read Binary Data into a QVector

J.Hilk

Without seeing more,

I would suggest union

union
{
    double f;
    unsigned char chr;
} u;

this should do the trick as long as you pass the correct byteArrey/length.

code: untested from my side

BadHombre

@VRonin so i tested your code but i don ' t get the right values :/

matlab code : 
matrix_save=[A' B' C D' E' ];
fileID = fopen('folder/filename.bin','w');
fwrite(fileID,matrix_save','double');
fclose(fileID);

this is the code i wrote for c++ .. A B C D E are some Matrix..with numbers in it

  ifstream file("E:\\file.bin");

        char * buffer;
        long size;
        file.seekg(0, std::ios::end);
        size = file.tellg();
        file.seekg(0, std::ios::beg);
        buffer = new char[size];
        file.read(buffer, size);
        file.close();

        double* double_values = (double*)buffer;//reinterpret as doubles
        vector<double> buffer2(double_values, double_values + (size / sizeof(double)));

VRonin

the code I posted above should work then

J.Hilk

@BadHombre
I totally missed, that you're using <fsteam>

You could also use this code sample:

ifstream file("E:\\file.bin");
double read;

        if ( file) {
            file.read( reinterpret_cast<char*>( &read ), sizeof read );
            qDebug() << read;
        }

If MatLap has the same byteorder of course :)

Edit, I acutally found the Source, from when i was researching that.

BadHombre

@VRonin doesn't work :/

this are the values in the Qvector

0
0
3.1568e-58
-3.03415e-298
2.08159e+74
2.92534e+211
-2.01411e+84
3.21098e+39
0
0

and this are the numbers from matlab which it should have :/

0	0	-5,99983766238420	-5,99651426671551	-5,22786749086602	5,99821019538431	5,99832392614572	5,22649512993718	0	0

VRonin

If you can temporarily change the matlab file with from fwrite(fileID,matrix_save','double'); to fwrite(fileID,matrix_save','b'); and see if it works in that case?

BadHombre

@VRonin I get an error

Error using fwrite
Invalid precision.

Error in Untitled (line 28)
fwrite(fileID,matrix_save','b');

VRonin

@BadHombre my bad, I deleted an argument by mistake, I meant fwrite(fileID,matrix_save','double','b');

BadHombre

@VRonin now i get the right number ... what was the Problem ? Is it possible to get the right numbers without changing the matlab code ?

J.Hilk

@BadHombre

the b paramter tells matlab to write in Big-endian ordering

VRonin

@BadHombre said in Read Binary Data into a QVector:

what was the Problem ?

@J-Hilk is super correct!

Is it possible to get the right numbers without changing the matlab code ?

Yes and no. If your matlab code and your program are assured to run on the same system and you know the default byte order of that system yes.

if you know your system is big-endian then my original code will just work, if you know your system is little endian (like in your case) add fileStream.setByteOrder(QDataStream::LittleEndian); before the for(;;)

There is no way with your current matlab code to make the file readable in another system without a doubt

EDIT

Actually you can hack your way around the "you know the default byte order of that system" but there is no way out of the "your matlab code and your program are assured to run on the same system" part

bool is_big_endian()
{
    union {
        quint32 i;
        char c[4];
    } bint = {0x01020304};
    return bint.c[0] == 1; 
}

QFile file1("C:/Qt/Test/test.bin");
QVector<double> outVect;
if (file1.open(QIODevice::ReadOnly)){
QDataStream fileStream(&file1);
if(!is_big_endian())
fileStream.setByteOrder(QDataStream::LittleEndian);
for(;;){
double tempValue;
fileStream.startTransaction();
fileStream >> tempValue;
if(fileStream.commitTransaction())
outVect << tempValue;
else
break;
}
}