Hi,
I need to write scientific data in a 16 bits unsigned int multi dimensional
array.
My data in java is in memory in a 32 bit signed int multi dimensional array.
The in memory array and the dataset have the same dimensions.
This how I have create my hdf5 dataset:
Datatype dtype = testFile.createDatatype(Datatype.CLASS_INTEGER, 2,
Datatype.NATIVE, Datatype.SIGN_NONE);
Dataset dataset = testFile.createScalarDS("radiance", g1, dtype, dims2D,
null, null, 0, null);
I get a conversion exception when I try to write the full array with
dataset.write()
java.lang.ClassCastException: [[I cannot be cast to [I
at ncsa.hdf.object.Dataset.convertToUnsignedC(Dataset.java:1110)
at ncsa.hdf.object.h5.H5ScalarDS.write(H5ScalarDS.java:905)
at eumetsat.sandbox.Sandbox.create_dset(Sandbox.java:234)
at eumetsat.sandbox.Sandbox.main(Sandbox.java:299)
Could you please let me know what I am doing wrong or point me out to a
piece of code example doing the same thing.
Many thanks
Guillaume
···
--
View this message in context: http://hdf-forum.184993.n3.nabble.com/writing-16-bits-unsigned-int-with-java-hdf-tp4025219.html
Sent from the hdf-forum mailing list archive at Nabble.com.
When data is written from 32-bit singed integer in memory to 16-bit unsigned
integer in file, it will be converted to unsigned using Dataset.convertToUnsignedC().
Our current implementation of Dataset.convertToUnsignedC() does not support
multiple dimension arrays. It only works for 1-D arrays for performance reason.
We will add the issue (supporting for N-D arrays) to our work list.
Reading and writing N-D arrays is not efficient. We encourage users to read to 1D array
and write data in 1D, which is very efficient.
As a workaround, you could convert the data by yourself before you write the data.
See the example code below:
···
==========
private static void testH5Write2D(final String filename) throws Exception
{
long[] dims = {10, 5};
int[][] data = new int[(int)dims[0]][(int)dims[1]];
H5File file = new H5File(filename, H5File.CREATE);
file.open();
for (int i=0; i<data.length; i++)
for (int j=0; j<data[0].length; j++)
data[i][j] =(int) ((i+1)*j);
Datatype dtype = file.createDatatype(Datatype.CLASS_INTEGER, 2, Datatype.NATIVE, Datatype.SIGN_NONE);
Dataset dataset = file.createScalarDS("dset", null, dtype, dims, null, null, 0, null);
dataset.init();
short[][] tmp = new short[(int)dims[0]][(int)dims[1]];
for (int i=0; i<data.length; i++)
for (int j=0; j<data[0].length; j++)
tmp[i][j] = (short) data[i][j];
dataset.write(tmp);
file.close();
}
On 7/4/2012 9:44 AM, gaubert wrote:
Hi,
I need to write scientific data in a 16 bits unsigned int multi dimensional
array.
My data in java is in memory in a 32 bit signed int multi dimensional array.
The in memory array and the dataset have the same dimensions.
This how I have create my hdf5 dataset:
Datatype dtype = testFile.createDatatype(Datatype.CLASS_INTEGER, 2,
Datatype.NATIVE, Datatype.SIGN_NONE);
Dataset dataset = testFile.createScalarDS("radiance", g1, dtype, dims2D,
null, null, 0, null);
I get a conversion exception when I try to write the full array with
dataset.write()
java.lang.ClassCastException: [[I cannot be cast to [I
at ncsa.hdf.object.Dataset.convertToUnsignedC(Dataset.java:1110)
at ncsa.hdf.object.h5.H5ScalarDS.write(H5ScalarDS.java:905)
at eumetsat.sandbox.Sandbox.create_dset(Sandbox.java:234)
at eumetsat.sandbox.Sandbox.main(Sandbox.java:299)
Could you please let me know what I am doing wrong or point me out to a
piece of code example doing the same thing.
Many thanks
Guillaume
--
View this message in context: http://hdf-forum.184993.n3.nabble.com/writing-16-bits-unsigned-int-with-java-hdf-tp4025219.html
Sent from the hdf-forum mailing list archive at Nabble.com.
_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@hdfgroup.org
http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org
