Hello all,
I tried to be precise in my subject.
I am completely new to hdf5 and do not know whether I use the correct words.
What I want to do is maybe best illustrated by a python script that already
does what I want on the python side; I now want to define
this same structure on the C++ side for reading and writing access. Simple data
exchange.
Short description:
I have three arrays of double: om1, om2, tau.
These arrays have a runtime-determined length N.
These arrays represent a snapshot of the state of a dynamical system.
I want to write these three arrays to disk for every time step of my simulation.
Here is the implementation on the python side (writing only one time step):
import scipy
from tables import *
N = 100
class Particle(IsDescription):
om1 = Float64Col(shape=(N+1))
om2 = Float64Col(shape=(N+1))
tau = Float64Col(shape=(N+1))
pass
fileh = openFile("data.h5", mode = "w")
root = fileh.root
group = fileh.createGroup(root, "mygroup")
gparticles = root.mygroup
table = fileh.createTable(gparticles, "mytable", Particle, "A Snapshot")
om1=scipy.randn(N+1)
om2=scipy.randn(N+1)
tau=scipy.randn(N+!)
particle['om1'] = om1
particle['om2'] = om2
particle['tau'] = tau
# Flush the table buffers
table.flush()
fileh.close()
This thing works exactly as expected. I am, however, not capable of defining
the proper compound struct on the C++ side that would be able to read and write
this thing. I am capable of creating the correct compound struct IF I FIX THE
SIZE N+1 of the arrays on the C++ side
at compile time.
I have not found a way to do this, if N is determined at runtime.
Here is the code that writes a similar file on the C++ side; as I say, with
compile-time-determined size of the arrays (the group part on the python side
is not yet implemented):
/***********************************************************************
* Example for compound structures with multiple datatypes. *
* blw 4/16/03 *
***********************************************************************/
/* Include the HDF5 library */
#include "hdf5.h"
/* System libraries to include */
#include <stdlib.h>
#include <assert.h>
/* Name of file for database output */
#define DATAFILE "compound_native.h5"
/* Name of dataset to create in datafile */
#define DATASETNAME "CompoundNative"
/* Dataset dimensions */
#define LENGTH 15
#define RANK 1
#define ARRAY_RANK 1
int
main(void)
{
const int ARRAY_DIM = 101;
/* Structure and array for compound types */
typedef struct Array1Struct {
double om1[ARRAY_DIM];
double om2[ARRAY_DIM];
double tau[ARRAY_DIM];
} Array1Struct;
Array1Struct Array1[LENGTH];
hid_t Array1Structid; /* File datatype identifier */
hid_t array_tid; /* Array datatype handle */
hid_t datafile, dataset;
hid_t dataspace;
herr_t status;
hsize_t dim[] = {LENGTH}; /* Dataspace dimensions */
hsize_t array_dim[] = {ARRAY_DIM}; /* Array dimensions */
int m, n; /* Array init loop vars */
/* Initialize the data in the arrays */
for (m = 0; m< LENGTH; m++) {
for (n = 0; n < ARRAY_DIM; n++) {
Array1[m].om1[n] = m+n;
}
for (n = 0; n < ARRAY_DIM; n++) {
Array1[m].om2[n] = m+n;
}
for (n = 0; n < ARRAY_DIM; n++) {
Array1[m].tau[n] = m+n;
}
}
/* Create the dataspace */
dataspace = H5Screate_simple(RANK, dim, NULL);
/* Create the file */
datafile = H5Fcreate(DATAFILE, H5F_ACC_TRUNC, H5P_DEFAULT,
H5P_DEFAULT);
/* Create the array data type */
array_tid = H5Tarray_create(H5T_NATIVE_DOUBLE, ARRAY_RANK,
array_dim, NULL);
/* Create the memory data type */
Array1Structid = H5Tcreate (H5T_COMPOUND, sizeof(Array1Struct));
H5Tinsert(Array1Structid, "om1", HOFFSET(Array1Struct, om1),
array_tid);
H5Tinsert(Array1Structid, "om2", HOFFSET(Array1Struct, om2),
array_tid);
H5Tinsert(Array1Structid, "tau", HOFFSET(Array1Struct, tau),
array_tid);
/* Create the dataset */
dataset = H5Dcreate(datafile, DATASETNAME, Array1Structid,
dataspace, H5P_DEFAULT);
/* Write data to the dataset */
status = H5Dwrite(dataset, Array1Structid, H5S_ALL, H5S_ALL,
H5P_DEFAULT, Array1);
/* Release resources */
H5Tclose(Array1Structid);
H5Tclose(array_tid);
H5Sclose(dataspace);
H5Dclose(dataset);
H5Fclose(datafile);
/* Return value */
return 0;
}
My aim is to be able to read, write,edit files both on the c++ and on the
python side for data exchange.
All help appreciated.
_______________________________________________
Hdf-forum is for HDF software users discussion.
[email protected]
http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org
