HDF5 back end

1. HDF5 back end

1 HDF5 back end

1.1 Template for HDF5 definitions

#define $GROUP$_GROUP_NAME          "$group$"
#define $GROUP_NUM$_NAME            "$group_num$"
#define $GROUP_DSET$_NAME           "$group_dset$"
#define $GROUP_STR$_NAME            "$group_str$"

1.2 Template for HDF5 structures

typedef struct trexio_hdf5_s {
  trexio_t     parent ;
  hid_t        file_id;
  hid_t        $group$_group;
} trexio_hdf5_t;

1.3 Template for HDF5 init/deinit

trexio_exit_code
trexio_hdf5_init (trexio_t* const file)
{

  trexio_hdf5_t* const f = (trexio_hdf5_t*) file;

  /* If file doesn't exist, create it */
  int f_exists = 0;
  struct stat st;

  if (stat(file->file_name, &st) == 0) f_exists = 1;

  if (f_exists == 1) {

    switch (file->mode) {
    case 'r':
      // reading the existing file -> open as RDONLY
      f->file_id = H5Fopen(file->file_name, H5F_ACC_RDONLY, H5P_DEFAULT);
      break;
    case 'w':
      // writing the existing file -> open as RDWRITE
      f->file_id = H5Fopen(file->file_name, H5F_ACC_RDWR, H5P_DEFAULT);
      break;
    }

  } else {

    switch (file->mode) {
    case 'r':
      // reading non-existing file -> error
      return TREXIO_FAILURE;
    case 'w':
      // writing non-existing file -> create it
      f->file_id = H5Fcreate(file->file_name, H5F_ACC_EXCL, H5P_DEFAULT, H5P_DEFAULT);
      break;
    }

  }

  /* Create or open groups in the hdf5 file assuming that they exist if file exists */
  switch (file->mode) {
    case 'r':
      f->$group$_group = H5Gopen(f->file_id, $GROUP$_GROUP_NAME, H5P_DEFAULT);
      break;
    case 'w':
      if (f_exists == 1) {
        f->$group$_group = H5Gopen(f->file_id, $GROUP$_GROUP_NAME, H5P_DEFAULT);
      } else {
        f->$group$_group = H5Gcreate(f->file_id, $GROUP$_GROUP_NAME, H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
      }
      break;
  }
  if (f->$group$_group <= 0L) return TREXIO_INVALID_ID;

  return TREXIO_SUCCESS;
}

trexio_exit_code
trexio_hdf5_deinit (trexio_t* const file)
{

  trexio_hdf5_t* f = (trexio_hdf5_t*) file;

  H5Gclose(f->$group$_group);
  f->$group$_group = 0;

  H5Fclose(f->file_id);
  f->file_id = 0;

  return TREXIO_SUCCESS;

}

1.4 Template for HDF5 has/read/write the numerical attribute

trexio_exit_code
trexio_hdf5_read_$group_num$ (trexio_t* const file, $group_num_dtype_double$* const num)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;
  if (num  == NULL) return TREXIO_INVALID_ARG_2;

  const trexio_hdf5_t* f = (const trexio_hdf5_t*) file;
  /* Quit if the dimensioning attribute is missing in the file */
  if (H5Aexists(f->$group$_group, $GROUP_NUM$_NAME) == 0) return TREXIO_FAILURE;

  /* Read the $group_num$ attribute of $group$ group */
  const hid_t num_id = H5Aopen(f->$group$_group, $GROUP_NUM$_NAME, H5P_DEFAULT);
  if (num_id <= 0) return TREXIO_INVALID_ID;

  const herr_t status = H5Aread(num_id, H5T_$GROUP_NUM_H5_DTYPE$, num);

  H5Aclose(num_id);

  if (status < 0) return TREXIO_FAILURE;

  return TREXIO_SUCCESS;

}

trexio_exit_code
trexio_hdf5_write_$group_num$ (trexio_t* const file, const $group_num_dtype_double$ num)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;

  trexio_hdf5_t* const f = (trexio_hdf5_t*) file;

  /* Write the dimensioning variables */
  const hid_t dtype = H5Tcopy(H5T_$GROUP_NUM_H5_DTYPE$);
  const hid_t dspace = H5Screate(H5S_SCALAR);

  const hid_t num_id = H5Acreate(f->$group$_group, $GROUP_NUM$_NAME, 
                                 dtype, dspace, H5P_DEFAULT, H5P_DEFAULT);
  if (num_id <= 0) {
    H5Sclose(dspace);
    H5Tclose(dtype);
    return TREXIO_INVALID_ID;
  }

  const herr_t status = H5Awrite(num_id, dtype, &(num));
  if (status < 0) {
    H5Aclose(num_id);
    H5Sclose(dspace);
    H5Tclose(dtype);
    return TREXIO_FAILURE;
  }

  H5Sclose(dspace);
  H5Aclose(num_id);
  H5Tclose(dtype);

  return TREXIO_SUCCESS;
}

trexio_exit_code
trexio_hdf5_has_$group_num$ (trexio_t* const file)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;

  const trexio_hdf5_t* f = (const trexio_hdf5_t*) file;

  htri_t status = H5Aexists(f->$group$_group, $GROUP_NUM$_NAME);
  /* H5Aexists returns positive value if attribute exists, 0 if does not, negative if error */
  if (status > 0){
    return TREXIO_SUCCESS;
  } else if (status == 0) {
    return TREXIO_HAS_NOT;
  } else {
    return TREXIO_FAILURE;
  }

}

1.5 Template for HDF5 has/read/write the dataset of numerical data

trexio_exit_code
trexio_hdf5_read_$group_dset$ (trexio_t* const file, $group_dset_dtype$* const $group_dset$, const uint32_t rank, const uint64_t* dims)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;
  if ($group_dset$  == NULL) return TREXIO_INVALID_ARG_2;

  const trexio_hdf5_t* f = (const trexio_hdf5_t*) file;

  // open the dataset to get its dimensions
  hid_t dset_id = H5Dopen(f->$group$_group, $GROUP_DSET$_NAME, H5P_DEFAULT);
  if (dset_id <= 0) return TREXIO_INVALID_ID;

  // allocate space for the dimensions to be read
  hsize_t* ddims = CALLOC( (int) rank, hsize_t);
  if (ddims == NULL) return TREXIO_FAILURE;

  // get the dataspace of the dataset
  hid_t dspace_id = H5Dget_space(dset_id);
  // get the rank and dimensions of the dataset
  int rrank = H5Sget_simple_extent_dims(dspace_id, ddims, NULL);

  // check that dimensions are consistent
  if (rrank != (int) rank) {
    FREE(ddims);
    H5Sclose(dspace_id);
    H5Dclose(dset_id);
    return TREXIO_INVALID_ARG_3;
  }

  for (uint32_t i=0; i<rank; ++i){
    if (ddims[i] != dims[i]) {
      FREE(ddims);
      H5Sclose(dspace_id);
      H5Dclose(dset_id);
      return TREXIO_INVALID_ARG_4;
    }
  }

  FREE(ddims);
  H5Sclose(dspace_id);
  H5Dclose(dset_id);

  /* High-level H5LT API. No need to deal with dataspaces and datatypes */
  herr_t status = H5LTread_dataset(f->$group$_group,
                                   $GROUP_DSET$_NAME,
                                   H5T_$GROUP_DSET_H5_DTYPE$,
                                   $group_dset$);
  if (status < 0) return TREXIO_FAILURE;

  return TREXIO_SUCCESS;
}

trexio_exit_code
trexio_hdf5_write_$group_dset$ (trexio_t* const file, const $group_dset_dtype$* $group_dset$, const uint32_t rank, const uint64_t* dims)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;
  if ($group_dset$  == NULL) return TREXIO_INVALID_ARG_2;

  trexio_hdf5_t* f = (trexio_hdf5_t*) file;

  if ( H5LTfind_dataset(f->$group$_group, $GROUP_DSET$_NAME) != 1 ) {

    const herr_t status = H5LTmake_dataset(f->$group$_group,
                                           $GROUP_DSET$_NAME,
                                           (int) rank, (const hsize_t*) dims,
                                           H5T_$GROUP_DSET_H5_DTYPE$,
                                           $group_dset$);
    if (status < 0) return TREXIO_FAILURE;

  } else {

    hid_t dset_id = H5Dopen(f->$group$_group, $GROUP_DSET$_NAME, H5P_DEFAULT);
    if (dset_id <= 0) return TREXIO_INVALID_ID;

    const herr_t status = H5Dwrite(dset_id,
                                   H5T_$GROUP_DSET_H5_DTYPE$,
                                   H5S_ALL, H5S_ALL, H5P_DEFAULT,
                                   $group_dset$);

    H5Dclose(dset_id);
    if (status < 0) return TREXIO_FAILURE;

  }

  return TREXIO_SUCCESS;

}

trexio_exit_code
trexio_hdf5_has_$group_dset$ (trexio_t* const file)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;

  trexio_hdf5_t* f = (trexio_hdf5_t*) file;

  herr_t status = H5LTfind_dataset(f->$group$_group, $GROUP_DSET$_NAME);
  /* H5LTfind_dataset returns 1 if dataset exists, 0 otherwise */
  if (status == 1){
    return TREXIO_SUCCESS;
  } else if (status == 0) {
    return TREXIO_HAS_NOT;
  } else {
    return TREXIO_FAILURE;
  }

}

1.6 Template for HDF5 has/read/write the dataset of strings

trexio_exit_code
trexio_hdf5_read_$group_dset$ (trexio_t* const file, char* const $group_dset$, const uint32_t rank, const uint64_t* dims, const uint32_t max_str_len)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;
  if ($group_dset$  == NULL) return TREXIO_INVALID_ARG_2;

  const trexio_hdf5_t* f = (const trexio_hdf5_t*) file;

  herr_t status;

  // open the dataset to get its dimensions
  hid_t dset_id = H5Dopen(f->$group$_group, $GROUP_DSET$_NAME, H5P_DEFAULT);
  if (dset_id <= 0) return TREXIO_INVALID_ID;

  // allocate space for the dimensions to be read
  hsize_t* ddims = CALLOC( (int) rank, hsize_t);
  if (ddims == NULL) {
    H5Dclose(dset_id);
    return TREXIO_ALLOCATION_FAILED;
  }

  hid_t dspace = H5Dget_space(dset_id); 
  if (dset_id <= 0) {
    FREE(ddims);
    H5Dclose(dset_id); 
    return TREXIO_INVALID_ID;
  }

  // get the rank of the dataset in a file
  int rrank = H5Sget_simple_extent_dims(dspace, ddims, NULL);

  if (rrank != (int) rank) {
    FREE(ddims);
    H5Dclose(dset_id);
    H5Sclose(dspace);
    return TREXIO_INVALID_ARG_3;
  }

  for (int i=0; i<rrank; i++) {
    if (ddims[i] != dims[i]) {
      H5Dclose(dset_id);
      H5Sclose(dspace);
      FREE(ddims);
      return TREXIO_INVALID_ARG_4;
    }
  }
  FREE(ddims);

  hid_t memtype = H5Tcopy (H5T_C_S1);
  status = H5Tset_size(memtype, H5T_VARIABLE);
  if (status < 0 || memtype <= 0) {
    H5Dclose(dset_id);
    H5Sclose(dspace);
    return TREXIO_FAILURE;
  }

  char** rdata = CALLOC(dims[0], char*);
  if (rdata == NULL) {
    H5Dclose(dset_id);
    H5Sclose(dspace);
    H5Tclose(memtype); 
    return TREXIO_ALLOCATION_FAILED;
  }

  status = H5Dread(dset_id, memtype, H5S_ALL, H5S_ALL, H5P_DEFAULT, rdata);
  if (status < 0) {
    FREE(rdata);
    H5Dclose(dset_id);
    H5Sclose(dspace);
    H5Tclose(memtype); 
    return TREXIO_FAILURE;
  }

  // copy contents of temporary rdata buffer into the group_dset otherwise they are lost
  // after calling H5Treclaim or H5Dvlen_reclaim functions
  strcpy($group_dset$, "");
  for (uint64_t i=0; i<dims[0]; i++) {
    strncat($group_dset$, rdata[i], max_str_len);
    strcat($group_dset$, TREXIO_DELIM);
  }

  // H5Dvlen_reclaim is deprecated and replaced by H5Treclaim in HDF5 v.1.12.0
  #if (H5_VERS_MAJOR <= 1 && H5_VERS_MINOR < 12)
    status = H5Dvlen_reclaim(memtype, dspace, H5P_DEFAULT, rdata);
  #else
    status = H5Treclaim(memtype, dspace, H5P_DEFAULT, rdata);
  #endif

  if (status < 0) {
    FREE(rdata);
    H5Dclose(dset_id);
    H5Sclose(dspace);
    H5Tclose(memtype); 
    return TREXIO_FAILURE;
  }

  FREE(rdata); 
  H5Dclose(dset_id);
  H5Sclose(dspace);
  H5Tclose(memtype);

  return TREXIO_SUCCESS;
}

trexio_exit_code
trexio_hdf5_write_$group_dset$ (trexio_t* const file, const char** $group_dset$, const uint32_t rank, const uint64_t* dims)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;
  if ($group_dset$  == NULL) return TREXIO_INVALID_ARG_2;

  trexio_hdf5_t* f = (trexio_hdf5_t*) file;

  herr_t status;
  hid_t dset_id;

  /* we are going to write variable-length strings */
  hid_t memtype = H5Tcopy (H5T_C_S1);
  if (memtype <= 0) return TREXIO_INVALID_ID;

  status = H5Tset_size (memtype, H5T_VARIABLE);
  if (status < 0) return TREXIO_FAILURE;

  if ( H5LTfind_dataset(f->$group$_group, $GROUP_DSET$_NAME) != 1 ) {

    /* code to create dataset */   
    hid_t filetype = H5Tcopy (H5T_FORTRAN_S1);
    if (filetype <= 0) return TREXIO_INVALID_ID;

    status = H5Tset_size (filetype, H5T_VARIABLE);
    if (status < 0) return TREXIO_FAILURE;

    hid_t dspace = H5Screate_simple( (int) rank, (const hsize_t*) dims, NULL);
    if (dspace <= 0) return TREXIO_INVALID_ID;

    dset_id = H5Dcreate (f->$group$_group, $GROUP_DSET$_NAME, filetype, dspace,
                         H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
    if (dset_id <= 0) return TREXIO_INVALID_ID;

    status = H5Dwrite (dset_id, memtype,
                       H5S_ALL, H5S_ALL, H5P_DEFAULT,
                       $group_dset$);

    H5Dclose (dset_id);
    H5Sclose (dspace);
    H5Tclose (filetype);
    H5Tclose (memtype);

    if (status < 0) return TREXIO_FAILURE;

  } else {

    dset_id = H5Dopen(f->$group$_group, $GROUP_DSET$_NAME, H5P_DEFAULT);
    if (dset_id <= 0) return TREXIO_INVALID_ID;

    /* code to write dataset */
    status = H5Dwrite(dset_id, memtype,
                      H5S_ALL, H5S_ALL, H5P_DEFAULT,
                      $group_dset$);

    H5Dclose(dset_id);
    H5Tclose(memtype);

    if (status < 0) return TREXIO_FAILURE;

  }

  return TREXIO_SUCCESS;

}

trexio_exit_code
trexio_hdf5_has_$group_dset$ (trexio_t* const file)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;

  trexio_hdf5_t* f = (trexio_hdf5_t*) file;

  herr_t status = H5LTfind_dataset(f->$group$_group, $GROUP_DSET$_NAME);
  /* H5LTfind_dataset returns 1 if dataset exists, 0 otherwise */
  if (status == 1){
    return TREXIO_SUCCESS;
  } else if (status == 0) {
    return TREXIO_HAS_NOT;
  } else {
    return TREXIO_FAILURE;
  }

}

1.7 Template for HDF5 has/read/write the string attribute

trexio_exit_code
trexio_hdf5_read_$group_str$ (trexio_t* const file, char* const str, const uint32_t max_str_len)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;
  if (str  == NULL) return TREXIO_INVALID_ARG_2;

  const trexio_hdf5_t* f = (const trexio_hdf5_t*) file;
  /* Quit if the string attribute is missing in the file */
  if (H5Aexists(f->$group$_group, $GROUP_STR$_NAME) == 0) return TREXIO_HAS_NOT;

  /* Read the $group_str$ attribute of $group$ group */
  const hid_t str_id = H5Aopen(f->$group$_group, $GROUP_STR$_NAME, H5P_DEFAULT);
  if (str_id <= 0) return TREXIO_INVALID_ID;

  const hid_t ftype_id = H5Aget_type(str_id);
  if (ftype_id <= 0) return TREXIO_INVALID_ID;
  uint64_t sdim = H5Tget_size(ftype_id);
  if (sdim <= 0) return TREXIO_FAILURE;
  sdim++;                         /* Make room for null terminator */

  const hid_t mem_id = H5Tcopy(H5T_C_S1);
  if (mem_id <= 0) return TREXIO_INVALID_ID;

  herr_t status;
  status = (max_str_len+1) > sdim ? H5Tset_size(mem_id, sdim) : H5Tset_size(mem_id, max_str_len+1) ;
  if (status < 0) return TREXIO_FAILURE;

  status = H5Aread(str_id, mem_id, str);
  if (status < 0) return TREXIO_FAILURE;

  H5Aclose(str_id);
  H5Tclose(mem_id);
  H5Tclose(ftype_id);

  return TREXIO_SUCCESS;

}

trexio_exit_code
trexio_hdf5_write_$group_str$ (trexio_t* const file, const char* str)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;
  if (str  == NULL) return TREXIO_INVALID_ARG_2;

  trexio_hdf5_t* const f = (trexio_hdf5_t*) file;


  /* Setup the dataspace */
  const hid_t dtype_id = H5Tcopy(H5T_C_S1);
  if (dtype_id <= 0) return TREXIO_INVALID_ID;

  size_t str_attr_len = strlen(str) + 1;

  herr_t status;
  status = H5Tset_size(dtype_id, str_attr_len);
  if (status < 0) return TREXIO_FAILURE;

  status = H5Tset_strpad(dtype_id, H5T_STR_NULLTERM);
  if (status < 0) return TREXIO_FAILURE;

  const hid_t dspace_id = H5Screate(H5S_SCALAR);
  if (dspace_id <= 0) return TREXIO_INVALID_ID;

  /* Create the $group_str$ attribute of $group$ group */
  const hid_t str_id = H5Acreate(f->$group$_group, $GROUP_STR$_NAME, dtype_id, dspace_id,
                       H5P_DEFAULT, H5P_DEFAULT);

  if (str_id <= 0) {
    H5Sclose(dspace_id);
    H5Tclose(dtype_id);
    return TREXIO_INVALID_ID;
  }

  status = H5Awrite(str_id, dtype_id, str);
  if (status < 0) {
    H5Aclose(str_id);
    H5Sclose(dspace_id);
    H5Tclose(dtype_id);
    return TREXIO_FAILURE;
  }

  H5Aclose(str_id);
  H5Sclose(dspace_id);
  H5Tclose(dtype_id);
  return TREXIO_SUCCESS;

}

trexio_exit_code
trexio_hdf5_has_$group_str$ (trexio_t* const file)
{

  if (file == NULL) return TREXIO_INVALID_ARG_1;

  const trexio_hdf5_t* f = (const trexio_hdf5_t*) file;

  htri_t status = H5Aexists(f->$group$_group, $GROUP_STR$_NAME);
  /* H5Aexists returns positive value if attribute exists, 0 if does not, negative if error */
  if (status > 0){
    return TREXIO_SUCCESS;
  } else if (status == 0) {
    return TREXIO_HAS_NOT;
  } else {
    return TREXIO_FAILURE;
  }

}

HDF5 back end

Table of Contents

1 HDF5 back end

1.1 Template for HDF5 definitions

1.2 Template for HDF5 structures

1.3 Template for HDF5 init/deinit

1.4 Template for HDF5 has/read/write the numerical attribute

1.5 Template for HDF5 has/read/write the dataset of numerical data

1.6 Template for HDF5 has/read/write the dataset of strings

1.7 Template for HDF5 has/read/write the string attribute