add -l options to h5fuse

src/H5FDsubfiling/H5FDsubfiling.c

@@ -594,6 +594,115 @@ H5Pget_fapl_subfiling(hid_t fapl_id, H5FD_subfiling_config_t *config_out)
     H5_SUBFILING_FUNC_LEAVE_API;
 } /* end H5Pget_fapl_subfiling() */
 
+herr_t
+H5FDsubfiling_get_file_mapping(hid_t file_id, char ***filenames, size_t *len)
+{
+    subfiling_context_t *sf_context    = NULL;
+    H5FD_t              *driver        = NULL;
+    H5F_t               *file_ptr      = NULL;
+    char               **filenames_arr = NULL;
+    char                *filepath      = NULL;
+    char                *subfile_dir   = NULL;
+    char                *base          = NULL;
+    int                  num_subfiles  = 0;
+    int                  num_digits    = 0;
+    herr_t               ret_value     = SUCCEED;
+
+    if (file_id < 0)
+        H5_SUBFILING_GOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "invalid file ID");
+    if (!filenames)
+        H5_SUBFILING_GOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "`filenames` was NULL");
+    if (!len)
+        H5_SUBFILING_GOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "`len` was NULL");
+
+    *filenames = NULL;
+    *len       = 0;
+
+    if (NULL == (file_ptr = H5VL_object(file_id)))
+        H5_SUBFILING_GOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "invalid file ID");
+
+    if (H5F_shared_get_file_driver(H5F_SHARED(file_ptr), &driver) < 0)
+        H5_SUBFILING_GOTO_ERROR(H5E_FILE, H5E_CANTGET, FAIL, "can't get driver structure from file ID");
+
+    if (NULL == (sf_context = H5_get_subfiling_object(((H5FD_subfiling_t *)driver)->context_id)))
+        H5_SUBFILING_GOTO_ERROR(H5E_FILE, H5E_CANTGET, FAIL, "can't get subfiling context from ID");
+
+    if (!sf_context->topology)
+        H5_SUBFILING_GOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "application topology hasn't been initialized yet for this file");
+
+    assert(sf_context->h5_file_id != UINT64_MAX);
+    assert(sf_context->h5_filename);
+    assert(sf_context->sf_num_subfiles > 0);
+    assert(sf_context->topology);
+
+    if (sf_context->topology->rank_is_ioc) {
+        assert(sf_context->sf_fids);
+        assert(sf_context->sf_num_fids > 0);
+
+        /* Get the basename of the full HDF5 filename */
+        if (H5_basename(sf_context->h5_filename, &base) < 0)
+            H5_SUBFILING_GOTO_ERROR(H5E_RESOURCE, H5E_CANTALLOC, FAIL, "can't get HDF5 file basename");
+
+        /*
+         * Get the directory prefix where subfiles will be placed.
+         * Under normal circumstances, the subfiles are co-located
+         * with the HDF5 file, but users may specify a different
+         * directory name.
+         */
+        if (sf_context->subfile_prefix) {
+            if (NULL == (subfile_dir = H5MM_strdup(sf_context->subfile_prefix)))
+                H5_SUBFILING_GOTO_ERROR(H5E_RESOURCE, H5E_CANTALLOC, FAIL, "couldn't copy subfile prefix");
+        }
+        else {
+            if (H5_dirname(sf_context->h5_filename, &subfile_dir) < 0)
+                H5_SUBFILING_GOTO_ERROR(H5E_RESOURCE, H5E_CANTALLOC, FAIL, "couldn't get HDF5 file dirname");
+        }
+
+        num_subfiles = sf_context->sf_num_subfiles;
+        num_digits   = (int)(log10(num_subfiles) + 1);
+
+        if (NULL == (filenames_arr = calloc(1, (size_t)sf_context->sf_num_fids * sizeof(char *))))
+            H5_SUBFILING_GOTO_ERROR(H5E_RESOURCE, H5E_CANTALLOC, FAIL, "couldn't allocate filenames array");
+
+        for (int i = 0; i < sf_context->sf_num_fids; i++) {
+            int subfile_idx;
+
+            if (NULL == (filepath = malloc(PATH_MAX)))
+                H5_SUBFILING_GOTO_ERROR(H5E_RESOURCE, H5E_CANTALLOC, FAIL,
+                                        "couldn't allocate space for subfile filename");
+
+            subfile_idx = (i * sf_context->topology->n_io_concentrators) + sf_context->topology->ioc_idx + 1;
+
+            snprintf(filepath, PATH_MAX, "%s/" H5FD_SUBFILING_FILENAME_TEMPLATE, subfile_dir, base,
+                             sf_context->h5_file_id, num_digits, subfile_idx, num_subfiles);
+
+            filenames_arr[i] = filepath;
+            filepath         = NULL;
+        }
+
+        *filenames = filenames_arr;
+        *len       = (size_t)sf_context->sf_num_fids;
+    }
+
+done:
+    if (ret_value < 0) {
+        if (filenames_arr) {
+            for (int i = 0; i < sf_context->sf_num_fids; i++) {
+                free(filenames_arr[i]);
+            }
+        }
+        free(filenames_arr);
+
+        *filenames = NULL;
+        *len       = 0;
+    }
+
+    H5MM_free(base);
+    H5MM_free(subfile_dir);
+
+    H5_SUBFILING_FUNC_LEAVE_API;
+}
+
 static herr_t
 H5FD__subfiling_get_default_config(hid_t fapl_id, H5FD_subfiling_config_t *config_out)
 {

src/H5FDsubfiling/H5FDsubfiling.h

@@ -406,6 +406,12 @@ H5_DLL herr_t H5Pset_fapl_subfiling(hid_t fapl_id, const H5FD_subfiling_config_t
  */
 H5_DLL herr_t H5Pget_fapl_subfiling(hid_t fapl_id, H5FD_subfiling_config_t *config_out);
 
+/*
+ * The caller must call H5free_memory on each of the entries in `filenames`, as well
+ * as on `filenames` itself.
+ */
+H5_DLL herr_t H5FDsubfiling_get_file_mapping(hid_t file_id, char ***filenames, size_t *len);
+
 #ifdef __cplusplus
 }
 #endif

testpar/t_subfiling_vfd.c

@@ -253,6 +253,27 @@ test_create_and_close(void)
     file_id = H5Fcreate(SUBF_FILENAME, H5F_ACC_TRUNC, H5P_DEFAULT, fapl_id);
     VRFY((file_id >= 0), "H5Fcreate succeeded");
 
+    char **filenames = NULL;
+    size_t len = 0;
+    H5FDsubfiling_get_file_mapping(file_id, &filenames, &len);
+
+    if (len > 0) {
+        int l_mpi_rank;
+        MPI_Comm_rank(MPI_COMM_WORLD, &l_mpi_rank);
+        printf("I AM MPI RANK %d RESPONSIBLE FOR SUBFILES: [", l_mpi_rank);
+        for (size_t i = 0; i < len; i++) {
+            if (i > 0)
+                printf(", ");
+            printf("%s", filenames[i]);
+        }
+        printf("]\n\n");
+
+        for (size_t i = 0; i < len; i++) {
+            H5free_memory(filenames[i]);
+        }
+        H5free_memory(filenames);
+    }
+
     VRFY((H5Fclose(file_id) >= 0), "File close succeeded");
 
     H5E_BEGIN_TRY

utils/subfiling_vfd/h5fuse.in

@@ -26,10 +26,11 @@ function usage {
          configuration file either as a command-line argument or the script will
          search for the *.config file in the current directory."
    echo ""
-   echo "usage: h5fuse [-f filename] [-h] [-p] [-q] [-r] [-v] "
+   echo "usage: h5fuse [-f filename] [-h] [-p] [-q] [-r] [-v] [-l subfile1,subfile2,...]"
    echo "-f filename  Subfile configuration file."
    echo "-h           Print this help."
    echo "-q           Quiet all output. [no]"
+   echo "-l           Comma seperated list of subfiles to process"
    echo "-p           h5fuse is being run in parallel, with more than one rank. [no]"
    echo "-r           Remove subfiles after being processed. [no]"
    echo "-v           Verbose output. [no]"
@@ -172,7 +173,8 @@ verbose="false"
 quiet="false"
 rm_subf="false"
 parallel="false"
-while getopts "hpqrvf:" option; do
+subf_list=()
+while getopts "hpqrvf:l:" option; do
    case $option in
       f) # subfiling configuration file
          file_config=$OPTARG;;
@@ -187,6 +189,9 @@ while getopts "hpqrvf:" option; do
          rm_subf="true";;
       v) # verbose output
          verbose="true";;
+      l) # process a list of subfiles
+         # Split the comma-sperated list into an array
+         IFS=', ' read -r -a subf_list <<< "${OPTARG}";;
      \?) # Invalid option
          echo -e "$RED ERROR: Invalid option ${BLD}-${OPTARG}${RED} $NC"
          usage
@@ -259,17 +264,48 @@ fi
 nsubfiles=${#subfiles[@]}
 
 # Get the number of local subfiles
+
 subfiles_loc=()
 subfiles_size=()
-for i in "${subfiles[@]}"; do
-    subfile="${subfile_dir}/${i}"
-    if [ -f "${subfile}" ]; then
-        subfiles_loc+=("$subfile")
-        subfiles_size+=($(wc -c "${subfile}" | awk '{print $1}'))
-    else
-        subfiles_size+=(0)
-    fi
-done
+
+len=${#subf_list[@]}
+
+if [[ "${len}" == "0" ]]; then
+
+  for i in "${subfiles[@]}"; do
+      subfile="${subfile_dir}/${i}"
+      if [ -f "${subfile}" ]; then
+          subfiles_loc+=("$subfile")
+          subfiles_size+=($(wc -c "${subfile}" | awk '{print $1}'))
+      else
+          subfiles_size+=(0)
+      fi
+  done
+
+else
+
+  for ((i = 0; i < $nsubfiles; i++)); do
+    subfiles_size[i]=0
+  done
+
+  idx=0
+  for j in "${subf_list[@]}"; do
+      subfile1="${subfile_dir}/${j}"
+      for k in "${subfiles[@]}"; do
+          subfile="${subfile_dir}/${k}"
+          if [[ "${subfile1}" == "${subfile}" ]]; then
+            subfiles_loc+=("$subfile1")
+            subfiles_size[$idx]=$(wc -c "${subfile1}" | awk '{print $1}')
+            ((idx++))
+            break
+          fi
+      done
+      if [[ "${idx}" == "${len}" ]]; then
+        break;
+      fi
+  done
+
+fi
 
 if [ "$quiet" == "false" ]; then
     TIMEFORMAT="COMPLETION TIME = %R s"