Removed code which assumed that input files would need temporarily closed/reopened. HDF5 doesn't seem to have a limitation in the number of open files.

git-svn-id: http://svn.cactuscode.org/arrangements/CactusPUGHIO/IOHDF5/trunk@160 4825ed28-b72c-4eae-9704-e50c059e567d

1 files changed, 45 insertions, 122 deletions

diff --git a/src/util/hdf5_recombiner.c b/src/util/hdf5_recombiner.c
index bdcdee8..fd305d3 100644
--- a/src/util/hdf5_recombiner.c
+++ b/src/util/hdf5_recombiner.c
@@ -9,17 +9,11 @@
    @version   $Id$
  @@*/
 
-#include "cctk.h"
-
-#include <hdf5.h>
-#include <errno.h> 
-#include <stdio.h> 
+#include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
-#include <sys/stat.h>
-#ifdef HAVE_UNISTD_H
-#include <unistd.h>     /* sysconf(3) */
-#endif
+
+#include "cctk.h"
 #include "CactusPUGHIO/IOHDF5Util/src/ioHDF5UtilGH.h"
 
 /* the rcs ID and its dummy function to use it */
@@ -30,13 +24,9 @@ CCTK_FILEVERSION(CactusPUGHIO_IOHDF5_util_hdf5_recombiner_c)
 /*****************************************************************************/
 /*                           macro definitions                               */
 /*****************************************************************************/
-/* uncomment the following to get some debugging output */
-/* #define IOHDF5_DEBUG  1 */
-
 /* macro to do an HDF5 call, check its return code, and print a warning
    in case of an error */
 #define CHECK_ERROR(hdf5_call)                                                \
-          do                                                                  \
           {                                                                   \
             int _error_code = hdf5_call;                                      \
                                                                               \
@@ -48,7 +38,7 @@ CCTK_FILEVERSION(CactusPUGHIO_IOHDF5_util_hdf5_recombiner_c)
                                 __LINE__, #hdf5_call, _error_code);           \
               nerrors++;                                                      \
             }                                                                 \
-          } while (0)
+          }
 
 
 /*****************************************************************************/
@@ -62,7 +52,6 @@ CCTK_FILEVERSION(CactusPUGHIO_IOHDF5_util_hdf5_recombiner_c)
 static char *pathname = NULL;      /* pathname of the current object */
 static hid_t *infiles = NULL;      /* list of input file handles */
 static hid_t outfile = -1 ;        /* output file handle */
-static int max_filehandles = 0;    /* maximum number of open files */
 static char **infilenames = NULL;  /* list of input filenames */
 static int nprocs = 0;             /* total number of processors */
 static int ioproc_every = 0;       /* I/O was done on every N'th processor */
@@ -107,65 +96,38 @@ static int RecombineGroupData (const char *groupname);
                0 for success, negative return values indicate an error
    @endreturndesc
 @@*/
-int main (int argc, char *argv[])
+int main (int argc, const char *const argv[])
 {
-  int infile;
   int unchunked;
-  struct stat fileinfo;
   char *tmp, *template;
-  hid_t group, attr, dataspace;
+  hid_t infile, group, attr, dataspace;
 
 
   /* say hello to the user */
-  printf ("\n");
-  printf ("   -----------------------------\n");
-  printf ("   Cactus 4 HDF5 File Recombiner\n");
-  printf ("   -----------------------------\n");
-  printf ("\n");
+  printf ("\n   -----------------------------"
+          "\n   Cactus 4 HDF5 File Recombiner"
+          "\n   -----------------------------\n\n");
 
   /* give some help if called with incorrect number of parameters */
   if (argc != 3)
   {
-    fprintf (stderr, "Usage: %s <chunked_infile0> <unchunked_outfile>\n",
-             argv[0]);
-    fprintf (stderr, "   eg, %s alp.file_0.h5 alp.h5\n\n", argv[0]);
+    fprintf (stderr, "Usage: %s <chunked_infile0> <unchunked_outfile>\n"
+                     "   eg, %s alp.file_0.h5 alp.h5\n\n", argv[0], argv[0]);
     return (0);
   }
 
-  /* determine maximum number of files opened simultaneously */
-  max_filehandles = sysconf (_SC_OPEN_MAX);
-  if (max_filehandles < 0)
-  {
-    fprintf (stderr, "WARNING: Cannot determine filehandle limit ! "
-                     "Assuming no limit...\n");
-  }
-  /* subtract stdin, stdout, stderr, and output filehandle */
-  max_filehandles -= 4;
-
   /* open (first) input file */
-  if (stat (argv[1], &fileinfo))
-  {
-    fprintf (stderr, "ERROR: Cannot open input file '%s': %s\n\n",
-             argv[1], strerror (errno));
-    return (-1);
-  }
-  if (! H5Fis_hdf5 (argv[1]))
+  H5E_BEGIN_TRY
   {
-    fprintf (stderr, "ERROR: Input file '%s' is not an HDF5 file !\n\n",
-             argv[1]);
-    return (-1);
-  }
-  infiles = (hid_t *) malloc (1 * sizeof (hid_t));
-  infiles[0] = H5Fopen (argv[1], H5F_ACC_RDONLY, H5P_DEFAULT);
-  if (infiles[0] < 0)
+    infile = H5Fopen (argv[1], H5F_ACC_RDONLY, H5P_DEFAULT);
+  } H5E_END_TRY
+  if (infile < 0)
   {
-    fprintf (stderr, "ERROR: Cannot open input file '%s' as an HDF5 file !\n\n",
-             argv[1]);
+    fprintf (stderr, "ERROR: Cannot open HDF5 input file '%s' !\n\n", argv[1]);
     return (-1);
   }
-
   /* read the file set info from the GLOBAL_ATTRIBUTES_GROUP group */
-  CHECK_ERROR (group = H5Gopen (infiles[0], GLOBAL_ATTRIBUTES_GROUP));
+  CHECK_ERROR (group = H5Gopen (infile, GLOBAL_ATTRIBUTES_GROUP));
   if (group < 0)
   {
     fprintf (stderr, "ERROR: Cannot open attribute group '"
@@ -255,17 +217,14 @@ int main (int argc, char *argv[])
   {
     ninfiles++;
   }
-  if (max_filehandles < 0)
-  {
-    max_filehandles = ninfiles;
-  }
 
   printf ("Recombining HDF5 data from %d processors, "
           "output was written to %d file(s)\n\n", nprocs, ninfiles);
 
   /* allocate arrays for input file descriptors and filenames */
-  infiles = (hid_t *) realloc (infiles, ninfiles * sizeof (hid_t));
-  infilenames = (char **) malloc (ninfiles * sizeof (char *));
+  infiles = malloc (ninfiles * sizeof (hid_t));
+  infiles[0] = infile;
+  infilenames = malloc (ninfiles * sizeof (char *));
 
   /* now get all chunked input filenames and check that they can be opened */
   if (ninfiles == 1)
@@ -289,7 +248,7 @@ int main (int argc, char *argv[])
     }
 
     /* build the input filename template */
-    template = (char *) malloc (strlen (argv[1]) + 2);
+    template = malloc (strlen (argv[1]) + 2);
     strcpy (template, argv[1]);
     template[tmp - argv[1] + 6] = 0;
     strcat (template, "%d.h5");
@@ -298,7 +257,7 @@ int main (int argc, char *argv[])
     for (infile = 0; infile < ninfiles; infile++)
     {
       /* build the input filename */
-      infilenames[infile] = (char *) malloc (strlen (template) + 10);
+      infilenames[infile] = malloc (strlen (template) + 10);
       sprintf (infilenames[infile], template, infile);
       infiles[infile] = H5Fopen (infilenames[infile], H5F_ACC_RDONLY,
                                  H5P_DEFAULT);
@@ -308,12 +267,6 @@ int main (int argc, char *argv[])
                  infilenames[infile]);
         return (-1);
       }
-
-      /* close file if filehandle limit would be exceeded */
-      if (infile > max_filehandles)
-      {
-        CHECK_ERROR (H5Fclose (infiles[infile]));
-      }
     }
 
     free (template);
@@ -352,10 +305,6 @@ int main (int argc, char *argv[])
   /* finally, close all open files */
   for (infile = 0; infile < ninfiles; infile++)
   {
-    if (infile <= max_filehandles)
-    {
-      CHECK_ERROR (H5Fclose (infiles[infile]));
-    }
     free (infilenames[infile]);
   }
   free (infiles);
@@ -423,13 +372,12 @@ static herr_t CopyObject (hid_t from,
   char *current_pathname;
   int retval;
   size_t objectsize;
-  void *value;
+  void *data;
 
 
   /* build the full pathname for the current to object to process */
   current_pathname = pathname;
-  pathname = (char *) malloc (strlen (current_pathname) +
-                              strlen (objectname) + 2);
+  pathname = malloc (strlen (current_pathname) + strlen (objectname) + 2);
   sprintf (pathname, "%s/%s", current_pathname, objectname);
 
   /* get the output object identifier */
@@ -466,19 +414,13 @@ static herr_t CopyObject (hid_t from,
     CHECK_ERROR (dataspace = H5Dget_space (from));
     CHECK_ERROR (to = H5Dcreate (to, objectname, datatype, dataspace,
                                  H5P_DEFAULT));
-    objectsize = H5Tget_size (datatype);
-    if (H5Sis_simple (dataspace) > 0)
-    {
-      objectsize *= H5Sget_simple_extent_npoints (dataspace);
-    }
+    objectsize = H5Dget_storage_size (from);
     if (objectsize > 0)
     {
-      value = malloc (objectsize);
-      CHECK_ERROR (H5Dread (from, datatype, H5S_ALL, H5S_ALL, H5P_DEFAULT,
-                            value));
-      CHECK_ERROR (H5Dwrite (to, datatype, H5S_ALL, H5S_ALL, H5P_DEFAULT,
-                             value));
-      free (value);
+      data = malloc (objectsize);
+      CHECK_ERROR (H5Dread (from, datatype, H5S_ALL, H5S_ALL,H5P_DEFAULT,data));
+      CHECK_ERROR (H5Dwrite (to, datatype, H5S_ALL, H5S_ALL, H5P_DEFAULT,data));
+      free (data);
     }
     CHECK_ERROR (H5Aiterate (from, NULL, CopyAttribute, &to));
     CHECK_ERROR (H5Dclose (to));
@@ -516,7 +458,7 @@ static herr_t CopyObject (hid_t from,
    @vtype      hid_t
    @vio        in
    @endvar
-   @var        attrname
+   @var        name
    @vdesc      name of the current attribute
    @vtype      const char *
    @vio        in
@@ -534,7 +476,7 @@ static herr_t CopyObject (hid_t from,
    @endreturndesc
 @@*/
 static herr_t CopyAttribute (hid_t from,
-                             const char *attrname,
+                             const char *name,
                              void *_to)
 {
   hid_t attr, datatype, dataspace, to;
@@ -547,7 +489,7 @@ static herr_t CopyAttribute (hid_t from,
 
   /* open the attribute given by its name, get type, dataspace, and value
      and just copy it */
-  CHECK_ERROR (attr = H5Aopen_name (from, attrname));
+  CHECK_ERROR (attr = H5Aopen_name (from, name));
   CHECK_ERROR (datatype = H5Aget_type (attr));
   CHECK_ERROR (dataspace = H5Aget_space (attr));
   attrsize = H5Tget_size (datatype);
@@ -560,8 +502,7 @@ static herr_t CopyAttribute (hid_t from,
     value = malloc (attrsize);
     CHECK_ERROR (H5Aread (attr, datatype, value));
     CHECK_ERROR (H5Aclose (attr));
-    CHECK_ERROR (attr = H5Acreate (to, attrname, datatype, dataspace,
-                                    H5P_DEFAULT));
+    CHECK_ERROR (attr = H5Acreate (to, name, datatype, dataspace, H5P_DEFAULT));
     CHECK_ERROR (H5Awrite (attr, datatype, value));
     free (value);
   }
@@ -635,15 +576,12 @@ static int RecombineGroupData (const char *groupname)
                      "of type integer !\n", groupname);
     ndims = 0;
   }
+  global_size = NULL;
   if (ndims > 0)
   {
-    global_size = (hsize_t *) calloc (ndims, sizeof (hsize_t));
+    global_size = calloc (ndims, sizeof (hsize_t));
     CHECK_ERROR (H5Aread (attr, H5T_NATIVE_HSIZE, global_size));
   }
-  else
-  {
-    global_size = NULL;
-  }
   CHECK_ERROR (H5Sclose (dataspace));
   CHECK_ERROR (H5Tclose (datatype));
   CHECK_ERROR (H5Aclose (attr));
@@ -656,7 +594,7 @@ static int RecombineGroupData (const char *groupname)
   }
 
   /* allocate string buffer holding the possible dataset chunk names */
-  chunkname = (char *) malloc (strlen (groupname) + sizeof ("chunk") + 20);
+  chunkname = malloc (strlen (groupname) + sizeof ("chunk") + 20);
 
   /* the unchunked dataset's dimensions are taken from the 'global_size'
      attribute, with the least changing dimension being the first element */
@@ -672,8 +610,8 @@ static int RecombineGroupData (const char *groupname)
   free (global_size);
 
   /* allocate buffers to read a 'chunk_origin' attribute and the chunk dims */
-  chunk_origin = (hssize_t *) calloc (ndims, sizeof (hssize_t));
-  chunk_dims   = (hsize_t  *) calloc (ndims, sizeof (hsize_t));
+  chunk_origin = calloc (ndims, sizeof (hssize_t));
+  chunk_dims   = calloc (ndims, sizeof (hsize_t));
 
   unchunked_dataset = unchunked_datatype = -1;
 
@@ -688,16 +626,6 @@ static int RecombineGroupData (const char *groupname)
       nchunks = nprocs % ninfiles;
     }
 
-    /* re-open the file if it was closed before */
-    if (infile > max_filehandles)
-    {
-#ifdef DEBUG
-      printf ("reopening input file '%s'\n", infilenames[infile]);
-#endif
-      CHECK_ERROR (infiles[infile] = H5Fopen (infilenames[infile],
-                                              H5F_ACC_RDONLY, H5P_DEFAULT));
-    }
-
     /* loop over all chunks of this input file */
     for (chunk = 0; chunk < nchunks; chunk++)
     {
@@ -807,13 +735,17 @@ static int RecombineGroupData (const char *groupname)
       if (chunksize > 0)
       {
         /* give some info output */
-        printf ("    - file %d chunk %d\n", infile, chunk + ioproc_every*infile);
+        printf ("    - file %d chunk %d\n", infile,chunk + ioproc_every*infile);
         printf ("      chunk dimensions: [%d", (int) chunk_dims[ndims - 1]);
         for (chunk_ndims = 1; chunk_ndims < ndims; chunk_ndims++)
-            printf (", %d", (int) chunk_dims[ndims - chunk_ndims - 1]);
+        {
+          printf (", %d", (int) chunk_dims[ndims - chunk_ndims - 1]);
+        }
         printf ("]     chunk origin: [%d", (int) chunk_origin[ndims - 1]);
         for (chunk_ndims = 1; chunk_ndims < ndims; chunk_ndims++)
-            printf (", %d", (int) chunk_origin[ndims - chunk_ndims - 1]);
+        {
+          printf (", %d", (int) chunk_origin[ndims - chunk_ndims - 1]);
+        }
         printf ("]\n");
 
         /* read the chunk and write it to the unchunked dataset */
@@ -829,15 +761,6 @@ static int RecombineGroupData (const char *groupname)
       CHECK_ERROR (H5Sclose (chunked_dataspace));
 
     } /* end of loop over all chunks in this input file */
-
-    /* close input file if filehandle limit would be exceeded */
-    if (infile > max_filehandles)
-    {
-#ifdef DEBUG
-      printf ("temporarily closing input file '%s'\n", infilenames[infile]);
-#endif
-      CHECK_ERROR (H5Fclose (infiles[infile]));
-    }
   } /* end of loop over all input files */
 
   /* close objects and free allocated resources */