HDFFV-8611 change h5repack to save the root group creation order

Added test and new arguments to control the the input file parsing.

5 files changed, 134 insertions, 35 deletions

diff --git a/tools/src/h5repack/h5repack.c b/tools/src/h5repack/h5repack.c
index 3c9ce90..bc8527b 100644
--- a/tools/src/h5repack/h5repack.c
+++ b/tools/src/h5repack/h5repack.c
@@ -784,6 +784,8 @@ static int check_objects(const char* fname, pack_opt_t *options) {
     *-------------------------------------------------------------------------
     */
 
+    /* Initialize indexing options */
+    h5trav_set_index(sort_by, sort_order);
     /* init table */
     trav_table_init(&travt);
 
diff --git a/tools/src/h5repack/h5repack.h b/tools/src/h5repack/h5repack.h
index e36f0ef..e5ae03f 100644
--- a/tools/src/h5repack/h5repack.h
+++ b/tools/src/h5repack/h5repack.h
@@ -29,10 +29,10 @@
 #define MAX_COMPACT_DSIZE  64512  /* max data size for compact layout. -1k for header size */
 
 /* File space default information */
-#define FS_PAGESIZE_DEF		4096
-#define FS_STRATEGY_DEF        	H5F_FSPACE_STRATEGY_FSM_AGGR
-#define FS_PERSIST_DEF         	FALSE
-#define FS_THRESHOLD_DEF       	1
+#define FS_PAGESIZE_DEF        4096
+#define FS_STRATEGY_DEF            H5F_FSPACE_STRATEGY_FSM_AGGR
+#define FS_PERSIST_DEF             FALSE
+#define FS_THRESHOLD_DEF           1
 
 
 /*-------------------------------------------------------------------------
@@ -120,10 +120,10 @@ typedef struct {
  hsize_t         meta_block_size;  /* metadata aggregation block size (for H5Pset_meta_block_size) */
  hsize_t         threshold;        /* alignment threshold for H5Pset_alignment */
  hsize_t         alignment;        /* alignment for H5Pset_alignment */
- H5F_fspace_strategy_t fs_strategy;	/* File space handling strategy */
- int		fs_persist; 		/* Free space section threshold */
- long		fs_threshold; 		/* Free space section threshold */
- long long 	fs_pagesize;   		/* File space page size */
+ H5F_fspace_strategy_t fs_strategy;    /* File space handling strategy */
+ int        fs_persist;         /* Free space section threshold */
+ long        fs_threshold;         /* Free space section threshold */
+ long long     fs_pagesize;           /* File space page size */
 } pack_opt_t;
 
 
@@ -150,10 +150,10 @@ int h5repack_end(pack_opt_t *options);
 int h5repack_verify(const char *in_fname, const char *out_fname, pack_opt_t *options);
 int h5repack_cmp_pl(const char *fname1, const char *fname2);
 
-/* Note: The below copy_named_datatype(), named_datatype_free(), copy_attr() 
- * and struct named_dt_t were located in h5repack_copy.c as static prior to 
- * bugfix1726. 
- * Made shared functions as copy_attr() was needed in h5repack_refs.c. 
+/* Note: The below copy_named_datatype(), named_datatype_free(), copy_attr()
+ * and struct named_dt_t were located in h5repack_copy.c as static prior to
+ * bugfix1726.
+ * Made shared functions as copy_attr() was needed in h5repack_refs.c.
  * However copy_attr() may be obsoleted when H5Acopy is available and put back
  * others to static in h5repack_copy.c.
  */
diff --git a/tools/src/h5repack/h5repack_copy.c b/tools/src/h5repack/h5repack_copy.c
index ec3e287..42fedbc 100644
--- a/tools/src/h5repack/h5repack_copy.c
+++ b/tools/src/h5repack/h5repack_copy.c
@@ -103,6 +103,7 @@ int copy_objects(const char* fnamein, const char* fnameout, pack_opt_t *options)
     hbool_t       in_persist;             /* Persist free-space status from input file */
     hsize_t       in_threshold;           /* Free-space section threshold from input file */
     hsize_t       in_pagesize;            /* File space page size from input file */
+    unsigned      crt_order_flags;        /* group creation order flag */
 
     /*-------------------------------------------------------------------------
      * open input file
@@ -116,6 +117,8 @@ int copy_objects(const char* fnamein, const char* fnameout, pack_opt_t *options)
     /* get user block size and file space strategy/persist/threshold */
     {
         hid_t fcpl_in; /* file creation property list ID for input file */
+        hid_t grp_in = -1;   /* group ID */
+        hid_t gcpl_in = -1;  /* group creation property list */
 
         if ((fcpl_in = H5Fget_create_plist(fidin)) < 0) {
             error_msg("failed to retrieve file creation property list\n");
@@ -127,17 +130,28 @@ int copy_objects(const char* fnamein, const char* fnameout, pack_opt_t *options)
             HGOTO_ERROR(FAIL, H5E_tools_min_id_g, "H5Pclose failed");
         }
 
-    /* If the -S option is not set, get "strategy" from the input file */
-    if(H5Pget_file_space_strategy(fcpl_in, &in_strategy, &in_persist, &in_threshold) < 0) {
-        error_msg("failed to retrieve file space strategy\n");
-        HGOTO_ERROR(FAIL, H5E_tools_min_id_g, "H5Pclose failed");
-    }
+        /* If the -S option is not set, get "strategy" from the input file */
+        if(H5Pget_file_space_strategy(fcpl_in, &in_strategy, &in_persist, &in_threshold) < 0) {
+            error_msg("failed to retrieve file space strategy\n");
+            HGOTO_ERROR(FAIL, H5E_tools_min_id_g, "H5Pclose failed");
+        }
 
-    /* If the -G option is not set, get "pagesize" from the input file */
-    if(H5Pget_file_space_page_size(fcpl_in, &in_pagesize) < 0) {
-        error_msg("failed to retrieve file space threshold\n");
-        HGOTO_ERROR(FAIL, H5E_tools_min_id_g, "H5Pclose failed");
-    }
+        /* If the -G option is not set, get "pagesize" from the input file */
+        if(H5Pget_file_space_page_size(fcpl_in, &in_pagesize) < 0) {
+            error_msg("failed to retrieve file space threshold\n");
+            HGOTO_ERROR(FAIL, H5E_tools_min_id_g, "H5Pclose failed");
+        }
+        /* open root group */
+        if ((grp_in = H5Gopen2(fidin, "/", H5P_DEFAULT)) < 0)
+            HGOTO_ERROR(FAIL, H5E_tools_min_id_g, "H5Gopen2 failed");
+
+        /* get root group creation property list */
+        if ((gcpl_in = H5Gget_create_plist(grp_in)) < 0)
+            HGOTO_ERROR(FAIL, H5E_tools_min_id_g, "H5Gget_create_plist failed");
+
+        /* query and set the group creation properties */
+        if (H5Pget_link_creation_order(gcpl_in, &crt_order_flags) < 0)
+            HGOTO_ERROR(FAIL, H5E_tools_min_id_g, "H5Pget_link_creation_order failed");
 
         if (H5Pclose(fcpl_in) < 0) {
             error_msg("failed to close property list\n");
@@ -309,6 +323,9 @@ print_user_block(fnamein, fidin);
         }
     }
 
+    if(H5Pset_link_creation_order(fcpl, crt_order_flags ) < 0)
+        HGOTO_ERROR(FAIL, H5E_tools_min_id_g, "H5Pset_link_creation_order failed");
+
     /* Set file space info to those from input file */
     set_strategy = in_strategy;
     set_persist = in_persist;
@@ -375,6 +392,8 @@ print_user_block(fnamein, fidin);
      *-------------------------------------------------------------------------
      */
 
+    /* Initialize indexing options */
+    h5trav_set_index(sort_by, sort_order);
     /* init table */
     trav_table_init(&travt);
 
diff --git a/tools/src/h5repack/h5repack_main.c b/tools/src/h5repack/h5repack_main.c
index 339dc83..31b1e14 100644
--- a/tools/src/h5repack/h5repack_main.c
+++ b/tools/src/h5repack/h5repack_main.c
@@ -31,7 +31,7 @@ const char *outfile = NULL;
  * Command-line options: The user can specify short or long-named
  * parameters.
  */
-static const char *s_opts = "hVvf:l:m:e:nLc:d:s:u:b:M:t:a:i:o:S:P:T:G:E";
+static const char *s_opts = "hVvf:l:m:e:nLc:d:s:u:b:M:t:a:i:o:S:P:T:G:q:z:E";
 static struct long_options l_opts[] = {
     { "help", no_arg, 'h' },
     { "version", no_arg, 'V' },
@@ -56,6 +56,8 @@ static struct long_options l_opts[] = {
     { "fs_persist", require_arg, 'P' },
     { "fs_threshold", require_arg, 'T' },
     { "fs_pagesize", require_arg, 'G' },
+    { "sort_by", require_arg, 'q' },
+    { "sort_order", require_arg, 'z' },
     { "enable-error-stack", no_arg, 'E' },
     { NULL, 0, '\0' }
 };
@@ -90,6 +92,8 @@ static void usage(const char *prog) {
     PRINTVALSTREAM(rawoutstream, "   -M A, --metadata_block_size=A  Metadata block size for H5Pset_meta_block_size\n");
     PRINTVALSTREAM(rawoutstream, "   -t T, --threshold=T     Threshold value for H5Pset_alignment\n");
     PRINTVALSTREAM(rawoutstream, "   -a A, --alignment=A     Alignment value for H5Pset_alignment\n");
+    PRINTVALSTREAM(rawoutstream, "   -q Q, --sort_by=Q       Sort groups and attributes by index Q\n");
+    PRINTVALSTREAM(rawoutstream, "   -z Z, --sort_order=Z    Sort groups and attributes by order Z\n");
     PRINTVALSTREAM(rawoutstream, "   -f FILT, --filter=FILT  Filter type\n");
     PRINTVALSTREAM(rawoutstream, "   -l LAYT, --layout=LAYT  Layout type\n");
     PRINTVALSTREAM(rawoutstream, "   -S FS_STRATEGY, --fs_strategy=FS_STRATEGY  File space management strategy for H5Pset_file_space_strategy\n");
@@ -103,6 +107,8 @@ static void usage(const char *prog) {
     PRINTVALSTREAM(rawoutstream, "    U - is a filename.\n");
     PRINTVALSTREAM(rawoutstream, "    T - is an integer\n");
     PRINTVALSTREAM(rawoutstream, "    A - is an integer greater than zero\n");
+    PRINTVALSTREAM(rawoutstream, "    Q - is the sort index type for the input file. It can be \"name\" or \"creation_order\" (default)\n");
+    PRINTVALSTREAM(rawoutstream, "    Z - is the sort order type for the input file. It can be \"descending\" or \"ascending\" (default)\n");
     PRINTVALSTREAM(rawoutstream, "    B - is the user block size, any value that is 512 or greater and is\n");
     PRINTVALSTREAM(rawoutstream, "        a power of 2 (1024 default)\n");
     PRINTVALSTREAM(rawoutstream, "    F - is the shared object header message type, any of <dspace|dtype|fill|\n");
@@ -370,6 +376,52 @@ done:
 }
 
 /*-------------------------------------------------------------------------
+ * Function:    set_sort_by
+ *
+ * Purpose: set the "by" form of sorting by translating from a string input
+ *          parameter to a H5_index_t return value
+ *          current sort values are [creation_order | name]
+ *
+ * Return: H5_index_t form of sort or H5_INDEX_UNKNOWN if none found
+ *-------------------------------------------------------------------------
+ */
+static H5_index_t
+set_sort_by(const char *form)
+{
+    H5_index_t idx_type = H5_INDEX_UNKNOWN;
+
+    if (HDstrcmp(form,"name")==0) /* H5_INDEX_NAME */
+        idx_type = H5_INDEX_NAME;
+    else if (HDstrcmp(form,"creation_order")==0) /* H5_INDEX_CRT_ORDER */
+        idx_type = H5_INDEX_CRT_ORDER;
+
+    return idx_type;
+}
+
+/*-------------------------------------------------------------------------
+ * Function:    set_sort_order
+ *
+ * Purpose: set the order of sorting by translating from a string input
+ *          parameter to a H5_iter_order_t return value
+ *          current order values are [ascending | descending ]
+ *
+ * Return: H5_iter_order_t form of order or H5_ITER_UNKNOWN if none found
+ *-------------------------------------------------------------------------
+ */
+static H5_iter_order_t
+set_sort_order(const char *form)
+{
+    H5_iter_order_t iter_order = H5_ITER_UNKNOWN;
+
+    if (HDstrcmp(form,"ascending")==0) /* H5_ITER_INC */
+        iter_order = H5_ITER_INC;
+    else if (HDstrcmp(form,"descending")==0) /* H5_ITER_DEC */
+        iter_order = H5_ITER_DEC;
+
+    return iter_order;
+}
+
+/*-------------------------------------------------------------------------
  * Function: parse_command_line
  *
  * Purpose: parse command line input
@@ -553,14 +605,14 @@ int parse_command_line(int argc, const char **argv, pack_opt_t* options)
                 break;
 
             case 'P':
-                options->fs_persist = HDatoi( opt_arg );
+                options->fs_persist = HDatoi(opt_arg);
                 if(options->fs_persist == 0)
                     /* To distinguish the "specified" zero value */
                     options->fs_persist = -1;
                 break;
 
             case 'T':
-                options->fs_threshold = HDatol( opt_arg );
+                options->fs_threshold = HDatol(opt_arg);
                 if(options->fs_threshold == 0)
                     /* To distinguish the "specified" zero value */
                     options->fs_threshold = -1;
@@ -570,12 +622,30 @@ int parse_command_line(int argc, const char **argv, pack_opt_t* options)
                 options->fs_pagesize = HDstrtoll(opt_arg, NULL, 0);
                 if(options->fs_pagesize == 0)
                     /* To distinguish the "specified" zero value */
-                   options->fs_pagesize = -1;	
+                   options->fs_pagesize = -1;
                 break;
 
-	    case 'E':
-		enable_error_stack = TRUE;
-		break;
+            case 'q':
+                if((sort_by = set_sort_by(opt_arg)) < 0) {
+                    error_msg(" failed to set sort by form <%s>\n", opt_arg);
+                    h5tools_setstatus(EXIT_FAILURE);
+                    ret_value = -1;
+                    goto done;
+                }
+                break;
+
+            case 'z':
+                if((sort_order = set_sort_order(opt_arg)) < 0) {
+                    error_msg(" failed to set sort order form <%s>\n", opt_arg);
+                    h5tools_setstatus(EXIT_FAILURE);
+                    ret_value = -1;
+                    goto done;
+                }
+                break;
+
+            case 'E':
+            enable_error_stack = TRUE;
+            break;
 
             default:
                 break;
@@ -644,6 +714,9 @@ int main(int argc, const char **argv)
     /* initialize options  */
     h5repack_init(&options, 0, FALSE);
 
+    /* Initialize default indexing options */
+    sort_by = H5_INDEX_CRT_ORDER;
+
     if (parse_command_line(argc, argv, &options) < 0)
         goto done;
 
diff --git a/tools/src/h5repack/h5repack_verify.c b/tools/src/h5repack/h5repack_verify.c
index 7e79aa7..4ff4875 100644
--- a/tools/src/h5repack/h5repack_verify.c
+++ b/tools/src/h5repack/h5repack_verify.c
@@ -55,11 +55,11 @@ h5repack_verify(const char *in_fname, const char *out_fname, pack_opt_t *options
     trav_table_t *travt = NULL;
     int          ok = 1;
     hid_t       fcpl_in     = -1;   /* file creation property for input file */
-    hid_t   	fcpl_out    = -1;   /* file creation property for output file */
-    H5F_fspace_strategy_t in_strategy, out_strategy;	/* file space handling strategy for in/output file */
-    hbool_t 	in_persist, out_persist;		/* free-space persist status for in/output file */
-    hsize_t	in_threshold, out_threshold;		/* free-space section threshold for in/output file */
-    hsize_t	in_pagesize, out_pagesize;		/* file space page size for input/output file */
+    hid_t       fcpl_out    = -1;   /* file creation property for output file */
+    H5F_fspace_strategy_t in_strategy, out_strategy;    /* file space handling strategy for in/output file */
+    hbool_t     in_persist, out_persist;        /* free-space persist status for in/output file */
+    hsize_t    in_threshold, out_threshold;        /* free-space section threshold for in/output file */
+    hsize_t    in_pagesize, out_pagesize;        /* file space page size for input/output file */
 
     /* open the output file */
     if((fidout = H5Fopen(out_fname, H5F_ACC_RDONLY, H5P_DEFAULT)) < 0 )
@@ -122,6 +122,8 @@ h5repack_verify(const char *in_fname, const char *out_fname, pack_opt_t *options
     if(options->all_filter == 1 || options->all_layout == 1)
     {
 
+        /* Initialize indexing options */
+        h5trav_set_index(sort_by, sort_order);
         /* init table */
         trav_table_init(&travt);
 
@@ -455,6 +457,9 @@ int h5repack_cmp_pl(const char *fname1,
     * get file table list of objects
     *-------------------------------------------------------------------------
     */
+    /* Initialize indexing options */
+    h5trav_set_index(sort_by, sort_order);
+    /* init table */
     trav_table_init(&trav);
     if(h5trav_gettable(fid1, trav) < 0)
         HGOTO_ERROR(FAIL, H5E_tools_min_id_g, "h5trav_gettable failed");
@@ -643,7 +648,7 @@ int verify_filters(hid_t pid, hid_t tid, int nfilters, filter_info_t *filter)
         {
 
             case H5Z_FILTER_NONE:
-        	break;
+            break;
 
             case H5Z_FILTER_SHUFFLE:
                 /* 1 private client value is returned by DCPL */