Modified to load the blast scores from a text file into the HDF5 file.

1 files changed, 45 insertions, 8 deletions

diff --git a/src/load/load_blast_scores.c b/src/load/load_blast_scores.c
index 42e6bd9..e6a6fd3 100644
--- a/src/load/load_blast_scores.c
+++ b/src/load/load_blast_scores.c
@@ -1,3 +1,4 @@
+#define _GNU_SOURCE
 #include "error/check_error.h"
 #include "error/check_h5_error.h"
 #include "model/blast_scores_data.h"
@@ -6,7 +7,19 @@
 #include <hdf5_hl.h>
 #include <string.h>
 #include <stdlib.h>
-
+#include <stdio.h>
+
+/*
+ * A simple sanity check can be performed by comparing the HDF5
+ * content with the content of the input text file.  Data block output
+ * from the first command should be equal to the first 5 lines of the
+ * input file while data block output from the second command should
+ * be equal to the last 5 lines of the input file.
+ *
+ *   h5dump --dataset=blast --start "0" --count "5" influenza.h5
+ *
+ *   h5dump --dataset=blast --start "5749892" --count "5" influenza.h5
+ */
 void
 load_blast_scores (hid_t file_id, const char *file_name)
 {
@@ -21,14 +34,15 @@ load_blast_scores (hid_t file_id, const char *file_name)
   int *fill_data = NULL;
   int compress = 0;
 
-  blast_scores_data p_data[1000];
+  blast_scores_data p_data[10000];
   FILE *dat = fopen (file_name, "r");
   if (dat == NULL)
     check_error (__FILE__, __LINE__);
   char *line = NULL;
   size_t len = 0;
-  int current_line = 0;
   int i = -1;
+  int written = 0;
+  int current_line = 0;
 
   while (getline (&line, &len, dat) != -1)
     {
@@ -65,8 +79,10 @@ load_blast_scores (hid_t file_id, const char *file_name)
       token = strsep (&running, ",");
       p_data[i].evalue = strtod (token, NULL);
 
-      if (current_line == 1)
+      if (written == 0)
         {
+
+          printf ("Creating blast table.\n");
           
           const char *blast_scores_data_field_names[BLAST_SCORES_DATA_FIELD_NUM] =
             BLAST_SCORES_DATA_FIELD_NAMES;
@@ -84,13 +100,17 @@ load_blast_scores (hid_t file_id, const char *file_name)
           if (status < 0)
             check_h5_error (__FILE__, __LINE__);
 
+          written = 1;
+          i = -1;
+          current_line = 0;
+
         }
 
-      if ((i % 1000 == 0) && (i > 0))
+      if ((current_line % 10000 == 0) && (current_line > 0))
         {
-
+          
           herr_t status =
-            H5TBappend_records (file_id, "blast", 1000,
+            H5TBappend_records (file_id, "blast", 10000,
                                 dst_size, dst_offset, dst_sizes,
                                 &p_data[0]);
           if(status < 0)
@@ -100,9 +120,11 @@ load_blast_scores (hid_t file_id, const char *file_name)
           if (status < 0)
             check_h5_error (__FILE__, __LINE__);
 
-          printf ("Processed %i of records.\n", current_line);
+          printf ("Processed %i records.\n", current_line);
 
+          written += 10000;
           i = -1;
+
         }
 
       if (running)
@@ -110,6 +132,21 @@ load_blast_scores (hid_t file_id, const char *file_name)
 
     } // End for each line of the input file.
 
+  if (i >= 0)
+    {
+      herr_t status =
+        H5TBappend_records (file_id, "blast", i+1,
+                            dst_size, dst_offset, dst_sizes, &p_data[0]);
+      if (status < 0)
+        check_h5_error (__FILE__, __LINE__);
+ 
+      printf ("Wrote %i records.\n", written + i + 1);
+    }
+
+  herr_t status = H5Fflush (file_id, H5F_SCOPE_GLOBAL);
+  if (status < 0)
+    check_h5_error (__FILE__, __LINE__);
+
   if (line)
     free (line);