Modified to convert the year field from text to numeric during the load.

1 files changed, 17 insertions, 9 deletions

diff --git a/src/load_influenza_aa_dat.c b/src/load_influenza_aa_dat.c
index 5af8a72..493c7db 100644
--- a/src/load_influenza_aa_dat.c
+++ b/src/load_influenza_aa_dat.c
@@ -11,7 +11,7 @@
 #include <stdlib.h>
 
 #define NFIELDS (hsize_t) 11
-#define TABLE_NAME "Protein Sequences"
+#define TABLE_NAME "influenza_aa.dat"
 
 void
 load_influenza_aa_dat (hid_t file_id)
@@ -26,7 +26,7 @@ load_influenza_aa_dat (hid_t file_id)
     int genome_segment_number;
     char subtype[7];
     char country[25];
-    char year[8];
+    int year;
     int sequence_length;
     char virus_name[196];
     char age[17];
@@ -93,9 +93,7 @@ load_influenza_aa_dat (hid_t file_id)
   H5Tset_size (country_type, 25 );
   field_type[4] = country_type;
 
-  hid_t year_type = H5Tcopy ( H5T_C_S1 );
-  H5Tset_size (year_type, 8);
-  field_type[5] = year_type;
+  field_type[5] = H5T_NATIVE_INT;
 
   field_type[6] = H5T_NATIVE_INT;
 
@@ -175,8 +173,19 @@ load_influenza_aa_dat (hid_t file_id)
     strncpy(p_data.country, strsep (&running, "\t"),
             sizeof(p_data.country));
     
-    strncpy (p_data.year, strsep (&running, "\t"),
-             sizeof(p_data.year));
+    /*
+     * Convert the year field from text to numeric.  Unknown and empty
+     * values are assigned a numeric value of zero.
+     */
+    token = strsep (&running, "\t");
+    if (strcmp (token, "\0") == 0)
+      p_data.year = 0;
+    else if (strcmp (token, "unknown") == 0)
+      p_data.year = 0;
+    else if (strcmp (token, "NON") == 0)
+      p_data.year = 0;
+    else
+      p_data.year = atoi(token);
 
     token = strsep (&running, "\t");
     if (strcmp (token, "\0") == 0)
@@ -197,7 +206,7 @@ load_influenza_aa_dat (hid_t file_id)
             sizeof(p_data.full_length_indicator));
 
     if (current_line == 1)     
-      H5TBmake_table ("Protein Sequences", file_id, TABLE_NAME,NFIELDS,1,
+      H5TBmake_table ("influenza_aa.dat", file_id, TABLE_NAME,NFIELDS,1,
                       dst_size,field_names, dst_offset, field_type,
                       chunk_size, fill_data, compress, &p_data);
     else     
@@ -218,7 +227,6 @@ load_influenza_aa_dat (hid_t file_id)
   H5Tclose (host_type);
   H5Tclose (subtype_type);
   H5Tclose (country_type);
-  H5Tclose (year_type);
   H5Tclose (virus_name_type);
   H5Tclose (age_type);
   H5Tclose (gender_type);