Added loading of the influenza.faa file.

4 files changed, 126 insertions, 8 deletions

diff --git a/src/aggregator.c b/src/aggregator.c
index 5fb9d4a..36ea18c 100644
--- a/src/aggregator.c
+++ b/src/aggregator.c
@@ -16,26 +16,26 @@ main ()
   /*
    * Create the HDF5 file.
    */
-  //  hid_t file_id = H5Fcreate (FILE, H5F_ACC_TRUNC, H5P_DEFAULT, H5P_DEFAULT);
+  hid_t file_id = H5Fcreate (FILE, H5F_ACC_TRUNC, H5P_DEFAULT, H5P_DEFAULT);
 
   /*
    * Load the supplementary protein data file.
    */
-  //  load_influenza_aa_dat (file_id);
+  load_influenza_aa_dat (file_id);
 
   /*
    * Load the FASTA protein sequence data file.
    */
-  //  load_influenza_faa (file_id);
+  load_influenza_faa (file_id);
 
   /*
    * Close the HD5 file.
    */
-  //  herr_t status = H5Fclose (file_id);
-  //  if (status < 0)
-  //    check_h5_error (status, __FILE__, __LINE__);
+  herr_t status = H5Fclose (file_id);
+  if (status < 0)
+    check_h5_error (status, __FILE__, __LINE__);
 
-  assign_protein_type (0);
+  // assign_protein_type (0);
 
   return 0;
 }
diff --git a/src/check_ncbi_error.c b/src/check_ncbi_error.c
index 8e1c3b2..6071d1a 100644
--- a/src/check_ncbi_error.c
+++ b/src/check_ncbi_error.c
@@ -1,4 +1,7 @@
 #include "check_ncbi_error.h"
+#include <error.h>
+#include <stdlib.h>
+#include <blast.h>
 
 void
 check_ncbi_error (ValNodePtr error_returns,
diff --git a/src/load_influenza_aa_dat.c b/src/load_influenza_aa_dat.c
index 9ee3c46..f0d9ee5 100644
--- a/src/load_influenza_aa_dat.c
+++ b/src/load_influenza_aa_dat.c
@@ -158,7 +158,7 @@ load_influenza_aa_dat (hid_t file_id)
 
       current_line++;
       char *running = strdup (line);
-      char *token;
+      char *token = NULL;
 
       /*
        * Parse the line, handling the case of empty fields represented
diff --git a/src/load_influenza_faa.c b/src/load_influenza_faa.c
index 8fd0cd7..61bb99d 100644
--- a/src/load_influenza_faa.c
+++ b/src/load_influenza_faa.c
@@ -1,9 +1,124 @@
 #include "load_influenza_faa.h"
+#include "check_error.h"
+#include "check_h5_error.h"
+#include "hdf5_hl.h"
+#include <string.h>
+#include <stdlib.h>
 
 void
 load_influenza_faa (hid_t file_id)
 {
+  typedef struct
+  {
+    int gi;
+    char gb[9];
+    char description[196];
+  } sequence_data;
 
+  size_t dst_size = sizeof (sequence_data);
+  size_t dst_offset[3] =
+    { HOFFSET (sequence_data, gi),
+      HOFFSET (sequence_data, gb),
+      HOFFSET (sequence_data, description)
+    };
+
+  sequence_data dst_buf[1];
+
+  size_t dst_sizes[3] = {
+    sizeof (dst_buf[0].gi),
+    sizeof (dst_buf[0].gb),
+    sizeof (dst_buf[0].description)
+  };
+
+  hid_t field_type[3];
+
+  field_type[0] = H5T_NATIVE_INT;
+
+  hid_t gb_type = H5Tcopy (H5T_C_S1);
+  H5Tset_size (gb_type, 9);
+  field_type[1] = gb_type;
+
+  hid_t description_type = H5Tcopy (H5T_C_S1);
+  H5Tset_size (description_type, 196);
+  field_type[2] = description_type;
+
+  const char *field_names[3] = { "GI",
+                                 "GB",
+                                 "Description" };
+
+  hsize_t chunk_size = 10;
+  int *fill_data = NULL;
+  int compress = 0;
+
+  sequence_data p_data;
+  FILE *dat = fopen ("/home/don/exp004/genomes/INFLUENZA/influenza.faa",
+                     "r");
+  if (dat == NULL)
+    check_error (__FILE__, __LINE__);
+  char *line = NULL;
+  size_t len = 0;
+  int current_line = 0;
+
+  while (getline (&line, &len, dat) != -1)
+    {
+      current_line++;      
+
+      // Header line.
+      if (line[0] == '>')
+        {
+          char *running = strdup (line);
+          char *token = NULL;
+          
+          // Eat the ">gi".
+          strsep (&running, "|");
+          
+          // GI value.
+          token = strsep (&running, "|");
+          p_data.gi = atoi (token);
+
+          // Eat the "gb"
+          strsep (&running, "|");
+
+          // GB value.
+          strncpy (p_data.gb, strsep(&running, "|"), sizeof (p_data.gb));
+
+          // Description value.
+          strncpy (p_data.description, strsep (&running, "|"),
+                   sizeof (p_data.description));
+
+          if (current_line == 1)
+            {
+              herr_t status = H5TBmake_table ("influenza.faa", file_id,
+                                              "influenza.faa", 3, 1, dst_size,
+                                              field_names, dst_offset,
+                                              field_type, chunk_size,
+                                              fill_data, compress, &p_data);
+              if (status < 0)
+                check_h5_error (status, __FILE__, __LINE__);
+            }
+          else
+            {
+              herr_t status =
+                H5TBappend_records (file_id, "influenza.faa", 1, dst_size,
+                                    dst_offset, dst_sizes, &p_data);
+              if (status < 0)
+                check_h5_error (status, __FILE__, __LINE__);
+            }
+
+          if (running)
+            free (running);
+
+        }
+      
+    }
+
+  if (line)
+    free (line);
+
+  fclose (dat);
+
+  H5Tclose (gb_type);
+  H5Tclose (description_type);
 
   return;
 }