From 0871f6cf645c20673e45946c3ba3ddaa2ffb47aa Mon Sep 17 00:00:00 2001
From: Don Pellegrino <don@drexel.edu>
Date: Mon, 18 Jan 2010 18:46:10 +0000
Subject: Added error handling for the case where a GI record is not found in the BLAST database.  Added writing of the last bin of record updates to the file.

---
diff --git a/src/assign_protein_type.c b/src/assign_protein_type.c
index 166e787..ec3a959 100644
--- a/src/assign_protein_type.c
+++ b/src/assign_protein_type.c
@@ -83,6 +83,7 @@ assign_protein_type (hid_t file_id)
    * Assign protein types to records for which the field is empty.
    */
   printf ("Records to process: %i\n", (int)nrecords);
+  bool updates_pending = false;
   for (int i = 0; i < nrecords; i++)
     {
 
@@ -92,6 +93,14 @@ assign_protein_type (hid_t file_id)
 	 */
 	Int4 sequence_number = readdb_gi2seq (seqdb, dst_buf[i].gi, NULL);
 	BioseqPtr bsp = readdb_get_bioseq (seqdb, sequence_number);
+	if (bsp == NULL) 
+	  {
+	    error_at_line (EXIT_FAILURE, 0, __FILE__, __LINE__,
+			   "Unable to find BLAST record for gi|%i.  Ensure the BLAST "
+			   "database is up-to-date with the HDF5 record set.  See the "
+			   "BLAST formatdb.log file for details.\n",
+			   dst_buf[i].gi);
+	  }
 	
 	SeqAlignPtr seqalign = BioseqBlastEngine (bsp,
 						  "blastp",
@@ -123,6 +132,7 @@ assign_protein_type (hid_t file_id)
 	    SeqIdWrite (target_id, target_id_buf, PRINTID_FASTA_SHORT, BUFFER_LEN);
 	    strncpy (dst_buf[i].protein_type, &target_id_buf[6], 
 		     sizeof (dst_buf[i].protein_type));
+	    updates_pending = true;
 	  }
 
 	/*
@@ -140,25 +150,56 @@ assign_protein_type (hid_t file_id)
 	seqalign = SeqAlignSetFree (seqalign);
 	bsp = BioseqFree (bsp);
 
-	/*
-	 * Write the data out to the file.
-	 */
-	if ( (i % 1000 == 0) && (i > 0) )
-	  {
-	    status = H5TBwrite_records (file_id, "influenza.faa", i - 1000, 1000,
-					dst_size, dst_offset, dst_sizes, 
-					&dst_buf[i-1000]);
-	    if (status < 0)
-	      check_h5_error (status, __FILE__, __LINE__);
+      }
 
-	    status = H5Fflush (file_id, H5F_SCOPE_GLOBAL);
-	    if (status < 0)
-	      check_h5_error (status, __FILE__, __LINE__);
+      /*
+       * Write the data out to the file.
+       */
+      if ( (i % 1000 == 0) && (i > 0) && updates_pending)
+	{
+	  status = H5TBwrite_records (file_id, "influenza.faa", i - 1000, 1000,
+				      dst_size, dst_offset, dst_sizes, 
+				      &dst_buf[i-1000]);
+	  if (status < 0)
+	    check_h5_error (status, __FILE__, __LINE__);
+	  
+	  status = H5Fflush (file_id, H5F_SCOPE_GLOBAL);
+	  if (status < 0)
+	    check_h5_error (status, __FILE__, __LINE__);
+	  
+	  updates_pending = false;
+
+	  printf ("Processed %i of %i records.\n", i, (int)nrecords);
+	}
+      
+    }
 
-	    printf ("Processed %i of %i records.\n", i, (int)nrecords);
-	  }
-      }
+  /*
+   * Write out records from the last bin if it was less than 1000
+   * records in size.
+   */
+  if (updates_pending)
+    {
+      if ((int)nrecords < 1000) 
+	{
+	  status = H5TBwrite_records (file_id, "influenza.faa", 0, nrecords,
+				      dst_size, dst_offset, dst_sizes,
+				      dst_buf);
+	}
+      else
+	{
+	  status = H5TBwrite_records (file_id, "influenza.faa", nrecords - 1000, 1000,
+				      dst_size, dst_offset, dst_sizes, 
+				      &dst_buf[nrecords-1000]);
+	}
+      if (status < 0)
+	check_h5_error (status, __FILE__, __LINE__);
+      
+      status = H5Fflush (file_id, H5F_SCOPE_GLOBAL);
+      if (status < 0)
+	check_h5_error (status, __FILE__, __LINE__);
       
+      updates_pending = false;
     }
   
   free (dst_buf);
--
cgit v0.8.3.1-22-g547a