summaryrefslogtreecommitdiffstats
Side-by-side diff
Diffstat
-rw-r--r--src/aggregator.c2
-rw-r--r--src/assign_protein_type.c7
-rw-r--r--src/assign_protein_type.h3
-rw-r--r--src/check_error.c2
-rw-r--r--src/check_error.h3
-rw-r--r--src/check_h5_error.c2
-rw-r--r--src/check_h5_error.h2
-rw-r--r--src/check_ncbi_error.c3
-rw-r--r--src/check_ncbi_error.h2
-rw-r--r--src/load_influenza_aa_dat.c259
-rw-r--r--src/load_influenza_aa_dat.h3
-rw-r--r--src/load_influenza_faa.c2
-rw-r--r--src/load_influenza_faa.h3
13 files changed, 145 insertions, 148 deletions
diff --git a/src/load_influenza_aa_dat.c b/src/load_influenza_aa_dat.c
index 91ef415..9ee3c46 100644
--- a/src/load_influenza_aa_dat.c
+++ b/src/load_influenza_aa_dat.c
@@ -46,60 +46,63 @@ load_influenza_aa_dat (hid_t file_id)
* memory."
*/
size_t dst_size = sizeof (supplementary_data);
- size_t dst_offset[NFIELDS] = { HOFFSET ( supplementary_data, genbank_accession_number ),
- HOFFSET ( supplementary_data, host ),
- HOFFSET ( supplementary_data, genome_segment_number ),
- HOFFSET ( supplementary_data, subtype ),
- HOFFSET ( supplementary_data, country ),
- HOFFSET ( supplementary_data, year ),
- HOFFSET ( supplementary_data, sequence_length ),
- HOFFSET ( supplementary_data, virus_name ),
- HOFFSET ( supplementary_data, age ),
- HOFFSET ( supplementary_data, gender ),
- HOFFSET ( supplementary_data, full_length_indicator )};
+ size_t dst_offset[NFIELDS] =
+ { HOFFSET (supplementary_data, genbank_accession_number),
+ HOFFSET (supplementary_data, host),
+ HOFFSET (supplementary_data, genome_segment_number),
+ HOFFSET (supplementary_data, subtype),
+ HOFFSET (supplementary_data, country),
+ HOFFSET (supplementary_data, year),
+ HOFFSET (supplementary_data, sequence_length),
+ HOFFSET (supplementary_data, virus_name),
+ HOFFSET (supplementary_data, age),
+ HOFFSET (supplementary_data, gender),
+ HOFFSET (supplementary_data, full_length_indicator)
+ };
supplementary_data dst_buf[1];
- size_t dst_sizes[NFIELDS] = { sizeof ( dst_buf[0].genbank_accession_number ),
- sizeof ( dst_buf[0].host ),
- sizeof ( dst_buf[0].genome_segment_number ),
- sizeof ( dst_buf[0].subtype ),
- sizeof ( dst_buf[0].country ),
- sizeof ( dst_buf[0].year ),
- sizeof ( dst_buf[0].sequence_length ),
- sizeof ( dst_buf[0].virus_name ),
- sizeof ( dst_buf[0].age ),
- sizeof ( dst_buf[0].gender ),
- sizeof ( dst_buf[0].full_length_indicator)};
+ size_t dst_sizes[NFIELDS] = { sizeof (dst_buf[0].genbank_accession_number),
+ sizeof (dst_buf[0].host),
+ sizeof (dst_buf[0].genome_segment_number),
+ sizeof (dst_buf[0].subtype),
+ sizeof (dst_buf[0].country),
+ sizeof (dst_buf[0].year),
+ sizeof (dst_buf[0].sequence_length),
+ sizeof (dst_buf[0].virus_name),
+ sizeof (dst_buf[0].age),
+ sizeof (dst_buf[0].gender),
+ sizeof (dst_buf[0].full_length_indicator)
+ };
/*
* Map the native types to HDF5 types for each field.
*/
hid_t field_type[NFIELDS];
- hid_t genbank_accession_number_type = H5Tcopy ( H5T_C_S1 );
- H5Tset_size ( genbank_accession_number_type, 9 );
+ hid_t genbank_accession_number_type = H5Tcopy (H5T_C_S1);
+ H5Tset_size (genbank_accession_number_type, 9);
field_type[0] = genbank_accession_number_type;
- hid_t host_type = H5Tcopy ( H5T_C_S1 );
- H5Tset_size ( host_type, 15 );
+ hid_t host_type = H5Tcopy (H5T_C_S1);
+ H5Tset_size (host_type, 15);
field_type[1] = host_type;
-
+
field_type[2] = H5T_NATIVE_INT;
- hid_t subtype_type = H5Tcopy ( H5T_C_S1 );
- H5Tset_size (subtype_type, 7 );
+ hid_t subtype_type = H5Tcopy (H5T_C_S1);
+ H5Tset_size (subtype_type, 7);
field_type[3] = subtype_type;
- hid_t country_type = H5Tcopy ( H5T_C_S1 );
- H5Tset_size (country_type, 25 );
+ hid_t country_type = H5Tcopy (H5T_C_S1);
+ H5Tset_size (country_type, 25);
field_type[4] = country_type;
field_type[5] = H5T_NATIVE_INT;
field_type[6] = H5T_NATIVE_INT;
- hid_t virus_name_type = H5Tcopy ( H5T_C_S1 );
+ hid_t virus_name_type = H5Tcopy (H5T_C_S1);
H5Tset_size (virus_name_type, 196);
field_type[7] = virus_name_type;
@@ -118,18 +121,18 @@ load_influenza_aa_dat (hid_t file_id)
/*
* Labels used for the fields in the table.
*/
- const char *field_names[NFIELDS] =
- { "GenBank accession number",
- "Host",
- "Genome segment number",
- "Subtype",
- "Country",
- "Year",
- "Sequence length",
- "Virus name",
- "Age",
- "Gender",
- "Full-length Indicator" };
+ const char *field_names[NFIELDS] = { "GenBank accession number",
+ "Host",
+ "Genome segment number",
+ "Subtype",
+ "Country",
+ "Year",
+ "Sequence length",
+ "Virus name",
+ "Age",
+ "Gender",
+ "Full-length Indicator"
+ };
/*
* Table storage options.
@@ -142,7 +145,7 @@ load_influenza_aa_dat (hid_t file_id)
* Insert the records.
*/
supplementary_data p_data;
- FILE* dat = fopen ("/home/don/exp004/genomes/INFLUENZA/influenza_aa.dat",
+ FILE *dat = fopen ("/home/don/exp004/genomes/INFLUENZA/influenza_aa.dat",
"r");
if (dat == NULL)
check_error (__FILE__, __LINE__);
@@ -150,89 +153,89 @@ load_influenza_aa_dat (hid_t file_id)
size_t len = 0;
int current_line = 0;
- while (getline (&line, &len, dat) != -1) {
-
- current_line++;
- char *running = strdup (line);
- char *token;
-
- /*
- * Parse the line, handling the case of empty fields represented
- * by sequential delimiters.
- */
- strncpy(p_data.genbank_accession_number, strsep (&running, "\t"),
- sizeof(p_data.genbank_accession_number));
-
- strncpy(p_data.host, strsep (&running, "\t"),
- sizeof(p_data.host));
-
- token = strsep (&running, "\t");
- if (strcmp (token, "\0") == 0)
- p_data.genome_segment_number = 0;
- else
- p_data.genome_segment_number = atoi(token);
-
- strncpy(p_data.subtype, strsep (&running, "\t"),
- sizeof(p_data.subtype));
-
- strncpy(p_data.country, strsep (&running, "\t"),
- sizeof(p_data.country));
-
- /*
- * 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)
- p_data.sequence_length = 0;
- else
- p_data.sequence_length = atoi(token);
-
- strncpy(p_data.virus_name, strsep (&running, "\t"),
- sizeof(p_data.virus_name));
-
- strncpy(p_data.age, strsep (&running, "\t"),
- sizeof(p_data.age));
-
- strncpy(p_data.gender, strsep (&running, "\t"),
- sizeof(p_data.gender));
-
- strncpy(p_data.full_length_indicator, strsep (&running, "\t"),
- sizeof(p_data.full_length_indicator));
-
- if (current_line == 1)
- {
- herr_t status = 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);
- if (status < 0)
- check_h5_error (status, __FILE__, __LINE__);
- }
- else
- {
- herr_t status = H5TBappend_records (file_id, TABLE_NAME, 1, dst_size,
- dst_offset, dst_sizes, &p_data);
- if (status < 0)
- check_h5_error (status, __FILE__, __LINE__);
- }
-
- if (running)
- free (running);
-
- }
-
+ while (getline (&line, &len, dat) != -1)
+ {
+
+ current_line++;
+ char *running = strdup (line);
+ char *token;
+
+ /*
+ * Parse the line, handling the case of empty fields represented
+ * by sequential delimiters.
+ */
+ strncpy (p_data.genbank_accession_number, strsep (&running, "\t"),
+ sizeof (p_data.genbank_accession_number));
+
+ strncpy (p_data.host, strsep (&running, "\t"), sizeof (p_data.host));
+
+ token = strsep (&running, "\t");
+ if (strcmp (token, "\0") == 0)
+ p_data.genome_segment_number = 0;
+ else
+ p_data.genome_segment_number = atoi (token);
+
+ strncpy (p_data.subtype, strsep (&running, "\t"),
+ sizeof (p_data.subtype));
+
+ strncpy (p_data.country, strsep (&running, "\t"),
+ sizeof (p_data.country));
+
+ /*
+ * 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)
+ p_data.sequence_length = 0;
+ else
+ p_data.sequence_length = atoi (token);
+
+ strncpy (p_data.virus_name, strsep (&running, "\t"),
+ sizeof (p_data.virus_name));
+
+ strncpy (p_data.age, strsep (&running, "\t"), sizeof (p_data.age));
+
+ strncpy (p_data.gender, strsep (&running, "\t"),
+ sizeof (p_data.gender));
+
+ strncpy (p_data.full_length_indicator, strsep (&running, "\t"),
+ sizeof (p_data.full_length_indicator));
+
+ if (current_line == 1)
+ {
+ herr_t status = 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);
+ if (status < 0)
+ check_h5_error (status, __FILE__, __LINE__);
+ }
+ else
+ {
+ herr_t status =
+ H5TBappend_records (file_id, TABLE_NAME, 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);

Valid XHTML 1.0 Strict

Copyright © 2009 Don Pellegrino All Rights Reserved.