static int determine_outfp(void *data, GtError *err)
{
GtOutputFileInfo *ofi = (GtOutputFileInfo*) data;
GtFileMode file_mode;
int had_err = 0;
gt_error_check(err);
gt_assert(ofi);
if (!gt_str_length(ofi->output_filename))
*ofi->outfp = NULL; /* no output file given -> use stdout */
else { /* outputfile given -> create generic file pointer */
gt_assert(!(ofi->gzip && ofi->bzip2));
if (ofi->gzip)
file_mode = GT_FILE_MODE_GZIP;
else if (ofi->bzip2)
file_mode = GT_FILE_MODE_BZIP2;
else
file_mode = GT_FILE_MODE_UNCOMPRESSED;
if (file_mode != GT_FILE_MODE_UNCOMPRESSED &&
strcmp(gt_str_get(ofi->output_filename) +
gt_str_length(ofi->output_filename) -
strlen(gt_file_mode_suffix(file_mode)),
gt_file_mode_suffix(file_mode))) {
gt_warning("output file '%s' doesn't have correct suffix '%s', appending "
"it", gt_str_get(ofi->output_filename),
gt_file_mode_suffix(file_mode));
gt_str_append_cstr(ofi->output_filename, gt_file_mode_suffix(file_mode));
}
if (!ofi->force && gt_file_exists(gt_str_get(ofi->output_filename))) {
gt_error_set(err, "file \"%s\" exists already, use option -%s to "
"overwrite", gt_str_get(ofi->output_filename),
GT_FORCE_OPT_CSTR);
had_err = -1;
}
if (!had_err) {
*ofi->outfp = gt_file_xopen_file_mode(file_mode,
gt_str_get(ofi->output_filename),
"w");
gt_assert(*ofi->outfp);
}
}
return had_err;
}
static int store_in_subset_file(void *data, GthSA *sa,
const char *outputfilename, GtError *err)
{
Store_in_subset_file_data *store_in_subset_file_data =
(Store_in_subset_file_data*) data;
double split_determing_percentage = 0.0;
unsigned long filenum;
char filenamesuffix[4];
int had_err = 0;
gt_error_check(err);
/* filter before we do any further processing */
if (gth_sa_filter_filter_sa(store_in_subset_file_data->sa_filter, sa)) {
/* and free it afterwards */
gth_sa_delete(sa);
/* discard */
return 0;
}
/* check whether we got a new output file to process */
if (!store_in_subset_file_data->current_outputfilename) {
store_in_subset_file_data->current_outputfilename =
gt_cstr_dup(outputfilename);
}
else if (strcmp(store_in_subset_file_data->current_outputfilename,
outputfilename)) {
/* close current output files */
close_output_files(store_in_subset_file_data);
gt_free(store_in_subset_file_data->current_outputfilename);
}
/* determine in which file the current sa needs to be put */
switch (store_in_subset_file_data->gthsplitinfo->splitmode) {
case ALIGNMENTSCORE_SPLIT:
split_determing_percentage = gth_sa_score(sa);
strcpy(filenamesuffix, "scr");
break;
case COVERAGE_SPLIT:
split_determing_percentage = gth_sa_coverage(sa);
strcpy(filenamesuffix, "cov");
break;
default: gt_assert(0);
}
gt_assert(split_determing_percentage >= 0.0);
/* XXX: change into an assertion when coverage problem is fixed */
if (split_determing_percentage > 1.0)
split_determing_percentage = 1.0;
if (split_determing_percentage == 1.0)
filenum = store_in_subset_file_data->num_of_subset_files - 1;
else {
filenum = floor(split_determing_percentage * 100.0 /
store_in_subset_file_data->gthsplitinfo->range);
}
gt_assert(filenum < store_in_subset_file_data->num_of_subset_files);
/* make sure the file exists and is open */
if (!store_in_subset_file_data->subset_files[filenum]) {
gt_assert(store_in_subset_file_data->subset_filenames[filenum] == NULL);
store_in_subset_file_data->subset_filenames[filenum] = gt_str_new();
gt_str_append_cstr_nt(store_in_subset_file_data->subset_filenames[filenum],
outputfilename,
gt_file_basename_length(outputfilename));
gt_str_append_char(store_in_subset_file_data->subset_filenames[filenum],
'.');
gt_str_append_cstr(store_in_subset_file_data->subset_filenames[filenum],
filenamesuffix);
gt_str_append_ulong(store_in_subset_file_data->subset_filenames[filenum],
filenum *
store_in_subset_file_data->gthsplitinfo->range);
gt_str_append_char(store_in_subset_file_data->subset_filenames[filenum],
'-');
gt_str_append_ulong(store_in_subset_file_data->subset_filenames[filenum],
(filenum + 1) *
store_in_subset_file_data->gthsplitinfo->range);
gt_str_append_cstr(store_in_subset_file_data->subset_filenames[filenum],
gt_file_mode_suffix(store_in_subset_file_data
->gthsplitinfo->file_mode));
/* if not disabled by -force, check if file already exists */
if (!store_in_subset_file_data->gthsplitinfo->force) {
store_in_subset_file_data->subset_files[filenum] =
gt_file_open(store_in_subset_file_data->gthsplitinfo->file_mode,
gt_str_get(store_in_subset_file_data
->subset_filenames[filenum]), "r", NULL);
if (store_in_subset_file_data->subset_files[filenum]) {
gt_error_set(err, "file \"%s\" exists already. use option -%s to "
"overwrite", gt_str_get(store_in_subset_file_data
->subset_filenames[filenum]),
GT_FORCE_OPT_CSTR);
had_err = -1;
}
}
if (!had_err) {
/* open split file for writing */
store_in_subset_file_data->subset_files[filenum] =
gt_file_xopen_file_mode(store_in_subset_file_data->gthsplitinfo
->file_mode,
gt_str_get(store_in_subset_file_data
->subset_filenames[filenum]), "w");
/* store XML header in file */
gth_xml_show_leader(true,
store_in_subset_file_data->subset_files[filenum]);
}
}
/* put it there */
if (!had_err) {
gth_xml_inter_sa_visitor_set_outfp(store_in_subset_file_data->sa_visitor,
store_in_subset_file_data
->subset_files[filenum]);
gth_sa_visitor_visit_sa(store_in_subset_file_data->sa_visitor, sa);
}
/* adjust counter */
if (!had_err)
store_in_subset_file_data->subset_file_sa_counter[filenum]++;
/* and free it afterwards */
gth_sa_delete(sa);
return had_err;
}
static int split_fasta_file(const char *filename, unsigned long max_filesize,
bool force, GtError *err)
{
GtFile *srcfp = NULL, *destfp = NULL;
GtStr *destfilename = NULL;
unsigned long filenum = 0, bytecount = 0, separator_pos;
int read_bytes, had_err = 0;
char buf[BUFSIZ];
gt_error_check(err);
gt_assert(filename && max_filesize);
/* open source file */
srcfp = gt_file_xopen(filename, "r");
gt_assert(srcfp);
/* read start characters */
if ((read_bytes = gt_file_xread(srcfp, buf, BUFSIZ)) == 0) {
gt_error_set(err, "file \"%s\" is empty", filename);
had_err = -1;
}
bytecount += read_bytes;
/* make sure the file is in fasta format */
if (!had_err && buf[0] != '>') {
gt_error_set(err, "file is not in FASTA format");
had_err = -1;
}
if (!had_err) {
/* open destination file */
destfilename = gt_str_new();
gt_str_append_cstr_nt(destfilename, filename,
gt_file_basename_length(filename));
gt_str_append_char(destfilename, '.');
gt_str_append_ulong(destfilename, ++filenum);
gt_str_append_cstr(destfilename,
gt_file_mode_suffix(gt_file_mode(srcfp)));
if (!(destfp = gt_outputfile_xopen_forcecheck(gt_str_get(destfilename), "w",
force, err))) {
had_err = -1;
}
if (!had_err)
gt_file_xwrite(destfp, buf, read_bytes);
while (!had_err &&
(read_bytes = gt_file_xread(srcfp, buf, BUFSIZ)) != 0) {
if (bytecount + read_bytes > max_filesize) {
int offset = bytecount < max_filesize ? max_filesize - bytecount : 0;
if ((separator_pos = buf_contains_separator(buf, offset, read_bytes))) {
separator_pos--;
gt_assert(separator_pos < read_bytes);
if (separator_pos)
gt_file_xwrite(destfp, buf, separator_pos);
/* close current file */
gt_file_delete(destfp);
/* open new file */
gt_str_reset(destfilename);
gt_str_append_cstr_nt(destfilename, filename,
gt_file_basename_length(filename));
gt_str_append_char(destfilename, '.');
gt_str_append_ulong(destfilename, ++filenum);
gt_str_append_cstr(destfilename,
gt_file_mode_suffix(gt_file_mode(srcfp)));
if (!(destfp =
gt_outputfile_xopen_forcecheck(gt_str_get(destfilename), "w",
force, err))) {
had_err = -1;
break;
}
bytecount = read_bytes - separator_pos; /* reset */
gt_assert(buf[separator_pos] == '>');
gt_file_xwrite(destfp, buf + separator_pos,
read_bytes - separator_pos);
continue;
}
}
bytecount += read_bytes;
gt_file_xwrite(destfp, buf, read_bytes);
}
}
/* free */
gt_str_delete(destfilename);
/* close current file */
gt_file_delete(destfp);
/* close source file */
gt_file_delete(srcfp);
return had_err;
}