-
-/*
- * fastq-grep: regular expression searches of the sequences within a fastq file
+/*
+ * This file is part of fastq-tools.
+ *
+ * Copyright (c) 2011 by Daniel C. Jones <dcjones@cs.washington.edu>
*
- * Febuary 2011 / Daniel Jones <dcjones@cs.washington.edu>
+ * fastq-grep :
+ * Regular expression searches of the sequences within a FASTQ file.
*
*/
+#include "common.h"
+#include "parse.h"
#include <stdio.h>
#include <string.h>
#include <getopt.h>
#include <zlib.h>
#include <pcre.h>
-#include "kseq.h"
-KSEQ_INIT(gzFile, gzread)
-
#if defined(MSDOS) || defined(OS2) || defined(WIN32) || defined(__CYGWIN__)
# include <fcntl.h>
#endif
+static const char* prog_name = "fastq-grep";
+
+
void print_help()
{
- fprintf( stderr,
+ fprintf(stdout,
"fastq-grep [OPTION]... PATTERN [FILE]...\n"
"Search for PATTERN in the read sequences in each FILE or standard input.\n"
"PATTERN, by default, is a perl compatible regular expression.\n\n"
"Options:\n"
-" -h, --help print this message\n"
+" -i, --id match the read id (by default, sequence is matched)\n"
" -v, --invert-match select nonmatching entries\n"
+" -c, --count output only the number of matching sequences\n"
+" -h, --help print this message\n"
+" -V, --version output version information and exit\n"
);
}
static int invert_flag;
-static int help_flag;
-static int zout_flag;
+static int count_flag;
+static int id_flag;
-void print_fastq_entry( FILE* fout, kseq_t* seq )
-{
- fprintf(fout, "@%s\n%s\n+%s\n%s\n",
- seq->name.s,
- seq->seq.s,
- seq->comment.s,
- seq->qual.s );
-}
-
-
-void fastq_grep( gzFile fin, FILE* fout, pcre* re )
+void fastq_grep(FILE* fin, FILE* fout, pcre* re)
{
int rc;
int ovector[3];
+ size_t count = 0;
+
+ fastq_t* fqf = fastq_open(fin);
+ seq_t* seq = fastq_alloc_seq();
- kseq_t* seq;
- seq = kseq_init(fin);
+ while (fastq_next(fqf, seq)) {
- while (kseq_read(seq) >= 0) {
rc = pcre_exec(re, /* pattern */
- NULL, /* extre data */
- seq->seq.s, /* subject */
- seq->seq.l, /* subject length */
+ NULL, /* extra data */
+ id_flag ? seq->id1.s : seq->seq.s,
+ id_flag ? seq->id1.n : seq->seq.n,
0, /* subject offset */
0, /* options */
ovector, /* output vector */
3 ); /* output vector length */
- if ((invert_flag && rc == PCRE_ERROR_NOMATCH) || rc >= 0) {
- print_fastq_entry( fout, seq );
+ if ((invert_flag && rc == PCRE_ERROR_NOMATCH) || (!invert_flag && rc >= 0)) {
+ if (count_flag) count++;
+ else fastq_print(fout, seq);
}
}
- kseq_destroy(seq);
+ fastq_free_seq(seq);
+ fastq_close(fqf);
+
+ if (count_flag) fprintf(fout, "%zu\n", count);
}
int main(int argc, char* argv[])
{
+ SET_BINARY_MODE(stdin);
+ SET_BINARY_MODE(stdout);
+
const char* pat;
pcre* re;
const char* pat_error;
int pat_error_offset;
FILE* fin;
- gzFile gzfin;
- invert_flag = 0;
- help_flag = 0;
- zout_flag = 0;
+ invert_flag = 0;
+ count_flag = 0;
+ id_flag = 0;
int opt;
int opt_idx;
static struct option long_options[] =
{
- {"help", no_argument, &help_flag, 1},
+ {"id", no_argument, &id_flag, 1},
{"invert-match", no_argument, &invert_flag, 1},
+ {"count", no_argument, &count_flag, 1},
+ {"help", no_argument, NULL, 'h'},
+ {"version", no_argument, NULL, 'V'},
{0, 0, 0, 0}
};
while (1) {
- opt = getopt_long(argc, argv, "hv", long_options, &opt_idx);
+ opt = getopt_long(argc, argv, "ivchV", long_options, &opt_idx);
- if( opt == -1 ) break;
+ if (opt == -1) break;
switch (opt) {
case 0:
if (long_options[opt_idx].flag != 0) break;
if (optarg) {
- /* TODO */
}
break;
- case 'h':
- help_flag = 1;
+ case 'i':
+ id_flag = 1;
break;
case 'v':
invert_flag = 1;
break;
+ case 'c':
+ count_flag = 1;
+ break;
+
+ case 'h':
+ print_help();
+ return 0;
+
+ case 'V':
+ print_version(stdout, prog_name);
+ return 0;
+
+ case '?':
+ return 1;
+
default:
abort();
}
}
- if (help_flag) {
- print_help();
- return 0;
- }
-
if (optind >= argc) {
fprintf(stderr, "A pattern must be specified.\n");
return 1;
if (optind >= argc || (argc - optind == 1 && strcmp(argv[optind],"-") == 0)) {
- gzfin = gzdopen( fileno(stdin), "rb" );
- if (gzfin == NULL) {
- fprintf(stderr, "Malformed file 'stdin'.\n");
- return 1;
- }
-
- fastq_grep(gzfin, stdout, re);
-
- gzclose(gzfin);
+ fastq_grep(stdin, stdout, re);
}
else {
for (; optind < argc; optind++) {
continue;
}
- gzfin = gzdopen(fileno(fin), "rb");
- if (gzfin == NULL) {
- fprintf(stderr, "Malformed file '%s'.\n", argv[optind]);
- continue;
- }
-
- fastq_grep(gzfin, stdout, re);
+ fastq_grep(fin, stdout, re);
- gzclose(gzfin);
+ fclose(fin);
}
}
projects / fastq-tools.git / blobdiff