(limited) support for fasta

[fastq-tools.git] / src / fastq-grep.c

diff --git a/src/fastq-grep.c b/src/fastq-grep.c
index ca072ebc05fd234cab16e1190aa9a2ac1c9e7716..3eaffb79c53b89512a4ffe73d295981c7af7566f 100644 (file)
--- a/src/fastq-grep.c
+++ b/src/fastq-grep.c
@@ -1,22 +1,23 @@
 
-
-/* 
- * 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>
@@ -36,67 +37,65 @@ void print_help()
 "Options:\n"
 "  -h, --help              print this message\n"
 "  -v, --invert-match      select nonmatching entries\n"
+"  -c, --count             output only the number of matching sequences\n"
     );
 }
 
 static int invert_flag;
 static int help_flag;
-static int zout_flag;
+static int count_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;
 
-    kseq_t* seq;
-    seq = kseq_init(fin);
+    fastq_t* fqf = fastq_open(fin);
+    seq_t* seq = fastq_alloc_seq();
 
-    while (kseq_read(seq) >= 0) {
+    while (fastq_next(fqf, seq)) {
         rc = pcre_exec(re,          /* pattern */
                        NULL,        /* extre data */
                        seq->seq.s,  /* subject */
-                       seq->seq.l,  /* subject length */
+                       seq->seq.n,  /* subject length */
                        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 (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;
+    help_flag    = 0;
+    count_flag   = 0;
 
     int opt;
     int opt_idx;
@@ -106,11 +105,12 @@ int main(int argc, char* argv[])
         { 
           {"help", no_argument, &help_flag, 1},
           {"invert-match", no_argument, &invert_flag, 1},
+          {"count", no_argument, &count_flag, 1},
           {0, 0, 0, 0}
         };
 
     while (1) {
-        opt = getopt_long(argc, argv, "hv", long_options, &opt_idx);
+        opt = getopt_long(argc, argv, "hvc", long_options, &opt_idx);
 
         if( opt == -1 ) break;
 
@@ -118,7 +118,6 @@ int main(int argc, char* argv[])
             case 0:
                 if (long_options[opt_idx].flag != 0) break;
                 if (optarg) {
-                    /* TODO */
                 }
                 break;
 
@@ -130,6 +129,13 @@ int main(int argc, char* argv[])
                 invert_flag = 1;
                 break;
 
+            case 'c':
+                count_flag = 1;
+                break;
+
+            case '?':
+                return 1;
+
             default:
                 abort();
         }
@@ -157,15 +163,7 @@ int main(int argc, char* argv[])
 
 
     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++) {
@@ -175,15 +173,9 @@ int main(int argc, char* argv[])
                 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);
         }
     }