Much simpler faster code for parsing fastq files.

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

diff --git a/src/fastq-kmers.c b/src/fastq-kmers.c
index 43fa3026b8bf02919cced34663fc14fc4999d066..92f4314141133c14bd0e4dc8c7a6507fb76daf53 100644 (file)
--- a/src/fastq-kmers.c
+++ b/src/fastq-kmers.c
@@ -1,22 +1,23 @@
 
-/* 
- * fastq-kmers: kmer frequences within fastq files
+/*
+ * This file is part of fastq-tools.
  *
- * Febuary 2011 / Daniel Jones <dcjones@cs.washington.edu> 
+ * Copyright (c) 2011 by Daniel C. Jones <dcjones@cs.washington.edu>
+ *
+ * fastq-kmers :
+ * Tabulate k-mer frequencies with FASTQ files.
  *
  */
 
-
-
+#include "common.h"
+#include "parse.h"
+#include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
 #include <stdint.h>
 #include <getopt.h>
 #include <zlib.h>
 
-#include "kseq.h"
-KSEQ_INIT(gzFile, gzread)
-
 
 #if defined(MSDOS) || defined(OS2) || defined(WIN32) || defined(__CYGWIN__)
 #  include <fcntl.h>
@@ -26,6 +27,7 @@ KSEQ_INIT(gzFile, gzread)
 #  define SET_BINARY_MODE(file)
 #endif
 
+static const char* prog_name = "fastq-kmers";
 
 void print_help()
 {
@@ -34,12 +36,12 @@ void print_help()
 "Print kmer counts for the given kmer size.\n"
 "Output is in two tab-seperated columns for kmer and frequency.\n\n"
 "Options:\n"
+"  -k NUM, --size=NUM      kmer size (default: 1)\n"
 "  -h, --help              print this message\n"
-"  -k, --size              kmer size (default: 1)\n"
+"  -V, --version           output version information and exit\n"
     );
 }
 
-static int help_flag;
 static int k;
 
 int packkmer( const char* s, uint32_t* kmer, int k )
@@ -105,15 +107,16 @@ void unpackkmer( uint32_t kmer, char* s, int k )
 }
 
 
-void count_fastq_kmers(gzFile* fin, uint32_t* cs)
+void count_fastq_kmers(FILE* fin, uint32_t* cs)
 {
-    kseq_t* seq = kseq_init(fin);
+    seq_t* seq = seq_create();
+    fastq_t* fqf = fastq_create(fin);
     int i;
     int n;
     uint32_t kmer;
 
-    while (kseq_read(seq) >= 0) {
-        n = (int)seq->seq.l - k + 1;
+    while (fastq_read(fqf, seq)) {
+        n = (int)seq->seq.n - k + 1;
         for (i = 0; i < n; i++) {
             if( packkmer(seq->seq.s + i, &kmer, k) ) {
                 cs[kmer]++;
@@ -121,7 +124,8 @@ void count_fastq_kmers(gzFile* fin, uint32_t* cs)
         }
     }
 
-    kseq_destroy(seq);
+    seq_free(seq);
+    fastq_free(fqf);
 }
 
 
@@ -147,26 +151,25 @@ int main(int argc, char* argv[])
     SET_BINARY_MODE(stdin);
     SET_BINARY_MODE(stdout);
 
-    help_flag = 0;
     k = 1;
 
     uint32_t n;   /* number of kmers: 4^k */
     uint32_t* cs; /* counts */
 
     FILE* fin;
-    gzFile gzfin;
 
     int opt;
     int opt_idx;
     static struct option long_options[] =
         { 
-          {"help", no_argument, &help_flag, 1},
-          {"size", no_argument, 0, 0},
+          {"size", no_argument,    0, 0},
+          {"help", no_argument,    0, 'h'},
+          {"version", no_argument, 0, 'V'},
           {0, 0, 0, 0}
         };
 
     while (1) {
-        opt = getopt_long(argc, argv, "hk:", long_options, &opt_idx);
+        opt = getopt_long(argc, argv, "k:hV", long_options, &opt_idx);
 
         if( opt == -1 ) break;
 
@@ -180,14 +183,18 @@ int main(int argc, char* argv[])
                 }
                 break;
 
-            case 'h':
-                help_flag = 1;
-                break;
-
             case 'k':
                 k = atoi(optarg);
                 break;
 
+            case 'h':
+                print_help();
+                return 0;
+
+            case 'V':
+                print_version(stdout, prog_name);
+                return 0;
+
             case '?':
                 return 1;
 
@@ -196,11 +203,6 @@ int main(int argc, char* argv[])
         }
     }
 
-    if (help_flag) {
-        print_help();
-        return 0;
-    }
-
     if (k < 1) {
         fprintf(stderr, "Kmer size must be at least 1.");
         return 1;
@@ -222,15 +224,7 @@ 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;
-        }
-
-        count_fastq_kmers(gzfin, cs);
-
-        gzclose(gzfin);
+        count_fastq_kmers(stdin, cs);
     }
     else {
         for (; optind < argc; optind++) {
@@ -240,15 +234,7 @@ int main(int argc, char* argv[])
                 continue;
             }
 
-            gzfin = gzdopen(fileno(fin), "rb");
-            if (gzfin == NULL) {
-                fprintf(stderr, "Malformed file '%s'.\n", argv[optind]);
-                continue;
-            }
-
-            count_fastq_kmers(gzfin, cs);
-
-            gzclose(gzfin);
+            count_fastq_kmers(fin, cs);
         }
     }