Much simpler faster code for parsing fastq files.

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

diff --git a/src/fastq-grep.c b/src/fastq-grep.c
index 67c91901ead7d4a8320605c06ef4d72013966186..4d1603f7ced45f3d34e28d00e60204689031d4a8 100644 (file)
--- a/src/fastq-grep.c
+++ b/src/fastq-grep.c
@@ -28,62 +28,62 @@
 #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"
+"  -m, --mismatches=FILE   output mismatching entries to the given file\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 count_flag;
+static int id_flag;
 
 
 
-void print_fastq_entry(FILE* fout, seq_t* seq)
-{
-    fprintf(fout, "@%s\n%s\n+%s\n%s\n",
-                  seq->id1.s,
-                  seq->seq.s,
-                  seq->id2.s,
-                  seq->qual.s );
-}
-
-
-void fastq_grep(FILE* fin, FILE* fout, pcre* re)
+void fastq_grep(FILE* fin, FILE* fout, FILE* mismatch_file, pcre* re)
 {
     int rc;
     int ovector[3];
     size_t count = 0;
 
-    fastq_t* fqf = fastq_open(fin);
-    seq_t* seq = fastq_alloc_seq();
+    fastq_t* fqf = fastq_create(fin);
+    seq_t* seq = seq_create();
+
+    while (fastq_read(fqf, seq)) {
 
-    while (fastq_next(fqf, seq)) {
         rc = pcre_exec(re,          /* pattern */
-                       NULL,        /* extre data */
-                       seq->seq.s,  /* subject */
-                       seq->seq.n,  /* 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) {
+        if ((invert_flag && rc == PCRE_ERROR_NOMATCH) || (!invert_flag && rc >= 0)) {
             if (count_flag) count++;
-            else            print_fastq_entry(fout, seq);
+            else            fastq_print(fout, seq);
+        }
+        else if (mismatch_file) {
+            fastq_print(mismatch_file, seq);
         }
     }
 
-    fastq_free_seq(seq);
-    fastq_close(fqf);
+    seq_free(seq);
+    fastq_free(fqf);
 
     if (count_flag) fprintf(fout, "%zu\n", count);
 }
@@ -104,25 +104,29 @@ int main(int argc, char* argv[])
 
 
     invert_flag  = 0;
-    help_flag    = 0;
     count_flag   = 0;
+    id_flag      = 0;
 
     int opt;
     int opt_idx;
 
+    FILE* mismatch_file = NULL;
 
     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},
+          {"mismatches",   required_argument, NULL, 'm'},
+          {"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, "hvc", long_options, &opt_idx);
+        opt = getopt_long(argc, argv, "ivmchV", long_options, &opt_idx);
 
-        if( opt == -1 ) break;
+        if (opt == -1) break;
 
         switch (opt) {
             case 0:
@@ -131,18 +135,34 @@ int main(int argc, char* argv[])
                 }
                 break;
 
-            case 'h':
-                help_flag = 1;
+            case 'i':
+                id_flag = 1;
                 break;
 
             case 'v':
                 invert_flag = 1;
                 break;
 
+            case 'm':
+                mismatch_file = fopen(optarg, "w");
+                if (mismatch_file == NULL) {
+                    fprintf(stderr, "No such file '%s'.\n", optarg);
+                    return 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;
 
@@ -151,11 +171,6 @@ int main(int argc, char* argv[])
         }
     }
 
-    if (help_flag) {
-        print_help();
-        return 0;
-    }
-
     if (optind >= argc) {
         fprintf(stderr, "A pattern must be specified.\n");
         return 1;
@@ -173,7 +188,7 @@ int main(int argc, char* argv[])
 
 
     if (optind >= argc || (argc - optind == 1 && strcmp(argv[optind],"-") == 0)) {
-        fastq_grep(stdin, stdout, re);
+        fastq_grep(stdin, stdout, mismatch_file, re);
     }
     else {
         for (; optind < argc; optind++) {
@@ -183,13 +198,14 @@ int main(int argc, char* argv[])
                 continue;
             }
 
-            fastq_grep(fin, stdout, re);
+            fastq_grep(fin, stdout, mismatch_file, re);
 
             fclose(fin);
         }
     }
 
     pcre_free(re);
+    if (mismatch_file) fclose(mismatch_file);
 
     return 0;
 }