X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=src%2Ffastq-grep.c;h=4d1603f7ced45f3d34e28d00e60204689031d4a8;hb=d6b81119e55cf77c1e4c17b9e9026abc01e22b96;hp=d402cdeaa4e0423cf10e1e5410b8aed8a5080109;hpb=e25e049490086a631d12bc56785f1a1845877541;p=fastq-tools.git

diff --git a/src/fastq-grep.c b/src/fastq-grep.c
index d402cde..4d1603f 100644
--- a/src/fastq-grep.c
+++ b/src/fastq-grep.c
@@ -38,7 +38,9 @@ void print_help()
 "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"
+"  -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"
@@ -47,36 +49,41 @@ void print_help()
 
 static int invert_flag;
 static int count_flag;
+static int id_flag;
 
 
 
-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            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);
 }
@@ -98,14 +105,18 @@ int main(int argc, char* argv[])
 
     invert_flag  = 0;
     count_flag   = 0;
+    id_flag      = 0;
 
     int opt;
     int opt_idx;
 
+    FILE* mismatch_file = NULL;
 
     static struct option long_options[] =
         { 
+          {"id",           no_argument, &id_flag,     1},
           {"invert-match", no_argument, &invert_flag, 1},
+          {"mismatches",   required_argument, NULL, 'm'},
           {"count",        no_argument, &count_flag,  1},
           {"help",         no_argument, NULL, 'h'},
           {"version",      no_argument, NULL, 'V'},
@@ -113,9 +124,9 @@ int main(int argc, char* argv[])
         };
 
     while (1) {
-        opt = getopt_long(argc, argv, "vchV", 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:
@@ -124,10 +135,22 @@ int main(int argc, char* argv[])
                 }
                 break;
 
+            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;
@@ -165,7 +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++) {
@@ -175,13 +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;
 }