chore: pulled out sequence header function to be used by fastq

2 files changed, 22 insertions, 9 deletions

diff --git a/sys/libbio/io/fasta.c b/sys/libbio/io/fasta.c
index 5e659ab..e26fd70 100644
--- a/sys/libbio/io/fasta.c
+++ b/sys/libbio/io/fasta.c
@@ -96,6 +96,9 @@ push(struct Seqbuf **sb, int n, void *buf)
     return 0;
 }
 
+// -----------------------------------------------------------------------
+// Fasta files
+
 struct bio·FastaReader {
     byte      eof;
     io·Reader file;
@@ -132,9 +135,6 @@ fill(bio·FastaReader *rdr)
     return 0;
 }
 
-// -----------------------------------------------------------------------
-// Fasta files
-
 bio·FastaReader*
 bio·openfasta(io·Reader file, void *f, mem·Allocator heap, void *h)
 {
@@ -165,8 +165,9 @@ ERROR:
     return nil;
 }
 
+static
 error
-bio·readfasta(bio·FastaReader *rdr, bio·Seq *seq)
+readfasta(bio·FastaReader *rdr, bio·Seq *seq, byte tok)
 {
     error  err;
     byte  *beg;
@@ -179,7 +180,7 @@ bio·readfasta(bio·FastaReader *rdr, bio·Seq *seq)
 
     // NOTE: Can this case happen?
     Assert(rdr->b != rdr->bend);
-    if (*rdr->b++ != '>') {
+    if (*rdr->b++ != tok) {
         errorf("fasta format: expected '>', found '%c'", *rdr->b--);
         return 1;
     }
@@ -213,7 +214,7 @@ SEQL:
             beg = rdr->b + 1;
         }
 
-        if (*rdr->b == '>' || *rdr->b == '\0') {
+        if (*rdr->b == tok || *rdr->b == '\0') {
             goto SUCCESS;
         }
 
@@ -232,12 +233,26 @@ SUCCESS:
     push(&rdr->seq, rdr->b - beg, beg);
     put(&rdr->seq, '\0');
 
+    return 0;
+}
+
+error
+bio·readfasta(bio·FastaReader *rdr, bio·Seq *seq)
+{
+    error err;
+
+    err = readfasta(rdr, seq, '>');
+    if (err && err != EOF) {
+        errorf("parse fail: could not read record");
+        return err;
+    }
+
     seq->name = rdr->seq->b;
     seq->s    = rdr->seq->b + rdr->seq->off;
     seq->len  = rdr->seq->it - seq->s;
     seq->q    = nil;
 
-    return 0;
+    return err;
 }
 
 error
@@ -246,7 +261,6 @@ bio·closefasta(bio·FastaReader *rdr)
     mem·Allocator heap;
     void         *h;
 
-
     heap = rdr->seq->heap;
     h    = rdr->seq->h; 
     
diff --git a/sys/libbio/test.c b/sys/libbio/test.c
index d38bbc1..f301ead 100644
--- a/sys/libbio/test.c
+++ b/sys/libbio/test.c
@@ -3,7 +3,6 @@
 #include <libbio.h>
 
 #include <time.h>
-#include "kseq.h"
 
 // -----------------------------------------------------------------------
 // Point of entry for testing