1 files changed, 92 insertions, 6 deletions

diff --git a/sys/libbio/io/fasta.c b/sys/libbio/io/fasta.c
index e26fd70..bb6bfc7 100644
--- a/sys/libbio/io/fasta.c
+++ b/sys/libbio/io/fasta.c
@@ -167,7 +167,7 @@ ERROR:
 
 static
 error
-readfasta(bio·FastaReader *rdr, bio·Seq *seq, byte tok)
+readfasta(bio·FastaReader *rdr, bio·Seq *seq, byte hdr, byte stop)
 {
     error  err;
     byte  *beg;
@@ -180,8 +180,8 @@ readfasta(bio·FastaReader *rdr, bio·Seq *seq, byte tok)
 
     // NOTE: Can this case happen?
     Assert(rdr->b != rdr->bend);
-    if (*rdr->b++ != tok) {
-        errorf("fasta format: expected '>', found '%c'", *rdr->b--);
+    if (*rdr->b++ != hdr) {
+        errorf("fasta/q format: expected '%c', found '%c'", hdr, *rdr->b--);
         return 1;
     }
 
@@ -214,7 +214,7 @@ SEQL:
             beg = rdr->b + 1;
         }
 
-        if (*rdr->b == tok || *rdr->b == '\0') {
+        if (*rdr->b == stop || *rdr->b == '\0') {
             goto SUCCESS;
         }
 
@@ -241,9 +241,9 @@ bio·readfasta(bio·FastaReader *rdr, bio·Seq *seq)
 {
     error err;
 
-    err = readfasta(rdr, seq, '>');
+    err = readfasta(rdr, seq, '>', '>');
     if (err && err != EOF) {
-        errorf("parse fail: could not read record");
+        errorf("parse fail: could not read sequence of record");
         return err;
     }
 
@@ -272,3 +272,89 @@ bio·closefasta(bio·FastaReader *rdr)
 
 // -----------------------------------------------------------------------
 // Fastq files
+
+struct bio·FastqReader {
+    struct bio·FastaReader;
+};
+
+bio·FastqReader*
+bio·openfastq(io·Reader file, void *f, mem·Allocator heap, void *h)
+{
+    return (bio·FastqReader*)bio·openfasta(file, f, heap, h);
+}
+
+error
+bio·closefastq(bio·FastqReader *rdr)
+{
+    return bio·closefasta((bio·FastaReader*)rdr);
+}
+
+error
+bio·readfastq(bio·FastqReader *rdr, bio·Seq *seq)
+{
+    int   n;
+    byte *beg;
+    error err;
+
+    err = readfasta((bio·FastaReader*)rdr, seq, '@', '+');
+    if (err) {
+        errorf("parse fail: could not read sequence of record");
+        return err;
+    }
+
+    seq->len = rdr->seq->it - (rdr->seq->b + rdr->seq->off);
+
+    if (*rdr->b++ != '+') {
+        errorf("format error: no '+' character seperator found");
+        return -1;
+    }
+
+EATLN:
+    while (rdr->b != rdr->bend) {
+        if (*rdr->b++ == '\n') {
+            n = 0;
+            goto QUAL;
+        }
+    }
+
+    if (err = fill((bio·FastaReader*)rdr), err) {
+        errorf("read: could not populate buffer");
+        return 1;
+    }
+    goto EATLN;
+
+QUAL:
+    beg = rdr->b;
+    while (rdr->b != rdr->bend) {
+        if (*rdr->b == '\n') {
+            push(&rdr->seq, rdr->b - beg, beg);
+            beg = rdr->b + 1;
+        }
+
+        if (n++ == seq->len || *rdr->b == '\0') {
+            err = *rdr->b == '\0' ? EOF : 0;
+            goto SUCCESS;
+        }
+
+        rdr->b++;
+    }
+
+    push(&rdr->seq, rdr->b - beg, beg);
+
+    if (err = fill((bio·FastaReader*)rdr), err) {
+        errorf("read: could not populate buffer");
+        return 1;
+    }
+    goto QUAL;
+
+
+SUCCESS:
+    push(&rdr->seq, rdr->b - beg, beg);
+    put(&rdr->seq,  '\0');
+
+    seq->name = rdr->seq->b;
+    seq->s    = rdr->seq->b + rdr->seq->off;
+    seq->q    = seq->s + seq->len + 1;
+
+    return err;
+}