* synchronize bgzf in tabix and in samtools

[samtools.git] / bam.c

diff --git a/bam.c b/bam.c
index 41850bab46e190f33c0b279ec20e4bec663bb86f..96aace21c7ab7c261380ee2136e283022e9c834d 100644 (file)
--- a/bam.c
+++ b/bam.c
@@ -79,7 +79,7 @@ bam_header_t *bam_header_read(bamFile fp)
                // with ESPIPE.  Suppress the error message in this case.
                if (errno != ESPIPE) perror("[bam_header_read] bgzf_check_EOF");
        }
-       else if (i == 0) fprintf(stderr, "[bam_header_read] EOF marker is absent.\n");
+       else if (i == 0) fprintf(stderr, "[bam_header_read] EOF marker is absent. The input is probably truncated.\n");
        // read "BAM1"
        magic_len = bam_read(fp, buf, 4);
        if (magic_len != 4 || strncmp(buf, "BAM\001", 4) != 0) {
@@ -305,6 +305,22 @@ void bam_view1(const bam_header_t *header, const bam1_t *b)
        free(s);
 }
 
+int bam_validate1(const bam_header_t *header, const bam1_t *b)
+{
+       char *s;
+
+       if (b->core.tid < -1 || b->core.mtid < -1) return 0;
+       if (header && (b->core.tid >= header->n_targets || b->core.mtid >= header->n_targets)) return 0;
+
+       if (b->data_len < b->core.l_qname) return 0;
+       s = memchr(bam1_qname(b), '\0', b->core.l_qname);
+       if (s != &bam1_qname(b)[b->core.l_qname-1]) return 0;
+
+       // FIXME: Other fields could also be checked, especially the auxiliary data
+
+       return 1;
+}
+
 // FIXME: we should also check the LB tag associated with each alignment
 const char *bam_get_library(bam_header_t *h, const bam1_t *b)
 {