Imported Upstream version 0.1.7~dfsg

[samtools.git] / bam.c

diff --git a/bam.c b/bam.c
index 1ff4a5aba25028f4fccdd0735a9dcae594b6e9a8..ee7642b3434248c9b44ef918435f150f101cae00 100644 (file)
--- a/bam.c
+++ b/bam.c
@@ -1,38 +1,19 @@
 #include <stdio.h>
 #include <ctype.h>
+#include <errno.h>
 #include <assert.h>
 #include "bam.h"
 #include "bam_endian.h"
 #include "kstring.h"
+#include "sam_header.h"
 
 int bam_is_be = 0;
+char *bam_flag2char_table = "pPuUrR12sfd\0\0\0\0\0";
 
 /**************************
  * CIGAR related routines *
  **************************/
 
-int bam_segreg(int32_t pos, const bam1_core_t *c, const uint32_t *cigar, bam_segreg_t *reg)
-{
-       unsigned k;
-       int32_t x = c->pos, y = 0;
-       int state = 0;
-       for (k = 0; k < c->n_cigar; ++k) {
-               int op = cigar[k] & BAM_CIGAR_MASK; // operation
-               int l = cigar[k] >> BAM_CIGAR_SHIFT; // length
-               if (state == 0 && (op == BAM_CMATCH || op == BAM_CDEL || op == BAM_CINS) && x + l > pos) {
-                       reg->tbeg = x; reg->qbeg = y; reg->cbeg = k;
-                       state = 1;
-               }
-               if (op == BAM_CMATCH) { x += l; y += l; }
-               else if (op == BAM_CDEL || op == BAM_CREF_SKIP) x += l;
-               else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) y += l;
-               if (state == 1 && (op == BAM_CSOFT_CLIP || op == BAM_CHARD_CLIP || op == BAM_CREF_SKIP || k == c->n_cigar - 1)) {
-                       reg->tend = x; reg->qend = y; reg->cend = k;
-               }
-       }
-       return state? 0 : -1;
-}
-
 uint32_t bam_calend(const bam1_core_t *c, const uint32_t *cigar)
 {
        uint32_t k, end;
@@ -79,10 +60,9 @@ void bam_header_destroy(bam_header_t *header)
                free(header->target_len);
        }
        free(header->text);
-#ifndef BAM_NO_HASH
-       if (header->rg2lib) bam_strmap_destroy(header->rg2lib);
+       if (header->dict) sam_header_free(header->dict);
+       if (header->rg2lib) sam_tbl_destroy(header->rg2lib);
        bam_destroy_header_hash(header);
-#endif
        free(header);
 }
 
@@ -90,7 +70,15 @@ bam_header_t *bam_header_read(bamFile fp)
 {
        bam_header_t *header;
        char buf[4];
-       int32_t i, name_len;
+       int32_t i = 1, name_len;
+       // check EOF
+       i = bgzf_check_EOF(fp);
+       if (i < 0) {
+               // If the file is a pipe, checking the EOF marker will *always* fail
+               // 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");
        // read "BAM1"
        if (bam_read(fp, buf, 4) != 4) return 0;
        if (strncmp(buf, "BAM\001", 4)) {
@@ -236,7 +224,7 @@ int bam_write1(bamFile fp, const bam1_t *b)
        return bam_write1_core(fp, &b->core, b->data_len, b->data);
 }
 
-char *bam_format1(const bam_header_t *header, const bam1_t *b)
+char *bam_format1_core(const bam_header_t *header, const bam1_t *b, int of)
 {
        uint8_t *s = bam1_seq(b), *t = bam1_qual(b);
        int i;
@@ -244,7 +232,15 @@ char *bam_format1(const bam_header_t *header, const bam1_t *b)
        kstring_t str;
        str.l = str.m = 0; str.s = 0;
 
-       ksprintf(&str, "%s\t%d\t", bam1_qname(b), c->flag);
+       ksprintf(&str, "%s\t", bam1_qname(b));
+       if (of == BAM_OFDEC) ksprintf(&str, "%d\t", c->flag);
+       else if (of == BAM_OFHEX) ksprintf(&str, "0x%x\t", c->flag);
+       else { // BAM_OFSTR
+               for (i = 0; i < 16; ++i)
+                       if ((c->flag & 1<<i) && bam_flag2char_table[i])
+                               kputc(bam_flag2char_table[i], &str);
+               kputc('\t', &str);
+       }
        if (c->tid < 0) kputs("*\t", &str);
        else ksprintf(&str, "%s\t", header->target_name[c->tid]);
        ksprintf(&str, "%d\t%d\t", c->pos + 1, c->qual);
@@ -258,10 +254,12 @@ char *bam_format1(const bam_header_t *header, const bam1_t *b)
        else if (c->mtid == c->tid) kputs("=\t", &str);
        else ksprintf(&str, "%s\t", header->target_name[c->mtid]);
        ksprintf(&str, "%d\t%d\t", c->mpos + 1, c->isize);
-       for (i = 0; i < c->l_qseq; ++i) kputc(bam_nt16_rev_table[bam1_seqi(s, i)], &str);
-       kputc('\t', &str);
-       if (t[0] == 0xff) kputc('*', &str);
-       else for (i = 0; i < c->l_qseq; ++i) kputc(t[i] + 33, &str);
+       if (c->l_qseq) {
+               for (i = 0; i < c->l_qseq; ++i) kputc(bam_nt16_rev_table[bam1_seqi(s, i)], &str);
+               kputc('\t', &str);
+               if (t[0] == 0xff) kputc('*', &str);
+               else for (i = 0; i < c->l_qseq; ++i) kputc(t[i] + 33, &str);
+       } else ksprintf(&str, "*\t*");
        s = bam1_aux(b);
        while (s < b->data + b->data_len) {
                uint8_t type, key[2];
@@ -282,9 +280,24 @@ char *bam_format1(const bam_header_t *header, const bam1_t *b)
        return str.s;
 }
 
+char *bam_format1(const bam_header_t *header, const bam1_t *b)
+{
+       return bam_format1_core(header, b, BAM_OFDEC);
+}
+
 void bam_view1(const bam_header_t *header, const bam1_t *b)
 {
        char *s = bam_format1(header, b);
        printf("%s\n", s);
        free(s);
 }
+
+// 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)
+{
+       const uint8_t *rg;
+       if (h->dict == 0) h->dict = sam_header_parse2(h->text);
+       if (h->rg2lib == 0) h->rg2lib = sam_header2tbl(h->dict, "RG", "ID", "LB");
+       rg = bam_aux_get(b, "RG");
+       return (rg == 0)? 0 : sam_tbl_get(h->rg2lib, (const char*)(rg + 1));
+}