#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;
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);
}
{
bam_header_t *header;
char buf[4];
+ int magic_len;
int32_t i = 1, name_len;
// check EOF
i = bgzf_check_EOF(fp);
- if (i < 0) fprintf(stderr, "[bam_header_read] read from pipe; skip EOF checking.\n");
- else if (i == 0) fprintf(stderr, "[bam_header_read] EOF marker is absent.\n");
+ 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. The input is probably truncated.\n");
// read "BAM1"
- if (bam_read(fp, buf, 4) != 4) return 0;
- if (strncmp(buf, "BAM\001", 4)) {
- fprintf(stderr, "[bam_header_read] wrong header\n");
+ magic_len = bam_read(fp, buf, 4);
+ if (magic_len != 4 || strncmp(buf, "BAM\001", 4) != 0) {
+ fprintf(stderr, "[bam_header_read] invalid BAM binary header (this is not a BAM file).\n");
return 0;
}
header = bam_header_init();
bam_write(fp, &x, 4);
} else bam_write(fp, &header->target_len[i], 4);
}
+ bgzf_flush(fp);
return 0;
}
x[5] = c->mtid;
x[6] = c->mpos;
x[7] = c->isize;
+ bgzf_flush_try(fp, 4 + block_len);
if (bam_is_be) {
for (i = 0; i < 8; ++i) bam_swap_endian_4p(x + i);
y = block_len;
kstring_t str;
str.l = str.m = 0; str.s = 0;
- ksprintf(&str, "%s\t", bam1_qname(b));
- if (of == BAM_OFDEC) ksprintf(&str, "%d\t", c->flag);
+ kputsn(bam1_qname(b), c->l_qname-1, &str); kputc('\t', &str);
+ if (of == BAM_OFDEC) { kputw(c->flag, &str); kputc('\t', &str); }
else if (of == BAM_OFHEX) ksprintf(&str, "0x%x\t", c->flag);
else { // BAM_OFSTR
for (i = 0; i < 16; ++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);
+ if (c->tid < 0) kputsn("*\t", 2, &str);
+ else {
+ if (header) kputs(header->target_name[c->tid] , &str);
+ else kputw(c->tid, &str);
+ kputc('\t', &str);
+ }
+ kputw(c->pos + 1, &str); kputc('\t', &str); kputw(c->qual, &str); kputc('\t', &str);
if (c->n_cigar == 0) kputc('*', &str);
else {
- for (i = 0; i < c->n_cigar; ++i)
- ksprintf(&str, "%d%c", bam1_cigar(b)[i]>>BAM_CIGAR_SHIFT, "MIDNSHP"[bam1_cigar(b)[i]&BAM_CIGAR_MASK]);
+ for (i = 0; i < c->n_cigar; ++i) {
+ kputw(bam1_cigar(b)[i]>>BAM_CIGAR_SHIFT, &str);
+ kputc("MIDNSHP"[bam1_cigar(b)[i]&BAM_CIGAR_MASK], &str);
+ }
}
kputc('\t', &str);
- if (c->mtid < 0) kputs("*\t", &str);
- 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);
+ if (c->mtid < 0) kputsn("*\t", 2, &str);
+ else if (c->mtid == c->tid) kputsn("=\t", 2, &str);
+ else {
+ if (header) kputs(header->target_name[c->mtid], &str);
+ else kputw(c->mtid, &str);
+ kputc('\t', &str);
+ }
+ kputw(c->mpos + 1, &str); kputc('\t', &str); kputw(c->isize, &str); kputc('\t', &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*");
+ } else kputsn("*\t*", 3, &str);
s = bam1_aux(b);
while (s < b->data + b->data_len) {
uint8_t type, key[2];
key[0] = s[0]; key[1] = s[1];
s += 2; type = *s; ++s;
- ksprintf(&str, "\t%c%c:", key[0], key[1]);
- if (type == 'A') { ksprintf(&str, "A:%c", *s); ++s; }
- else if (type == 'C') { ksprintf(&str, "i:%u", *s); ++s; }
- else if (type == 'c') { ksprintf(&str, "i:%d", *s); ++s; }
- else if (type == 'S') { ksprintf(&str, "i:%u", *(uint16_t*)s); s += 2; }
- else if (type == 's') { ksprintf(&str, "i:%d", *(int16_t*)s); s += 2; }
- else if (type == 'I') { ksprintf(&str, "i:%u", *(uint32_t*)s); s += 4; }
- else if (type == 'i') { ksprintf(&str, "i:%d", *(int32_t*)s); s += 4; }
+ kputc('\t', &str); kputsn((char*)key, 2, &str); kputc(':', &str);
+ if (type == 'A') { kputsn("A:", 2, &str); kputc(*s, &str); ++s; }
+ else if (type == 'C') { kputsn("i:", 2, &str); kputw(*s, &str); ++s; }
+ else if (type == 'c') { kputsn("i:", 2, &str); kputw(*(int8_t*)s, &str); ++s; }
+ else if (type == 'S') { kputsn("i:", 2, &str); kputw(*(uint16_t*)s, &str); s += 2; }
+ else if (type == 's') { kputsn("i:", 2, &str); kputw(*(int16_t*)s, &str); s += 2; }
+ else if (type == 'I') { kputsn("i:", 2, &str); kputuw(*(uint32_t*)s, &str); s += 4; }
+ else if (type == 'i') { kputsn("i:", 2, &str); kputw(*(int32_t*)s, &str); s += 4; }
else if (type == 'f') { ksprintf(&str, "f:%g", *(float*)s); s += 4; }
else if (type == 'd') { ksprintf(&str, "d:%lg", *(double*)s); s += 8; }
- else if (type == 'Z' || type == 'H') { ksprintf(&str, "%c:", type); while (*s) kputc(*s++, &str); ++s; }
+ else if (type == 'Z' || type == 'H') { kputc(type, &str); kputc(':', &str); while (*s) kputc(*s++, &str); ++s; }
}
return str.s;
}
void bam_view1(const bam_header_t *header, const bam1_t *b)
{
char *s = bam_format1(header, b);
- printf("%s\n", s);
+ puts(s);
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)
+{
+ 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));
+}
projects / samtools.git / blobdiff