X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_pileup.c;h=f68f400ab5cd086dc88a33e14899abe6cd4e53bb;hb=a7dfcd368cda94d5847d3f2b42d4288399d77835;hp=46906dba429032c49920f61a4149c29ec43eae11;hpb=7a36e408af60d3a48eb8e2fda313f4ee748ad0fc;p=samtools.git

diff --git a/bam_pileup.c b/bam_pileup.c
index 46906db..f68f400 100644
--- a/bam_pileup.c
+++ b/bam_pileup.c
@@ -1,6 +1,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <ctype.h>
+#include <assert.h>
 #include "sam.h"
 
 typedef struct __linkbuf_t {
@@ -61,7 +62,7 @@ static inline int resolve_cigar(bam_pileup1_t *p, uint32_t pos)
 	int ret = 1, is_restart = 1;
 
 	if (c->flag&BAM_FUNMAP) return 0; // unmapped read
-	assert(x <= pos);
+	assert(x <= pos); // otherwise a bug
 	p->qpos = -1; p->indel = 0; p->is_del = p->is_head = p->is_tail = 0;
 	for (k = 0; k < c->n_cigar; ++k) {
 		int op = bam1_cigar(b)[k] & BAM_CIGAR_MASK; // operation
@@ -77,6 +78,10 @@ static inline int resolve_cigar(bam_pileup1_t *p, uint32_t pos)
 						int op_next = cigar&BAM_CIGAR_MASK; // next CIGAR operation
 						if (op_next == BAM_CDEL) p->indel = -(int32_t)(cigar>>BAM_CIGAR_SHIFT); // del
 						else if (op_next == BAM_CINS) p->indel = cigar>>BAM_CIGAR_SHIFT; // ins
+						if (op_next == BAM_CDEL || op_next == BAM_CINS) {
+							if (k + 2 < c->n_cigar) op_next = bam1_cigar(b)[k+2]&BAM_CIGAR_MASK;
+							else p->is_tail = 1;
+						}
 						if (op_next == BAM_CSOFT_CLIP || op_next == BAM_CREF_SKIP || op_next == BAM_CHARD_CLIP)
 							p->is_tail = 1; // tail
 					} else p->is_tail = 1; // this is the last operation; set tail
@@ -97,7 +102,7 @@ static inline int resolve_cigar(bam_pileup1_t *p, uint32_t pos)
 			break;
 		}
 	}
-	assert(x > pos);
+	assert(x > pos); // otherwise a bug
 	return ret;
 }
 
@@ -165,9 +170,13 @@ int bam_plbuf_push(const bam1_t *b, bam_plbuf_t *buf)
 		if (b->core.flag & buf->flag_mask) return 0;
 		bam_copy1(&buf->tail->b, b);
 		buf->tail->beg = b->core.pos; buf->tail->end = bam_calend(&b->core, bam1_cigar(b));
-		if (!(b->core.tid >= buf->max_tid || (b->core.tid == buf->max_tid && buf->tail->beg >= buf->max_pos))) {
-			fprintf(stderr, "[bam_pileup_core] the input is not sorted. Abort!\n");
-			abort();
+		if (b->core.tid < buf->max_tid) {
+			fprintf(stderr, "[bam_pileup_core] the input is not sorted (chromosomes out of order)\n");
+			return -1;
+		}
+		if ((b->core.tid == buf->max_tid) && (buf->tail->beg < buf->max_pos)) {
+			fprintf(stderr, "[bam_pileup_core] the input is not sorted (reads out of order)\n");
+			return -1;
 		}
 		buf->max_tid = b->core.tid; buf->max_pos = buf->tail->beg;
 		if (buf->tail->end > buf->pos || buf->tail->b.core.tid > buf->tid) {
@@ -196,7 +205,12 @@ int bam_plbuf_push(const bam1_t *b, bam_plbuf_t *buf)
 			buf->func(buf->tid, buf->pos, n_pu, buf->pu, buf->func_data);
 		}
 		// update tid and pos
-		if (buf->head->next) assert(buf->tid <= buf->head->b.core.tid); // otherwise, not sorted
+		if (buf->head->next) {
+			if (buf->tid > buf->head->b.core.tid) {
+				fprintf(stderr, "[bam_plbuf_push] unsorted input. Pileup aborts.\n");
+				return 1;
+			}
+		}
 		if (buf->tid < buf->head->b.core.tid) { // come to a new reference sequence
 			buf->tid = buf->head->b.core.tid; buf->pos = buf->head->beg; // jump to the next reference
 		} else if (buf->pos < buf->head->beg) { // here: tid == head->b.core.tid
@@ -206,3 +220,19 @@ int bam_plbuf_push(const bam1_t *b, bam_plbuf_t *buf)
 	}
 	return 0;
 }
+
+int bam_pileup_file(bamFile fp, int mask, bam_pileup_f func, void *func_data)
+{
+	bam_plbuf_t *buf;
+	int ret;
+	bam1_t *b;
+	b = bam_init1();
+	buf = bam_plbuf_init(func, func_data);
+	bam_plbuf_set_mask(buf, mask);
+	while ((ret = bam_read1(fp, b)) >= 0)
+		bam_plbuf_push(b, buf);
+	bam_plbuf_push(0, buf);
+	bam_plbuf_destroy(buf);
+	bam_destroy1(b);
+	return 0;
+}