X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=bam_index.c;h=f250907039156c5559d92bfff9254ef0d8c401c8;hb=be1abba58b415b449a7a9230f2221fd3960781db;hp=3fa950d5235297c36c598258ce20b80183b3f04b;hpb=adefe520aeefbaf64ffdc947bbe35db4bfe9d811;p=samtools.git

diff --git a/bam_index.c b/bam_index.c
index 3fa950d..f250907 100644
--- a/bam_index.c
+++ b/bam_index.c
@@ -120,7 +120,7 @@ static void merge_chunks(bam_index_t *idx)
 		index = idx->index[i];
 		for (k = kh_begin(index); k != kh_end(index); ++k) {
 			bam_binlist_t *p;
-			if (!kh_exist(index, k)) continue;
+			if (!kh_exist(index, k) || kh_key(index, k) == BAM_MAX_BIN) continue;
 			p = &kh_value(index, k);
 			m = 0;
 			for (l = 1; l < p->n; ++l) {
@@ -154,37 +154,54 @@ bam_index_t *bam_index_core(bamFile fp)
 	bam_header_t *h;
 	int i, ret;
 	bam_index_t *idx;
-	uint32_t last_bin, save_bin;
+	uint32_t last_bin, save_bin, recalculated_bin;
 	int32_t last_coor, last_tid, save_tid;
 	bam1_core_t *c;
 	uint64_t save_off, last_off, n_mapped, n_unmapped, off_beg, off_end, n_no_coor;
 
+	h = bam_header_read(fp);
+	if(h == 0) {
+	    fprintf(stderr, "[bam_index_core] Invalid BAM header.");
+	    return NULL;
+	}
+
 	idx = (bam_index_t*)calloc(1, sizeof(bam_index_t));
 	b = (bam1_t*)calloc(1, sizeof(bam1_t));
-	h = bam_header_read(fp);
 	c = &b->core;
 
 	idx->n = h->n_targets;
-	bam_header_destroy(h);
 	idx->index = (khash_t(i)**)calloc(idx->n, sizeof(void*));
 	for (i = 0; i < idx->n; ++i) idx->index[i] = kh_init(i);
 	idx->index2 = (bam_lidx_t*)calloc(idx->n, sizeof(bam_lidx_t));
 
 	save_bin = save_tid = last_tid = last_bin = 0xffffffffu;
 	save_off = last_off = bam_tell(fp); last_coor = 0xffffffffu;
-    n_mapped = n_unmapped = n_no_coor = off_end = 0;
+	n_mapped = n_unmapped = n_no_coor = off_end = 0;
 	off_beg = off_end = bam_tell(fp);
 	while ((ret = bam_read1(fp, b)) >= 0) {
 		if (c->tid < 0) ++n_no_coor;
-		if (last_tid != c->tid) { // change of chromosomes
+		if (last_tid < c->tid || (last_tid >= 0 && c->tid < 0)) { // change of chromosomes
 			last_tid = c->tid;
 			last_bin = 0xffffffffu;
-		} else if (last_coor > c->pos) {
+		} else if ((uint32_t)last_tid > (uint32_t)c->tid) {
+			fprintf(stderr, "[bam_index_core] the alignment is not sorted (%s): %d-th chr > %d-th chr\n",
+					bam1_qname(b), last_tid+1, c->tid+1);
+			return NULL;
+		} else if ((int32_t)c->tid >= 0 && last_coor > c->pos) {
 			fprintf(stderr, "[bam_index_core] the alignment is not sorted (%s): %u > %u in %d-th chr\n",
 					bam1_qname(b), last_coor, c->pos, c->tid+1);
-			exit(1);
+			return NULL;
 		}
-		if (c->tid >= 0) insert_offset2(&idx->index2[b->core.tid], b, last_off);
+		if (c->tid >= 0) {
+			recalculated_bin = bam_reg2bin(c->pos, bam_calend(c, bam1_cigar(b)));
+			if (c->bin != recalculated_bin) {
+				fprintf(stderr, "[bam_index_core] read '%s' mapped to '%s' at POS %d to %d has BIN %d but should be %d\n",
+					bam1_qname(b), h->target_name[c->tid], c->pos + 1, bam_calend(c, bam1_cigar(b)), c->bin, recalculated_bin);
+				fprintf(stderr, "[bam_index_core] Fix it by using BAM->SAM->BAM to force a recalculation of the BIN field\n");
+				return NULL;
+			}
+		}
+		if (c->tid >= 0 && !(c->flag & BAM_FUNMAP)) insert_offset2(&idx->index2[b->core.tid], b, last_off);
 		if (c->bin != last_bin) { // then possibly write the binning index
 			if (save_bin != 0xffffffffu) // save_bin==0xffffffffu only happens to the first record
 				insert_offset(idx->index[save_tid], save_bin, save_off, last_off);
@@ -203,22 +220,30 @@ bam_index_t *bam_index_core(bamFile fp)
 		if (bam_tell(fp) <= last_off) {
 			fprintf(stderr, "[bam_index_core] bug in BGZF/RAZF: %llx < %llx\n",
 					(unsigned long long)bam_tell(fp), (unsigned long long)last_off);
-			exit(1);
+			return NULL;
 		}
 		if (c->flag & BAM_FUNMAP) ++n_unmapped;
 		else ++n_mapped;
 		last_off = bam_tell(fp);
 		last_coor = b->core.pos;
 	}
+        bam_header_destroy(h);
 	if (save_tid >= 0) {
 		insert_offset(idx->index[save_tid], save_bin, save_off, bam_tell(fp));
-		insert_offset(idx->index[save_tid], BAM_MAX_BIN, off_beg, off_end);
+		insert_offset(idx->index[save_tid], BAM_MAX_BIN, off_beg, bam_tell(fp));
 		insert_offset(idx->index[save_tid], BAM_MAX_BIN, n_mapped, n_unmapped);
 	}
 	merge_chunks(idx);
 	fill_missing(idx);
-	if (ret >= 0)
-		while ((ret = bam_read1(fp, b)) >= 0) ++n_no_coor;
+	if (ret >= 0) {
+		while ((ret = bam_read1(fp, b)) >= 0) {
+			++n_no_coor;
+			if (c->tid >= 0 && n_no_coor) {
+				fprintf(stderr, "[bam_index_core] the alignment is not sorted: reads without coordinates prior to reads with coordinates.\n");
+				return NULL;
+			}
+		}
+	}
 	if (ret < -1) fprintf(stderr, "[bam_index_core] truncated file? Continue anyway. (%d)\n", ret);
 	free(b->data); free(b);
 	idx->n_no_coor = n_no_coor;
@@ -448,6 +473,7 @@ bam_index_t *bam_index_load(const char *fn)
 		strcat(strcpy(fnidx, fn), ".bai");
 		fprintf(stderr, "[bam_index_load] attempting to download the remote index file.\n");
 		download_from_remote(fnidx);
+        free(fnidx);
 		idx = bam_index_load_local(fn);
 	}
 	if (idx == 0) fprintf(stderr, "[bam_index_load] fail to load BAM index.\n");
@@ -466,6 +492,10 @@ int bam_index_build2(const char *fn, const char *_fnidx)
 	}
 	idx = bam_index_core(fp);
 	bam_close(fp);
+	if(idx == 0) {
+		fprintf(stderr, "[bam_index_build2] fail to index the BAM file.\n");
+		return -1;
+	}
 	if (_fnidx == 0) {
 		fnidx = (char*)calloc(strlen(fn) + 5, 1);
 		strcpy(fnidx, fn); strcat(fnidx, ".bai");
@@ -474,6 +504,7 @@ int bam_index_build2(const char *fn, const char *_fnidx)
 	if (fpidx == 0) {
 		fprintf(stderr, "[bam_index_build2] fail to create the index file.\n");
 		free(fnidx);
+        bam_index_destroy(idx);
 		return -1;
 	}
 	bam_index_save(idx, fpidx);
@@ -504,7 +535,7 @@ int bam_idxstats(int argc, char *argv[])
 	bam_index_t *idx;
 	bam_header_t *header;
 	bamFile fp;
-	int i, no_stats = 0;
+	int i;
 	if (argc < 2) {
 		fprintf(stderr, "Usage: samtools idxstats <in.bam>\n");
 		return 1;
@@ -522,12 +553,10 @@ int bam_idxstats(int argc, char *argv[])
 		k = kh_get(i, h, BAM_MAX_BIN);
 		if (k != kh_end(h))
 			printf("\t%llu\t%llu", (long long)kh_val(h, k).list[1].u, (long long)kh_val(h, k).list[1].v);
-		else no_stats = 1;
+		else printf("\t0\t0");
 		putchar('\n');
 	}
-	printf("*\t0");
-	if (!no_stats) printf("\t0\t%llu", (long long)idx->n_no_coor);
-	putchar('\n');
+	printf("*\t0\t0\t%llu\n", (long long)idx->n_no_coor);
 	bam_header_destroy(header);
 	bam_index_destroy(idx);
 	return 0;
@@ -610,6 +639,9 @@ bam_iter_t bam_iter_query(const bam_index_t *idx, int tid, int beg, int end)
 		}
 	}
 	free(bins);
+	if (n_off == 0) {
+		free(off); return iter;
+	}
 	{
 		bam1_t *b = (bam1_t*)calloc(1, sizeof(bam1_t));
 		int l;
@@ -658,17 +690,17 @@ void bam_iter_destroy(bam_iter_t iter)
 
 int bam_iter_read(bamFile fp, bam_iter_t iter, bam1_t *b)
 {
-	if (iter->finished) return -1;
-	if (iter->from_first) {
-		int ret = bam_read1(fp, b);
-		if (ret < 0) iter->finished = 1;
+	int ret;
+	if (iter && iter->finished) return -1;
+	if (iter == 0 || iter->from_first) {
+		ret = bam_read1(fp, b);
+		if (ret < 0 && iter) iter->finished = 1;
 		return ret;
 	}
 	if (iter->off == 0) return -1;
 	for (;;) {
-		int ret;
 		if (iter->curr_off == 0 || iter->curr_off >= iter->off[iter->i].v) { // then jump to the next chunk
-			if (iter->i == iter->n_off - 1) break; // no more chunks
+			if (iter->i == iter->n_off - 1) { ret = -1; break; } // no more chunks
 			if (iter->i >= 0) assert(iter->curr_off == iter->off[iter->i].v); // otherwise bug
 			if (iter->i < 0 || iter->off[iter->i].v != iter->off[iter->i+1].u) { // not adjacent chunks; then seek
 				bam_seek(fp, iter->off[iter->i+1].u, SEEK_SET);
@@ -676,23 +708,28 @@ int bam_iter_read(bamFile fp, bam_iter_t iter, bam1_t *b)
 			}
 			++iter->i;
 		}
-		if ((ret = bam_read1(fp, b)) > 0) {
+		if ((ret = bam_read1(fp, b)) >= 0) {
 			iter->curr_off = bam_tell(fp);
-			if (b->core.tid != iter->tid || b->core.pos >= iter->end) break; // no need to proceed
+			if (b->core.tid != iter->tid || b->core.pos >= iter->end) { // no need to proceed
+				ret = bam_validate1(NULL, b)? -1 : -5; // determine whether end of region or error
+				break;
+			}
 			else if (is_overlap(iter->beg, iter->end, b)) return ret;
-		} else break; // end of file
+		} else break; // end of file or error
 	}
 	iter->finished = 1;
-	return -1;
+	return ret;
 }
 
 int bam_fetch(bamFile fp, const bam_index_t *idx, int tid, int beg, int end, void *data, bam_fetch_f func)
 {
+	int ret;
 	bam_iter_t iter;
 	bam1_t *b;
 	b = bam_init1();
 	iter = bam_iter_query(idx, tid, beg, end);
-	while (bam_iter_read(fp, iter, b) >= 0) func(b, data);
+	while ((ret = bam_iter_read(fp, iter, b)) >= 0) func(b, data);
+	bam_iter_destroy(iter);
 	bam_destroy1(b);
-	return 0;
+	return (ret == -1)? 0 : ret;
 }