4 #include "bam_endian.h"
11 #define TAD_LIDX_SHIFT 13
27 static inline void insert_offset2(bcf_lidx_t *index2, int _beg, int _end, uint64_t offset)
30 beg = _beg >> TAD_LIDX_SHIFT;
31 end = (_end - 1) >> TAD_LIDX_SHIFT;
32 if (index2->m < end + 1) {
33 int old_m = index2->m;
35 kroundup32(index2->m);
36 index2->offset = (uint64_t*)realloc(index2->offset, index2->m * 8);
37 memset(index2->offset + old_m, 0, 8 * (index2->m - old_m));
40 if (index2->offset[beg] == 0) index2->offset[beg] = offset;
42 for (i = beg; i <= end; ++i)
43 if (index2->offset[i] == 0) index2->offset[i] = offset;
45 if (index2->n < end + 1) index2->n = end + 1;
48 bcf_idx_t *bcf_idx_core(bcf_t *bp, bcf_hdr_t *h)
51 int32_t last_coor, last_tid;
58 b = calloc(1, sizeof(bcf1_t));
59 str = calloc(1, sizeof(kstring_t));
60 idx = (bcf_idx_t*)calloc(1, sizeof(bcf_idx_t));
62 idx->index2 = calloc(h->n_ref, sizeof(bcf_lidx_t));
64 last_tid = 0xffffffffu;
65 last_off = bgzf_tell(fp); last_coor = 0xffffffffu;
66 while ((ret = bcf_read(bp, h, b)) > 0) {
68 if (last_tid != b->tid) { // change of chromosomes
70 } else if (last_coor > b->pos) {
71 fprintf(stderr, "[bcf_idx_core] the input is out of order\n");
72 free(str->s); free(str); free(idx); bcf_destroy(b);
76 end = b->pos + (tmp > 0? tmp : 1);
77 insert_offset2(&idx->index2[b->tid], b->pos, end, last_off);
78 last_off = bgzf_tell(fp);
81 free(str->s); free(str); bcf_destroy(b);
85 void bcf_idx_destroy(bcf_idx_t *idx)
89 for (i = 0; i < idx->n; ++i) free(idx->index2[i].offset);
98 void bcf_idx_save(const bcf_idx_t *idx, BGZF *fp)
101 ti_is_be = bam_is_big_endian();
102 bgzf_write(fp, "BCI\4", 4);
105 bgzf_write(fp, bam_swap_endian_4p(&x), 4);
106 } else bgzf_write(fp, &idx->n, 4);
107 for (i = 0; i < idx->n; ++i) {
108 bcf_lidx_t *index2 = idx->index2 + i;
109 // write linear index (index2)
112 bgzf_write(fp, bam_swap_endian_4p(&x), 4);
113 } else bgzf_write(fp, &index2->n, 4);
114 if (ti_is_be) { // big endian
116 for (x = 0; (int)x < index2->n; ++x)
117 bam_swap_endian_8p(&index2->offset[x]);
118 bgzf_write(fp, index2->offset, 8 * index2->n);
119 for (x = 0; (int)x < index2->n; ++x)
120 bam_swap_endian_8p(&index2->offset[x]);
121 } else bgzf_write(fp, index2->offset, 8 * index2->n);
125 static bcf_idx_t *bcf_idx_load_core(BGZF *fp)
130 ti_is_be = bam_is_big_endian();
132 fprintf(stderr, "[%s] fail to load index.\n", __func__);
135 bgzf_read(fp, magic, 4);
136 if (strncmp(magic, "BCI\4", 4)) {
137 fprintf(stderr, "[%s] wrong magic number.\n", __func__);
140 idx = (bcf_idx_t*)calloc(1, sizeof(bcf_idx_t));
141 bgzf_read(fp, &idx->n, 4);
142 if (ti_is_be) bam_swap_endian_4p(&idx->n);
143 idx->index2 = (bcf_lidx_t*)calloc(idx->n, sizeof(bcf_lidx_t));
144 for (i = 0; i < idx->n; ++i) {
145 bcf_lidx_t *index2 = idx->index2 + i;
147 bgzf_read(fp, &index2->n, 4);
148 if (ti_is_be) bam_swap_endian_4p(&index2->n);
149 index2->m = index2->n;
150 index2->offset = (uint64_t*)calloc(index2->m, 8);
151 bgzf_read(fp, index2->offset, index2->n * 8);
153 for (j = 0; j < index2->n; ++j) bam_swap_endian_8p(&index2->offset[j]);
158 bcf_idx_t *bcf_idx_load_local(const char *fnidx)
161 fp = bgzf_open(fnidx, "r");
163 bcf_idx_t *idx = bcf_idx_load_core(fp);
170 static void download_from_remote(const char *url)
172 const int buf_size = 1 * 1024 * 1024;
178 if (strstr(url, "ftp://") != url && strstr(url, "http://") != url) return;
180 for (fn = (char*)url + l - 1; fn >= url; --fn)
181 if (*fn == '/') break;
182 ++fn; // fn now points to the file name
183 fp_remote = knet_open(url, "r");
184 if (fp_remote == 0) {
185 fprintf(stderr, "[download_from_remote] fail to open remote file.\n");
188 if ((fp = fopen(fn, "w")) == 0) {
189 fprintf(stderr, "[download_from_remote] fail to create file in the working directory.\n");
190 knet_close(fp_remote);
193 buf = (uint8_t*)calloc(buf_size, 1);
194 while ((l = knet_read(fp_remote, buf, buf_size)) != 0)
195 fwrite(buf, 1, l, fp);
198 knet_close(fp_remote);
201 static void download_from_remote(const char *url)
207 static char *get_local_version(const char *fn)
210 char *fnidx = (char*)calloc(strlen(fn) + 5, 1);
211 strcat(strcpy(fnidx, fn), ".bci");
212 if ((strstr(fnidx, "ftp://") == fnidx || strstr(fnidx, "http://") == fnidx)) {
214 int l = strlen(fnidx);
215 for (p = fnidx + l - 1; p >= fnidx; --p)
216 if (*p == '/') break;
217 url = fnidx; fnidx = strdup(p + 1);
218 if (stat(fnidx, &sbuf) == 0) {
222 fprintf(stderr, "[%s] downloading the index file...\n", __func__);
223 download_from_remote(url);
226 if (stat(fnidx, &sbuf) == 0) return fnidx;
227 free(fnidx); return 0;
230 bcf_idx_t *bcf_idx_load(const char *fn)
233 char *fname = get_local_version(fn);
234 if (fname == 0) return 0;
235 idx = bcf_idx_load_local(fname);
240 int bcf_idx_build2(const char *fn, const char *_fnidx)
247 if ((bp = bcf_open(fn, "r")) == 0) {
248 fprintf(stderr, "[bcf_idx_build2] fail to open the BAM file.\n");
251 h = bcf_hdr_read(bp);
252 idx = bcf_idx_core(bp, h);
255 fnidx = (char*)calloc(strlen(fn) + 5, 1);
256 strcpy(fnidx, fn); strcat(fnidx, ".bci");
257 } else fnidx = strdup(_fnidx);
258 fpidx = bgzf_open(fnidx, "w");
260 fprintf(stderr, "[bcf_idx_build2] fail to create the index file.\n");
264 bcf_idx_save(idx, fpidx);
265 bcf_idx_destroy(idx);
271 int bcf_idx_build(const char *fn)
273 return bcf_idx_build2(fn, 0);
276 /********************************************
277 * parse a region in the format chr:beg-end *
278 ********************************************/
280 int bcf_parse_region(void *str2id, const char *str, int *tid, int *begin, int *end)
285 p = s = (char*)malloc(l+1);
286 /* squeeze out "," */
287 for (i = k = 0; i != l; ++i)
288 if (str[i] != ',' && !isspace(str[i])) s[k++] = str[i];
290 for (i = 0; i != k; ++i) if (s[i] == ':') break;
292 if ((*tid = bcf_str2id(str2id, s)) < 0) {
296 if (i == k) { /* dump the whole sequence */
297 *begin = 0; *end = 1<<29; free(s);
300 for (p = s + i + 1; i != k; ++i) if (s[i] == '-') break;
306 if (*begin > 0) --*begin;
308 if (*begin > *end) return -1;
312 /*******************************
313 * retrieve a specified region *
314 *******************************/
316 uint64_t bcf_idx_query(const bcf_idx_t *idx, int tid, int beg)
318 uint64_t min_off, *offset;
320 if (beg < 0) beg = 0;
321 offset = idx->index2[tid].offset;
322 for (i = beg>>TAD_LIDX_SHIFT; i < idx->index2[tid].n && offset[i] == 0; ++i);
323 min_off = (i == idx->index2[tid].n)? offset[idx->index2[tid].n-1] : offset[i];
327 int bcf_main_index(int argc, char *argv[])
330 fprintf(stderr, "Usage: bcftools index <in.bcf>\n");
333 bcf_idx_build(argv[1]);