8 KSORT_INIT_GENERIC(uint64_t)
11 KSTREAM_INIT(gzFile, gzread, 8192)
20 KHASH_MAP_INIT_STR(reg, bed_reglist_t)
24 typedef kh_reg_t reghash_t;
26 int *bed_index_core(int n, uint64_t *a, int *n_idx)
29 m = *n_idx = 0; idx = 0;
30 for (i = 0; i < n; ++i) {
32 beg = a[i]>>32 >> LIDX_SHIFT; end = ((uint32_t)a[i]) >> LIDX_SHIFT;
37 idx = realloc(idx, m * sizeof(int));
38 for (j = oldm; j < m; ++j) idx[j] = -1;
41 if (idx[beg] < 0) idx[beg] = i;
43 for (j = beg; j <= end; ++j)
44 if (idx[j] < 0) idx[j] = i;
51 void bed_index(void *_h)
53 reghash_t *h = (reghash_t*)_h;
55 for (k = 0; k < kh_end(h); ++k) {
57 bed_reglist_t *p = &kh_val(h, k);
58 if (p->idx) free(p->idx);
59 ks_introsort(uint64_t, p->n, p->a);
60 p->idx = bed_index_core(p->n, p->a, &p->m);
65 int bed_overlap_core(const bed_reglist_t *p, int beg, int end)
68 if (p->n == 0) return 0;
69 min_off = (beg>>LIDX_SHIFT >= p->n)? p->idx[p->n-1] : p->idx[beg>>LIDX_SHIFT];
70 if (min_off < 0) { // TODO: this block can be improved, but speed should not matter too much here
71 int n = beg>>LIDX_SHIFT;
72 if (n > p->n) n = p->n;
73 for (i = n - 1; i >= 0; --i)
74 if (p->idx[i] >= 0) break;
75 min_off = i >= 0? p->idx[i] : 0;
77 for (i = min_off; i < p->n; ++i) {
78 if ((int)(p->a[i]>>32) >= end) break; // out of range; no need to proceed
79 if ((int32_t)p->a[i] > beg && (int32_t)(p->a[i]>>32) < end)
80 return 1; // find the overlap; return
85 int bed_overlap(const void *_h, const char *chr, int beg, int end)
87 const reghash_t *h = (const reghash_t*)_h;
90 k = kh_get(reg, h, chr);
91 if (k == kh_end(h)) return 0;
92 return bed_overlap_core(&kh_val(h, k), beg, end);
95 void *bed_read(const char *fn)
97 reghash_t *h = kh_init(reg);
103 fp = strcmp(fn, "-")? gzopen(fn, "r") : gzdopen(fileno(stdin), "r");
104 if (fp == 0) return 0;
105 str = calloc(1, sizeof(kstring_t));
107 while (ks_getuntil(ks, 0, str, &dret) >= 0) { // read the chr name
108 int beg = -1, end = -1;
110 khint_t k = kh_get(reg, h, str->s);
111 if (k == kh_end(h)) { // absent from the hash table
113 char *s = strdup(str->s);
114 k = kh_put(reg, h, s, &ret);
115 memset(&kh_val(h, k), 0, sizeof(bed_reglist_t));
118 if (dret != '\n') { // if the lines has other characters
119 if (ks_getuntil(ks, 0, str, &dret) > 0 && isdigit(str->s[0])) {
120 beg = atoi(str->s); // begin
122 if (ks_getuntil(ks, 0, str, &dret) > 0 && isdigit(str->s[0]))
123 end = atoi(str->s); // end
127 if (dret != '\n') while ((dret = ks_getc(ks)) > 0 && dret != '\n'); // skip the rest of the line
128 if (end < 0 && beg > 0) end = beg, beg = beg - 1; // if there is only one column
129 if (beg >= 0 && end > beg) {
131 p->m = p->m? p->m<<1 : 4;
132 p->a = realloc(p->a, p->m * 8);
134 p->a[p->n++] = (uint64_t)beg<<32 | end;
139 free(str->s); free(str);
144 void bed_destroy(void *_h)
146 reghash_t *h = (reghash_t*)_h;
148 for (k = 0; k < kh_end(h); ++k) {
149 if (kh_exist(h, k)) {
150 free(kh_val(h, k).a);
151 free(kh_val(h, k).idx);
152 free((char*)kh_key(h, k));