10 int32_t line_len, line_blen;
14 KHASH_MAP_INIT_STR(s, faidx1_t)
20 #define ftello(fp) ftell(fp)
21 #define fseeko(fp, offset, whence) fseek(fp, offset, whence)
23 extern off_t ftello(FILE *stream);
24 extern int fseeko(FILE *stream, off_t offset, int whence);
27 #define razf_read(fp, buf, size) fread(buf, 1, size, fp)
28 #define razf_open(fn, mode) fopen(fn, mode)
29 #define razf_close(fp) fclose(fp)
30 #define razf_seek(fp, offset, whence) fseeko(fp, offset, whence)
31 #define razf_tell(fp) ftello(fp)
45 #define kroundup32(x) (--(x), (x)|=(x)>>1, (x)|=(x)>>2, (x)|=(x)>>4, (x)|=(x)>>8, (x)|=(x)>>16, ++(x))
48 static inline void fai_insert_index(faidx_t *idx, const char *name, int len, int line_len, int line_blen, uint64_t offset)
53 if (idx->n == idx->m) {
54 idx->m = idx->m? idx->m<<1 : 16;
55 idx->name = (char**)realloc(idx->name, sizeof(void*) * idx->m);
57 idx->name[idx->n] = strdup(name);
58 k = kh_put(s, idx->hash, idx->name[idx->n], &ret);
59 t.len = len; t.line_len = line_len; t.line_blen = line_blen; t.offset = offset;
60 kh_value(idx->hash, k) = t;
64 faidx_t *fai_build_core(RAZF *rz)
67 int l_name, m_name, ret;
68 int line_len, line_blen, state;
74 idx = (faidx_t*)calloc(1, sizeof(faidx_t));
75 idx->hash = kh_init(s);
76 name = 0; l_name = m_name = 0;
77 len = line_len = line_blen = -1; state = 0; l1 = l2 = -1; offset = 0;
78 while (razf_read(rz, &c, 1)) {
79 if (c == '\n') { // an empty line
81 offset = razf_tell(rz);
83 } else if ((state == 0 && len < 0) || state == 2) continue;
85 if (c == '>') { // fasta header
87 fai_insert_index(idx, name, len, line_len, line_blen, offset);
89 while ((ret = razf_read(rz, &c, 1)) != 0 && !isspace(c)) {
90 if (m_name < l_name + 2) {
93 name = (char*)realloc(name, m_name);
99 fprintf(stderr, "[fai_build_core] the last entry has no sequence\n");
100 free(name); fai_destroy(idx);
103 if (c != '\n') while (razf_read(rz, &c, 1) && c != '\n');
105 offset = razf_tell(rz);
108 fprintf(stderr, "[fai_build_core] inlined empty line is not allowed in sequence '%s'.\n", name);
109 free(name); fai_destroy(idx);
112 if (state == 2) state = 3;
116 if (isgraph(c)) ++l2;
117 } while ((ret = razf_read(rz, &c, 1)) && c != '\n');
118 if (state == 3 && l2) {
119 fprintf(stderr, "[fai_build_core] different line length in sequence '%s'.\n", name);
120 free(name); fai_destroy(idx);
124 if (state == 1) line_len = l1, line_blen = l2, state = 0;
125 else if (state == 0) {
126 if (l1 != line_len || l2 != line_blen) state = 2;
130 fai_insert_index(idx, name, len, line_len, line_blen, offset);
135 void fai_save(const faidx_t *fai, FILE *fp)
139 for (i = 0; i < fai->n; ++i) {
141 k = kh_get(s, fai->hash, fai->name[i]);
142 x = kh_value(fai->hash, k);
144 fprintf(fp, "%s\t%d\t%ld\t%d\t%d\n", fai->name[i], (int)x.len, (long)x.offset, (int)x.line_blen, (int)x.line_len);
146 fprintf(fp, "%s\t%d\t%lld\t%d\t%d\n", fai->name[i], (int)x.len, (long long)x.offset, (int)x.line_blen, (int)x.line_len);
151 faidx_t *fai_read(FILE *fp)
155 int len, line_len, line_blen;
161 fai = (faidx_t*)calloc(1, sizeof(faidx_t));
162 fai->hash = kh_init(s);
163 buf = (char*)calloc(0x10000, 1);
164 while (!feof(fp) && fgets(buf, 0x10000, fp)) {
165 for (p = buf; *p && isgraph(*p); ++p);
168 sscanf(p, "%d%ld%d%d", &len, &offset, &line_blen, &line_len);
170 sscanf(p, "%d%lld%d%d", &len, &offset, &line_blen, &line_len);
172 fai_insert_index(fai, buf, len, line_len, line_blen, offset);
178 void fai_destroy(faidx_t *fai)
181 for (i = 0; i < fai->n; ++i) free(fai->name[i]);
183 kh_destroy(s, fai->hash);
184 if (fai->rz) razf_close(fai->rz);
188 int fai_build(const char *fn)
194 str = (char*)calloc(strlen(fn) + 5, 1);
195 sprintf(str, "%s.fai", fn);
196 rz = razf_open(fn, "r");
198 fprintf(stderr, "[fai_build] fail to open the FASTA file %s\n",fn);
202 fai = fai_build_core(rz);
204 fp = fopen(str, "wb");
206 fprintf(stderr, "[fai_build] fail to write FASTA index %s\n",str);
207 fai_destroy(fai); free(str);
218 FILE *download_and_open(const char *fn)
220 const int buf_size = 1 * 1024 * 1024;
224 const char *url = fn;
227 for (p = fn + l - 1; p >= fn; --p)
228 if (*p == '/') break;
231 // First try to open a local copy
236 // If failed, download from remote and open
237 fp_remote = knet_open(url, "rb");
238 if (fp_remote == 0) {
239 fprintf(stderr, "[download_from_remote] fail to open remote file %s\n",url);
242 if ((fp = fopen(fn, "wb")) == 0) {
243 fprintf(stderr, "[download_from_remote] fail to create file in the working directory %s\n",fn);
244 knet_close(fp_remote);
247 buf = (uint8_t*)calloc(buf_size, 1);
248 while ((l = knet_read(fp_remote, buf, buf_size)) != 0)
249 fwrite(buf, 1, l, fp);
252 knet_close(fp_remote);
254 return fopen(fn, "r");
258 faidx_t *fai_load(const char *fn)
263 str = (char*)calloc(strlen(fn) + 5, 1);
264 sprintf(str, "%s.fai", fn);
267 if (strstr(fn, "ftp://") == fn || strstr(fn, "http://") == fn)
269 fp = download_and_open(str);
272 fprintf(stderr, "[fai_load] failed to open remote FASTA index %s\n", str);
279 fp = fopen(str, "rb");
281 fprintf(stderr, "[fai_load] build FASTA index.\n");
283 fp = fopen(str, "rb");
285 fprintf(stderr, "[fai_load] fail to open FASTA index.\n");
294 fai->rz = razf_open(fn, "rb");
297 fprintf(stderr, "[fai_load] fail to open FASTA file.\n");
303 char *fai_fetch(const faidx_t *fai, const char *str, int *len)
306 int i, l, k, name_end;
314 name_end = l = strlen(str);
315 s = (char*)malloc(l+1);
317 for (i = k = 0; i < l; ++i)
318 if (!isspace(str[i])) s[k++] = str[i];
320 // determine the sequence name
321 for (i = l - 1; i >= 0; --i) if (s[i] == ':') break; // look for colon from the end
322 if (i >= 0) name_end = i;
323 if (name_end < l) { // check if this is really the end
325 for (i = name_end + 1; i < l; ++i) {
326 if (s[i] == '-') ++n_hyphen;
327 else if (!isdigit(s[i]) && s[i] != ',') break;
329 if (i < l || n_hyphen > 1) name_end = l; // malformated region string; then take str as the name
331 iter = kh_get(s, h, s);
332 if (iter == kh_end(h)) { // cannot find the sequence name
333 iter = kh_get(s, h, str); // try str as the name
334 if (iter == kh_end(h)) {
337 } else s[name_end] = ':', name_end = l;
339 } else iter = kh_get(s, h, str);
340 if(iter == kh_end(h)) {
341 fprintf(stderr, "[fai_fetch] Warning - Reference %s not found in FASTA file, returning empty sequence\n", str);
345 val = kh_value(h, iter);
346 // parse the interval
348 for (i = k = name_end + 1; i < l; ++i)
349 if (s[i] != ',') s[k++] = s[i];
351 beg = atoi(s + name_end + 1);
352 for (i = name_end + 1; i != k; ++i) if (s[i] == '-') break;
353 end = i < k? atoi(s + i + 1) : val.len;
355 } else beg = 0, end = val.len;
356 if (beg >= val.len) beg = val.len;
357 if (end >= val.len) end = val.len;
358 if (beg > end) beg = end;
361 // now retrieve the sequence
363 s = (char*)malloc(end - beg + 2);
364 razf_seek(fai->rz, val.offset + beg / val.line_blen * val.line_len + beg % val.line_blen, SEEK_SET);
365 while (razf_read(fai->rz, &c, 1) == 1 && l < end - beg && !fai->rz->z_err)
366 if (isgraph(c)) s[l++] = c;
372 int faidx_main(int argc, char *argv[])
375 fprintf(stderr, "Usage: faidx <in.fasta> [<reg> [...]]\n");
378 if (argc == 2) fai_build(argv[1]);
383 fai = fai_load(argv[1]);
384 if (fai == 0) return 1;
385 for (i = 2; i != argc; ++i) {
386 printf(">%s\n", argv[i]);
387 s = fai_fetch(fai, argv[i], &l);
388 for (j = 0; j < l; j += 60) {
389 for (k = 0; k < 60 && k < l - j; ++k)
401 int faidx_fetch_nseq(const faidx_t *fai)
406 char *faidx_fetch_seq(const faidx_t *fai, char *c_name, int p_beg_i, int p_end_i, int *len)
415 iter = kh_get(s, fai->hash, c_name);
416 if(iter == kh_end(fai->hash)) return 0;
417 val = kh_value(fai->hash, iter);
418 if(p_end_i < p_beg_i) p_beg_i = p_end_i;
419 if(p_beg_i < 0) p_beg_i = 0;
420 else if(val.len <= p_beg_i) p_beg_i = val.len - 1;
421 if(p_end_i < 0) p_end_i = 0;
422 else if(val.len <= p_end_i) p_end_i = val.len - 1;
424 // Now retrieve the sequence
426 seq = (char*)malloc(p_end_i - p_beg_i + 2);
427 razf_seek(fai->rz, val.offset + p_beg_i / val.line_blen * val.line_len + p_beg_i % val.line_blen, SEEK_SET);
428 while (razf_read(fai->rz, &c, 1) == 1 && l < p_end_i - p_beg_i + 1)
429 if (isgraph(c)) seq[l++] = c;
436 int main(int argc, char *argv[]) { return faidx_main(argc, argv); }