13 KSTREAM_INIT(gzFile, gzread, 8192)
14 KHASH_MAP_INIT_STR(ref, uint64_t)
16 void bam_init_header_hash(bam_header_t *header);
17 void bam_destroy_header_hash(bam_header_t *header);
18 int32_t bam_get_tid(const bam_header_t *header, const char *seq_name);
20 unsigned char bam_nt16_table[256] = {
21 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
22 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
23 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
24 1, 2, 4, 8, 15,15,15,15, 15,15,15,15, 15, 0 /*=*/,15,15,
25 15, 1,14, 2, 13,15,15, 4, 11,15,15,12, 15, 3,15,15,
26 15,15, 5, 6, 8,15, 7, 9, 15,10,15,15, 15,15,15,15,
27 15, 1,14, 2, 13,15,15, 4, 11,15,15,12, 15, 3,15,15,
28 15,15, 5, 6, 8,15, 7, 9, 15,10,15,15, 15,15,15,15,
29 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
30 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
31 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
32 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
33 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
34 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
35 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
36 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15
39 unsigned short bam_char2flag_table[256] = {
40 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
41 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
42 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
43 0,BAM_FREAD1,BAM_FREAD2,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
44 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
45 BAM_FPROPER_PAIR,0,BAM_FMREVERSE,0, 0,BAM_FMUNMAP,0,0, 0,0,0,0, 0,0,0,0,
46 0,0,0,0, BAM_FDUP,0,BAM_FQCFAIL,0, 0,0,0,0, 0,0,0,0,
47 BAM_FPAIRED,0,BAM_FREVERSE,BAM_FSECONDARY, 0,BAM_FUNMAP,0,0, 0,0,0,0, 0,0,0,0,
48 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
49 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
50 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
51 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
52 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
53 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
54 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
55 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0
58 char *bam_nt16_rev_table = "=ACMGRSVTWYHKDBN";
68 char **__bam_get_lines(const char *fn, int *_n) // for bam_plcmd.c only
71 int n = 0, dret, m = 0;
72 gzFile fp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
75 str = (kstring_t*)calloc(1, sizeof(kstring_t));
77 while (ks_getuntil(ks, '\n', str, &dret) > 0) {
80 list = (char**)realloc(list, m * sizeof(char*));
82 if (str->s[str->l-1] == '\r')
83 str->s[--str->l] = '\0';
84 s = list[n++] = (char*)calloc(str->l + 1, 1);
89 free(str->s); free(str);
94 static bam_header_t *hash2header(const kh_ref_t *hash)
98 header = bam_header_init();
99 header->n_targets = kh_size(hash);
100 header->target_name = (char**)calloc(kh_size(hash), sizeof(char*));
101 header->target_len = (uint32_t*)calloc(kh_size(hash), 4);
102 for (k = kh_begin(hash); k != kh_end(hash); ++k) {
103 if (kh_exist(hash, k)) {
104 int i = (int)kh_value(hash, k);
105 header->target_name[i] = (char*)kh_key(hash, k);
106 header->target_len[i] = kh_value(hash, k)>>32;
109 bam_init_header_hash(header);
112 bam_header_t *sam_header_read2(const char *fn)
114 bam_header_t *header;
122 fp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
125 str = (kstring_t*)calloc(1, sizeof(kstring_t));
126 while (ks_getuntil(ks, 0, str, &dret) > 0) {
127 char *s = strdup(str->s);
130 ks_getuntil(ks, 0, str, &dret);
132 k = kh_put(ref, hash, s, &ret);
133 kh_value(hash, k) = (uint64_t)len<<32 | i;
135 while ((c = ks_getc(ks)) != '\n' && c != -1);
139 free(str->s); free(str);
140 fprintf(stderr, "[sam_header_read2] %d sequences loaded.\n", kh_size(hash));
141 header = hash2header(hash);
142 kh_destroy(ref, hash);
145 static inline uint8_t *alloc_data(bam1_t *b, int size)
147 if (b->m_data < size) {
149 kroundup32(b->m_data);
150 b->data = (uint8_t*)realloc(b->data, b->m_data);
154 static inline void parse_error(int64_t n_lines, const char * __restrict msg)
156 fprintf(stderr, "Parse error at line %lld: %s\n", (long long)n_lines, msg);
159 static inline void append_text(bam_header_t *header, kstring_t *str)
161 int x = header->l_text, y = header->l_text + str->l + 2; // 2 = 1 byte dret + 1 byte null
162 kroundup32(x); kroundup32(y);
163 if (x < y) header->text = (char*)realloc(header->text, y);
164 strncpy(header->text + header->l_text, str->s, str->l+1); // we cannot use strcpy() here.
165 header->l_text += str->l + 1;
166 header->text[header->l_text] = 0;
169 int sam_header_parse_rg(bam_header_t *h)
171 kstring_t *rgid, *rglib;
176 if (h == 0) return 0;
177 bam_strmap_destroy(h->rg2lib); h->rg2lib = 0;
178 if (h->l_text < 3) return 0;
180 h->rg2lib = bam_strmap_init();
181 rgid = calloc(1, sizeof(kstring_t));
182 rglib = calloc(1, sizeof(kstring_t));
184 while ((s = strstr(s, "@RG")) != 0) {
185 if (rgid->l && rglib->l) {
186 bam_strmap_put(h->rg2lib, rgid->s, rglib->s);
189 rgid->l = rglib->l = 0;
192 if ((p = strstr(s, "ID:")) != 0) {
194 for (p = q; *p && *p != '\t' && *p != '\r' && *p != '\n'; ++p);
195 kputsn(q, p - q, rgid);
197 fprintf(stderr, "[bam_header_parse] missing ID tag in @RG lines.\n");
201 if ((p = strstr(s, "LB:")) != 0) {
203 for (p = q; *p && *p != '\t' && *p != '\r' && *p != '\n'; ++p);
204 kputsn(q, p - q, rglib);
206 fprintf(stderr, "[bam_header_parse] missing LB tag in @RG lines.\n");
212 if (rgid->l && rglib->l) {
213 bam_strmap_put(h->rg2lib, rgid->s, rglib->s);
216 free(rgid->s); free(rgid);
217 free(rglib->s); free(rglib);
219 bam_strmap_destroy(h->rg2lib);
225 int sam_header_parse(bam_header_t *h)
231 free(h->target_len); free(h->target_name);
232 h->n_targets = 0; h->target_len = 0; h->target_name = 0;
233 if (h->l_text < 3) return 0;
234 // count number of @SQ
236 while ((s = strstr(s, "@SQ")) != 0) {
240 if (h->n_targets == 0) return 0;
241 h->target_len = (uint32_t*)calloc(h->n_targets, 4);
242 h->target_name = (char**)calloc(h->n_targets, sizeof(void*));
246 while ((s = strstr(s, "@SQ")) != 0) {
249 if ((p = strstr(s, "SN:")) != 0) {
251 for (p = q; *p && *p != '\t' && *p != '\r' && *p != '\n'; ++p);
252 h->target_name[i] = (char*)calloc(p - q + 1, 1);
253 strncpy(h->target_name[i], q, p - q);
254 } else goto header_err_ret;
256 if ((p = strstr(s, "LN:")) != 0) h->target_len[i] = strtol(p + 3, 0, 10);
257 else goto header_err_ret;
262 sam_header_parse_rg(h);
266 fprintf(stderr, "[bam_header_parse] missing SN or LN tag in @SQ lines.\n");
267 free(h->target_len); free(h->target_name);
268 h->n_targets = 0; h->target_len = 0; h->target_name = 0;
272 bam_header_t *sam_header_read(tamFile fp)
275 bam_header_t *header = bam_header_init();
276 kstring_t *str = fp->str;
277 while ((ret = ks_getuntil(fp->ks, KS_SEP_TAB, str, &dret)) >= 0 && str->s[0] == '@') { // skip header
278 str->s[str->l] = dret; // note that str->s is NOT null terminated!!
279 append_text(header, str);
281 ret = ks_getuntil(fp->ks, '\n', str, &dret);
282 str->s[str->l] = '\n'; // NOT null terminated!!
283 append_text(header, str);
287 sam_header_parse(header);
288 bam_init_header_hash(header);
293 int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
295 int ret, doff, doff0, dret, z = 0;
296 bam1_core_t *c = &b->core;
297 kstring_t *str = fp->str;
298 kstream_t *ks = fp->ks;
304 do { // special consideration for empty lines
305 ret = ks_getuntil(fp->ks, KS_SEP_TAB, str, &dret);
306 if (ret >= 0) z += str->l + 1;
309 if (ret < 0) return -1;
314 c->l_qname = strlen(str->s) + 1;
315 memcpy(alloc_data(b, doff + c->l_qname) + doff, str->s, c->l_qname);
321 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
322 flag = strtol((char*)str->s, &s, 0);
323 if (*s) { // not the end of the string
325 for (s = str->s; *s; ++s)
326 flag |= bam_char2flag_table[(int)*s];
331 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->tid = bam_get_tid(header, str->s);
332 if (c->tid < 0 && strcmp(str->s, "*")) {
333 if (header->n_targets == 0) {
334 fprintf(stderr, "[sam_read1] missing header? Abort!\n");
336 } else fprintf(stderr, "[sam_read1] reference '%s' is recognized as '*'.\n", str->s);
338 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->pos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
339 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->qual = isdigit(str->s[0])? atoi(str->s) : 0;
340 if (ret < 0) return -2;
347 if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -3;
349 if (str->s[0] != '*') {
350 for (s = str->s; *s; ++s) {
351 if (isalpha(*s)) ++c->n_cigar;
352 else if (!isdigit(*s)) parse_error(fp->n_lines, "invalid CIGAR character");
354 b->data = alloc_data(b, doff + c->n_cigar * 4);
355 for (i = 0, s = str->s; i != c->n_cigar; ++i) {
356 x = strtol(s, &t, 10);
358 if (op == 'M') op = BAM_CMATCH;
359 else if (op == 'I') op = BAM_CINS;
360 else if (op == 'D') op = BAM_CDEL;
361 else if (op == 'N') op = BAM_CREF_SKIP;
362 else if (op == 'S') op = BAM_CSOFT_CLIP;
363 else if (op == 'H') op = BAM_CHARD_CLIP;
364 else if (op == 'P') op = BAM_CPAD;
365 else parse_error(fp->n_lines, "invalid CIGAR operation");
367 bam1_cigar(b)[i] = x << BAM_CIGAR_SHIFT | op;
369 if (*s) parse_error(fp->n_lines, "unmatched CIGAR operation");
370 c->bin = bam_reg2bin(c->pos, bam_calend(c, bam1_cigar(b)));
371 doff += c->n_cigar * 4;
372 } else c->bin = bam_reg2bin(c->pos, c->pos + 1);
374 { // mtid, mpos, isize
375 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
376 c->mtid = strcmp(str->s, "=")? bam_get_tid(header, str->s) : c->tid;
377 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
378 c->mpos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
379 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
380 c->isize = (str->s[0] == '-' || isdigit(str->s[0]))? atoi(str->s) : 0;
381 if (ret < 0) return -4;
386 if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -5; // seq
388 c->l_qseq = strlen(str->s);
389 if (c->n_cigar && c->l_qseq != (int32_t)bam_cigar2qlen(c, bam1_cigar(b)))
390 parse_error(fp->n_lines, "CIGAR and sequence length are inconsistent");
391 p = (uint8_t*)alloc_data(b, doff + c->l_qseq + (c->l_qseq+1)/2) + doff;
392 memset(p, 0, (c->l_qseq+1)/2);
393 for (i = 0; i < c->l_qseq; ++i)
394 p[i/2] |= bam_nt16_table[(int)str->s[i]] << 4*(1-i%2);
395 if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -6; // qual
397 if (strcmp(str->s, "*") && c->l_qseq != strlen(str->s))
398 parse_error(fp->n_lines, "sequence and quality are inconsistent");
399 p += (c->l_qseq+1)/2;
400 if (strcmp(str->s, "*") == 0) for (i = 0; i < c->l_qseq; ++i) p[i] = 0xff;
401 else for (i = 0; i < c->l_qseq; ++i) p[i] = str->s[i] - 33;
402 doff += c->l_qseq + (c->l_qseq+1)/2;
405 if (dret != '\n' && dret != '\r') { // aux
406 while (ks_getuntil(ks, KS_SEP_TAB, str, &dret) >= 0) {
407 uint8_t *s, type, key[2];
409 if (str->l < 6 || str->s[2] != ':' || str->s[4] != ':')
410 parse_error(fp->n_lines, "missing colon in auxiliary data");
411 key[0] = str->s[0]; key[1] = str->s[1];
413 s = alloc_data(b, doff + 3) + doff;
414 s[0] = key[0]; s[1] = key[1]; s += 2; doff += 2;
415 if (type == 'A' || type == 'a' || type == 'c' || type == 'C') { // c and C for backward compatibility
416 s = alloc_data(b, doff + 2) + doff;
417 *s++ = 'A'; *s = str->s[5];
419 } else if (type == 'I' || type == 'i') {
421 s = alloc_data(b, doff + 5) + doff;
422 x = (long long)atoll(str->s + 5);
425 *s++ = 'c'; *(int8_t*)s = (int8_t)x;
427 } else if (x >= -32767) {
428 *s++ = 's'; *(int16_t*)s = (int16_t)x;
431 *s++ = 'i'; *(int32_t*)s = (int32_t)x;
433 if (x < -2147483648ll)
434 fprintf(stderr, "Parse warning at line %lld: integer %lld is out of range.",
435 (long long)fp->n_lines, x);
439 *s++ = 'C'; *s++ = (uint8_t)x;
441 } else if (x <= 65535) {
442 *s++ = 'S'; *(uint16_t*)s = (uint16_t)x;
445 *s++ = 'I'; *(uint32_t*)s = (uint32_t)x;
447 if (x > 4294967295ll)
448 fprintf(stderr, "Parse warning at line %lld: integer %lld is out of range.",
449 (long long)fp->n_lines, x);
452 } else if (type == 'f') {
453 s = alloc_data(b, doff + 5) + doff;
455 *(float*)s = (float)atof(str->s + 5);
457 } else if (type == 'd') {
458 s = alloc_data(b, doff + 9) + doff;
460 *(float*)s = (float)atof(str->s + 9);
462 } else if (type == 'Z' || type == 'H') {
463 int size = 1 + (str->l - 5) + 1;
464 if (type == 'H') { // check whether the hex string is valid
466 if ((str->l - 5) % 2 == 1) parse_error(fp->n_lines, "length of the hex string not even");
467 for (i = 0; i < str->l - 5; ++i) {
468 int c = toupper(str->s[5 + i]);
469 if (!((c >= '0' && c <= '9') || (c >= 'A' && c <= 'F')))
470 parse_error(fp->n_lines, "invalid hex character");
473 s = alloc_data(b, doff + size) + doff;
475 memcpy(s, str->s + 5, str->l - 5);
478 } else parse_error(fp->n_lines, "unrecognized type");
479 if (dret == '\n' || dret == '\r') break;
482 b->l_aux = doff - doff0;
487 tamFile sam_open(const char *fn)
490 gzFile gzfp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
491 if (gzfp == 0) return 0;
492 fp = (tamFile)calloc(1, sizeof(struct __tamFile_t));
493 fp->str = (kstring_t*)calloc(1, sizeof(kstring_t));
495 fp->ks = ks_init(fp->fp);
499 void sam_close(tamFile fp)
504 free(fp->str->s); free(fp->str);