16 KSTREAM_INIT(gzFile, gzread, 8192)
17 KHASH_MAP_INIT_STR(ref, uint64_t)
19 void bam_init_header_hash(bam_header_t *header);
20 void bam_destroy_header_hash(bam_header_t *header);
21 int32_t bam_get_tid(const bam_header_t *header, const char *seq_name);
23 unsigned char bam_nt16_table[256] = {
24 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
25 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
26 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
27 1, 2, 4, 8, 15,15,15,15, 15,15,15,15, 15, 0 /*=*/,15,15,
28 15, 1,14, 2, 13,15,15, 4, 11,15,15,12, 15, 3,15,15,
29 15,15, 5, 6, 8,15, 7, 9, 15,10,15,15, 15,15,15,15,
30 15, 1,14, 2, 13,15,15, 4, 11,15,15,12, 15, 3,15,15,
31 15,15, 5, 6, 8,15, 7, 9, 15,10,15,15, 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,
37 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
38 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
39 15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15
42 unsigned short bam_char2flag_table[256] = {
43 0,0,0,0, 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 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
46 0,BAM_FREAD1,BAM_FREAD2,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
47 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
48 BAM_FPROPER_PAIR,0,BAM_FMREVERSE,0, 0,BAM_FMUNMAP,0,0, 0,0,0,0, 0,0,0,0,
49 0,0,0,0, BAM_FDUP,0,BAM_FQCFAIL,0, 0,0,0,0, 0,0,0,0,
50 BAM_FPAIRED,0,BAM_FREVERSE,BAM_FSECONDARY, 0,BAM_FUNMAP,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,
56 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
57 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
58 0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0
61 char *bam_nt16_rev_table = "=ACMGRSVTWYHKDBN";
71 char **__bam_get_lines(const char *fn, int *_n) // for bam_plcmd.c only
74 int n = 0, dret, m = 0;
75 gzFile fp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
78 str = (kstring_t*)calloc(1, sizeof(kstring_t));
80 while (ks_getuntil(ks, '\n', str, &dret) > 0) {
83 list = (char**)realloc(list, m * sizeof(char*));
85 if (str->s[str->l-1] == '\r')
86 str->s[--str->l] = '\0';
87 s = list[n++] = (char*)calloc(str->l + 1, 1);
92 free(str->s); free(str);
97 static bam_header_t *hash2header(const kh_ref_t *hash)
101 header = bam_header_init();
102 header->n_targets = kh_size(hash);
103 header->target_name = (char**)calloc(kh_size(hash), sizeof(char*));
104 header->target_len = (uint32_t*)calloc(kh_size(hash), 4);
105 for (k = kh_begin(hash); k != kh_end(hash); ++k) {
106 if (kh_exist(hash, k)) {
107 int i = (int)kh_value(hash, k);
108 header->target_name[i] = (char*)kh_key(hash, k);
109 header->target_len[i] = kh_value(hash, k)>>32;
112 bam_init_header_hash(header);
115 bam_header_t *sam_header_read2(const char *fn)
117 bam_header_t *header;
124 if (fn == 0) return 0;
125 fp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
126 if (fp == 0) return 0;
129 str = (kstring_t*)calloc(1, sizeof(kstring_t));
130 while (ks_getuntil(ks, 0, str, &dret) > 0) {
131 char *s = strdup(str->s);
134 ks_getuntil(ks, 0, str, &dret);
136 k = kh_put(ref, hash, s, &ret);
137 kh_value(hash, k) = (uint64_t)len<<32 | i;
139 while ((c = ks_getc(ks)) != '\n' && c != -1);
143 free(str->s); free(str);
144 fprintf(stderr, "[sam_header_read2] %d sequences loaded.\n", kh_size(hash));
145 header = hash2header(hash);
146 kh_destroy(ref, hash);
149 static inline uint8_t *alloc_data(bam1_t *b, int size)
151 if (b->m_data < size) {
153 kroundup32(b->m_data);
154 b->data = (uint8_t*)realloc(b->data, b->m_data);
158 static inline void parse_error(int64_t n_lines, const char * __restrict msg)
160 fprintf(stderr, "Parse error at line %lld: %s\n", (long long)n_lines, msg);
163 static inline void append_text(bam_header_t *header, kstring_t *str)
165 int x = header->l_text, y = header->l_text + str->l + 2; // 2 = 1 byte dret + 1 byte null
166 kroundup32(x); kroundup32(y);
167 if (x < y) header->text = (char*)realloc(header->text, y);
168 strncpy(header->text + header->l_text, str->s, str->l+1); // we cannot use strcpy() here.
169 header->l_text += str->l + 1;
170 header->text[header->l_text] = 0;
173 int sam_header_parse_rg(bam_header_t *h)
175 kstring_t *rgid, *rglib;
180 if (h == 0) return 0;
181 bam_strmap_destroy(h->rg2lib); h->rg2lib = 0;
182 if (h->l_text < 3) return 0;
184 h->rg2lib = bam_strmap_init();
185 rgid = calloc(1, sizeof(kstring_t));
186 rglib = calloc(1, sizeof(kstring_t));
188 while ((s = strstr(s, "@RG")) != 0) {
189 if (rgid->l && rglib->l) {
190 bam_strmap_put(h->rg2lib, rgid->s, rglib->s);
193 rgid->l = rglib->l = 0;
196 if ((p = strstr(s, "ID:")) != 0) {
198 for (p = q; *p && *p != '\t' && *p != '\r' && *p != '\n'; ++p);
199 kputsn(q, p - q, rgid);
201 fprintf(stderr, "[bam_header_parse] missing ID tag in @RG lines.\n");
205 if ((p = strstr(s, "LB:")) != 0) {
207 for (p = q; *p && *p != '\t' && *p != '\r' && *p != '\n'; ++p);
208 kputsn(q, p - q, rglib);
210 fprintf(stderr, "[bam_header_parse] missing LB tag in @RG lines.\n");
216 if (rgid->l && rglib->l) {
217 bam_strmap_put(h->rg2lib, rgid->s, rglib->s);
220 free(rgid->s); free(rgid);
221 free(rglib->s); free(rglib);
223 bam_strmap_destroy(h->rg2lib);
229 int sam_header_parse(bam_header_t *h)
235 free(h->target_len); free(h->target_name);
236 h->n_targets = 0; h->target_len = 0; h->target_name = 0;
237 if (h->l_text < 3) return 0;
238 // count number of @SQ
240 while ((s = strstr(s, "@SQ")) != 0) {
244 if (h->n_targets == 0) return 0;
245 h->target_len = (uint32_t*)calloc(h->n_targets, 4);
246 h->target_name = (char**)calloc(h->n_targets, sizeof(void*));
250 while ((s = strstr(s, "@SQ")) != 0) {
253 if ((p = strstr(s, "SN:")) != 0) {
255 for (p = q; *p && *p != '\t' && *p != '\r' && *p != '\n'; ++p);
256 h->target_name[i] = (char*)calloc(p - q + 1, 1);
257 strncpy(h->target_name[i], q, p - q);
258 } else goto header_err_ret;
260 if ((p = strstr(s, "LN:")) != 0) h->target_len[i] = strtol(p + 3, 0, 10);
261 else goto header_err_ret;
266 sam_header_parse_rg(h);
270 fprintf(stderr, "[bam_header_parse] missing SN or LN tag in @SQ lines.\n");
271 free(h->target_len); free(h->target_name);
272 h->n_targets = 0; h->target_len = 0; h->target_name = 0;
276 bam_header_t *sam_header_read(tamFile fp)
279 bam_header_t *header = bam_header_init();
280 kstring_t *str = fp->str;
281 while ((ret = ks_getuntil(fp->ks, KS_SEP_TAB, str, &dret)) >= 0 && str->s[0] == '@') { // skip header
282 str->s[str->l] = dret; // note that str->s is NOT null terminated!!
283 append_text(header, str);
285 ret = ks_getuntil(fp->ks, '\n', str, &dret);
286 str->s[str->l] = '\n'; // NOT null terminated!!
287 append_text(header, str);
291 sam_header_parse(header);
292 bam_init_header_hash(header);
297 int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
299 int ret, doff, doff0, dret, z = 0;
300 bam1_core_t *c = &b->core;
301 kstring_t *str = fp->str;
302 kstream_t *ks = fp->ks;
308 do { // special consideration for empty lines
309 ret = ks_getuntil(fp->ks, KS_SEP_TAB, str, &dret);
310 if (ret >= 0) z += str->l + 1;
313 if (ret < 0) return -1;
318 c->l_qname = strlen(str->s) + 1;
319 memcpy(alloc_data(b, doff + c->l_qname) + doff, str->s, c->l_qname);
325 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
326 flag = strtol((char*)str->s, &s, 0);
327 if (*s) { // not the end of the string
329 for (s = str->s; *s; ++s)
330 flag |= bam_char2flag_table[(int)*s];
335 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->tid = bam_get_tid(header, str->s);
336 if (c->tid < 0 && strcmp(str->s, "*")) {
337 if (header->n_targets == 0) {
338 fprintf(stderr, "[sam_read1] missing header? Abort!\n");
340 } else fprintf(stderr, "[sam_read1] reference '%s' is recognized as '*'.\n", str->s);
342 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->pos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
343 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->qual = isdigit(str->s[0])? atoi(str->s) : 0;
344 if (ret < 0) return -2;
351 if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -3;
353 if (str->s[0] != '*') {
354 for (s = str->s; *s; ++s) {
355 if (isalpha(*s)) ++c->n_cigar;
356 else if (!isdigit(*s)) parse_error(fp->n_lines, "invalid CIGAR character");
358 b->data = alloc_data(b, doff + c->n_cigar * 4);
359 for (i = 0, s = str->s; i != c->n_cigar; ++i) {
360 x = strtol(s, &t, 10);
362 if (op == 'M' || op == '=' || op == 'X') op = BAM_CMATCH;
363 else if (op == 'I') op = BAM_CINS;
364 else if (op == 'D') op = BAM_CDEL;
365 else if (op == 'N') op = BAM_CREF_SKIP;
366 else if (op == 'S') op = BAM_CSOFT_CLIP;
367 else if (op == 'H') op = BAM_CHARD_CLIP;
368 else if (op == 'P') op = BAM_CPAD;
369 else parse_error(fp->n_lines, "invalid CIGAR operation");
371 bam1_cigar(b)[i] = x << BAM_CIGAR_SHIFT | op;
373 if (*s) parse_error(fp->n_lines, "unmatched CIGAR operation");
374 c->bin = bam_reg2bin(c->pos, bam_calend(c, bam1_cigar(b)));
375 doff += c->n_cigar * 4;
377 if (!(c->flag&BAM_FUNMAP)) {
378 fprintf(stderr, "Parse warning at line %lld: mapped sequence without CIGAR\n", (long long)fp->n_lines);
379 c->flag |= BAM_FUNMAP;
381 c->bin = bam_reg2bin(c->pos, c->pos + 1);
384 { // mtid, mpos, isize
385 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
386 c->mtid = strcmp(str->s, "=")? bam_get_tid(header, str->s) : c->tid;
387 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
388 c->mpos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
389 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
390 c->isize = (str->s[0] == '-' || isdigit(str->s[0]))? atoi(str->s) : 0;
391 if (ret < 0) return -4;
396 if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -5; // seq
398 if (strcmp(str->s, "*")) {
399 c->l_qseq = strlen(str->s);
400 if (c->n_cigar && c->l_qseq != (int32_t)bam_cigar2qlen(c, bam1_cigar(b)))
401 parse_error(fp->n_lines, "CIGAR and sequence length are inconsistent");
402 p = (uint8_t*)alloc_data(b, doff + c->l_qseq + (c->l_qseq+1)/2) + doff;
403 memset(p, 0, (c->l_qseq+1)/2);
404 for (i = 0; i < c->l_qseq; ++i)
405 p[i/2] |= bam_nt16_table[(int)str->s[i]] << 4*(1-i%2);
406 } else c->l_qseq = 0;
407 if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -6; // qual
409 if (strcmp(str->s, "*") && c->l_qseq != strlen(str->s))
410 parse_error(fp->n_lines, "sequence and quality are inconsistent");
411 p += (c->l_qseq+1)/2;
412 if (strcmp(str->s, "*") == 0) for (i = 0; i < c->l_qseq; ++i) p[i] = 0xff;
413 else for (i = 0; i < c->l_qseq; ++i) p[i] = str->s[i] - 33;
414 doff += c->l_qseq + (c->l_qseq+1)/2;
417 if (dret != '\n' && dret != '\r') { // aux
418 while (ks_getuntil(ks, KS_SEP_TAB, str, &dret) >= 0) {
419 uint8_t *s, type, key[2];
421 if (str->l < 6 || str->s[2] != ':' || str->s[4] != ':')
422 parse_error(fp->n_lines, "missing colon in auxiliary data");
423 key[0] = str->s[0]; key[1] = str->s[1];
425 s = alloc_data(b, doff + 3) + doff;
426 s[0] = key[0]; s[1] = key[1]; s += 2; doff += 2;
427 if (type == 'A' || type == 'a' || type == 'c' || type == 'C') { // c and C for backward compatibility
428 s = alloc_data(b, doff + 2) + doff;
429 *s++ = 'A'; *s = str->s[5];
431 } else if (type == 'I' || type == 'i') {
433 s = alloc_data(b, doff + 5) + doff;
434 x = (long long)atoll(str->s + 5);
437 *s++ = 'c'; *(int8_t*)s = (int8_t)x;
439 } else if (x >= -32767) {
440 *s++ = 's'; *(int16_t*)s = (int16_t)x;
443 *s++ = 'i'; *(int32_t*)s = (int32_t)x;
445 if (x < -2147483648ll)
446 fprintf(stderr, "Parse warning at line %lld: integer %lld is out of range.",
447 (long long)fp->n_lines, x);
451 *s++ = 'C'; *s++ = (uint8_t)x;
453 } else if (x <= 65535) {
454 *s++ = 'S'; *(uint16_t*)s = (uint16_t)x;
457 *s++ = 'I'; *(uint32_t*)s = (uint32_t)x;
459 if (x > 4294967295ll)
460 fprintf(stderr, "Parse warning at line %lld: integer %lld is out of range.",
461 (long long)fp->n_lines, x);
464 } else if (type == 'f') {
465 s = alloc_data(b, doff + 5) + doff;
467 *(float*)s = (float)atof(str->s + 5);
469 } else if (type == 'd') {
470 s = alloc_data(b, doff + 9) + doff;
472 *(float*)s = (float)atof(str->s + 9);
474 } else if (type == 'Z' || type == 'H') {
475 int size = 1 + (str->l - 5) + 1;
476 if (type == 'H') { // check whether the hex string is valid
478 if ((str->l - 5) % 2 == 1) parse_error(fp->n_lines, "length of the hex string not even");
479 for (i = 0; i < str->l - 5; ++i) {
480 int c = toupper(str->s[5 + i]);
481 if (!((c >= '0' && c <= '9') || (c >= 'A' && c <= 'F')))
482 parse_error(fp->n_lines, "invalid hex character");
485 s = alloc_data(b, doff + size) + doff;
487 memcpy(s, str->s + 5, str->l - 5);
490 } else parse_error(fp->n_lines, "unrecognized type");
491 if (dret == '\n' || dret == '\r') break;
494 b->l_aux = doff - doff0;
499 tamFile sam_open(const char *fn)
502 gzFile gzfp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "rb") : gzopen(fn, "rb");
503 if (gzfp == 0) return 0;
504 fp = (tamFile)calloc(1, sizeof(struct __tamFile_t));
505 fp->str = (kstring_t*)calloc(1, sizeof(kstring_t));
507 fp->ks = ks_init(fp->fp);
511 void sam_close(tamFile fp)
516 free(fp->str->s); free(fp->str);