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;
121 if (fn == 0) return 0;
122 fp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
123 if (fp == 0) return 0;
126 str = (kstring_t*)calloc(1, sizeof(kstring_t));
127 while (ks_getuntil(ks, 0, str, &dret) > 0) {
128 char *s = strdup(str->s);
131 ks_getuntil(ks, 0, str, &dret);
133 k = kh_put(ref, hash, s, &ret);
134 kh_value(hash, k) = (uint64_t)len<<32 | i;
136 while ((c = ks_getc(ks)) != '\n' && c != -1);
140 free(str->s); free(str);
141 fprintf(stderr, "[sam_header_read2] %d sequences loaded.\n", kh_size(hash));
142 header = hash2header(hash);
143 kh_destroy(ref, hash);
146 static inline uint8_t *alloc_data(bam1_t *b, int size)
148 if (b->m_data < size) {
150 kroundup32(b->m_data);
151 b->data = (uint8_t*)realloc(b->data, b->m_data);
155 static inline void parse_error(int64_t n_lines, const char * __restrict msg)
157 fprintf(stderr, "Parse error at line %lld: %s\n", (long long)n_lines, msg);
160 static inline void append_text(bam_header_t *header, kstring_t *str)
162 int x = header->l_text, y = header->l_text + str->l + 2; // 2 = 1 byte dret + 1 byte null
163 kroundup32(x); kroundup32(y);
164 if (x < y) header->text = (char*)realloc(header->text, y);
165 strncpy(header->text + header->l_text, str->s, str->l+1); // we cannot use strcpy() here.
166 header->l_text += str->l + 1;
167 header->text[header->l_text] = 0;
170 int sam_header_parse_rg(bam_header_t *h)
172 kstring_t *rgid, *rglib;
177 if (h == 0) return 0;
178 bam_strmap_destroy(h->rg2lib); h->rg2lib = 0;
179 if (h->l_text < 3) return 0;
181 h->rg2lib = bam_strmap_init();
182 rgid = calloc(1, sizeof(kstring_t));
183 rglib = calloc(1, sizeof(kstring_t));
185 while ((s = strstr(s, "@RG")) != 0) {
186 if (rgid->l && rglib->l) {
187 bam_strmap_put(h->rg2lib, rgid->s, rglib->s);
190 rgid->l = rglib->l = 0;
193 if ((p = strstr(s, "ID:")) != 0) {
195 for (p = q; *p && *p != '\t' && *p != '\r' && *p != '\n'; ++p);
196 kputsn(q, p - q, rgid);
198 fprintf(stderr, "[bam_header_parse] missing ID tag in @RG lines.\n");
202 if ((p = strstr(s, "LB:")) != 0) {
204 for (p = q; *p && *p != '\t' && *p != '\r' && *p != '\n'; ++p);
205 kputsn(q, p - q, rglib);
207 fprintf(stderr, "[bam_header_parse] missing LB tag in @RG lines.\n");
213 if (rgid->l && rglib->l) {
214 bam_strmap_put(h->rg2lib, rgid->s, rglib->s);
217 free(rgid->s); free(rgid);
218 free(rglib->s); free(rglib);
220 bam_strmap_destroy(h->rg2lib);
226 int sam_header_parse(bam_header_t *h)
232 free(h->target_len); free(h->target_name);
233 h->n_targets = 0; h->target_len = 0; h->target_name = 0;
234 if (h->l_text < 3) return 0;
235 // count number of @SQ
237 while ((s = strstr(s, "@SQ")) != 0) {
241 if (h->n_targets == 0) return 0;
242 h->target_len = (uint32_t*)calloc(h->n_targets, 4);
243 h->target_name = (char**)calloc(h->n_targets, sizeof(void*));
247 while ((s = strstr(s, "@SQ")) != 0) {
250 if ((p = strstr(s, "SN:")) != 0) {
252 for (p = q; *p && *p != '\t' && *p != '\r' && *p != '\n'; ++p);
253 h->target_name[i] = (char*)calloc(p - q + 1, 1);
254 strncpy(h->target_name[i], q, p - q);
255 } else goto header_err_ret;
257 if ((p = strstr(s, "LN:")) != 0) h->target_len[i] = strtol(p + 3, 0, 10);
258 else goto header_err_ret;
263 sam_header_parse_rg(h);
267 fprintf(stderr, "[bam_header_parse] missing SN or LN tag in @SQ lines.\n");
268 free(h->target_len); free(h->target_name);
269 h->n_targets = 0; h->target_len = 0; h->target_name = 0;
273 bam_header_t *sam_header_read(tamFile fp)
276 bam_header_t *header = bam_header_init();
277 kstring_t *str = fp->str;
278 while ((ret = ks_getuntil(fp->ks, KS_SEP_TAB, str, &dret)) >= 0 && str->s[0] == '@') { // skip header
279 str->s[str->l] = dret; // note that str->s is NOT null terminated!!
280 append_text(header, str);
282 ret = ks_getuntil(fp->ks, '\n', str, &dret);
283 str->s[str->l] = '\n'; // NOT null terminated!!
284 append_text(header, str);
288 sam_header_parse(header);
289 bam_init_header_hash(header);
294 int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
296 int ret, doff, doff0, dret, z = 0;
297 bam1_core_t *c = &b->core;
298 kstring_t *str = fp->str;
299 kstream_t *ks = fp->ks;
305 do { // special consideration for empty lines
306 ret = ks_getuntil(fp->ks, KS_SEP_TAB, str, &dret);
307 if (ret >= 0) z += str->l + 1;
310 if (ret < 0) return -1;
315 c->l_qname = strlen(str->s) + 1;
316 memcpy(alloc_data(b, doff + c->l_qname) + doff, str->s, c->l_qname);
322 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
323 flag = strtol((char*)str->s, &s, 0);
324 if (*s) { // not the end of the string
326 for (s = str->s; *s; ++s)
327 flag |= bam_char2flag_table[(int)*s];
332 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->tid = bam_get_tid(header, str->s);
333 if (c->tid < 0 && strcmp(str->s, "*")) {
334 if (header->n_targets == 0) {
335 fprintf(stderr, "[sam_read1] missing header? Abort!\n");
337 } else fprintf(stderr, "[sam_read1] reference '%s' is recognized as '*'.\n", str->s);
339 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->pos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
340 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->qual = isdigit(str->s[0])? atoi(str->s) : 0;
341 if (ret < 0) return -2;
348 if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -3;
350 if (str->s[0] != '*') {
351 for (s = str->s; *s; ++s) {
352 if (isalpha(*s)) ++c->n_cigar;
353 else if (!isdigit(*s)) parse_error(fp->n_lines, "invalid CIGAR character");
355 b->data = alloc_data(b, doff + c->n_cigar * 4);
356 for (i = 0, s = str->s; i != c->n_cigar; ++i) {
357 x = strtol(s, &t, 10);
359 if (op == 'M') op = BAM_CMATCH;
360 else if (op == 'I') op = BAM_CINS;
361 else if (op == 'D') op = BAM_CDEL;
362 else if (op == 'N') op = BAM_CREF_SKIP;
363 else if (op == 'S') op = BAM_CSOFT_CLIP;
364 else if (op == 'H') op = BAM_CHARD_CLIP;
365 else if (op == 'P') op = BAM_CPAD;
366 else parse_error(fp->n_lines, "invalid CIGAR operation");
368 bam1_cigar(b)[i] = x << BAM_CIGAR_SHIFT | op;
370 if (*s) parse_error(fp->n_lines, "unmatched CIGAR operation");
371 c->bin = bam_reg2bin(c->pos, bam_calend(c, bam1_cigar(b)));
372 doff += c->n_cigar * 4;
373 } else c->bin = bam_reg2bin(c->pos, c->pos + 1);
375 { // mtid, mpos, isize
376 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
377 c->mtid = strcmp(str->s, "=")? bam_get_tid(header, str->s) : c->tid;
378 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
379 c->mpos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
380 ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
381 c->isize = (str->s[0] == '-' || isdigit(str->s[0]))? atoi(str->s) : 0;
382 if (ret < 0) return -4;
387 if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -5; // seq
389 c->l_qseq = strlen(str->s);
390 if (c->n_cigar && c->l_qseq != (int32_t)bam_cigar2qlen(c, bam1_cigar(b)))
391 parse_error(fp->n_lines, "CIGAR and sequence length are inconsistent");
392 p = (uint8_t*)alloc_data(b, doff + c->l_qseq + (c->l_qseq+1)/2) + doff;
393 memset(p, 0, (c->l_qseq+1)/2);
394 for (i = 0; i < c->l_qseq; ++i)
395 p[i/2] |= bam_nt16_table[(int)str->s[i]] << 4*(1-i%2);
396 if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -6; // qual
398 if (strcmp(str->s, "*") && c->l_qseq != strlen(str->s))
399 parse_error(fp->n_lines, "sequence and quality are inconsistent");
400 p += (c->l_qseq+1)/2;
401 if (strcmp(str->s, "*") == 0) for (i = 0; i < c->l_qseq; ++i) p[i] = 0xff;
402 else for (i = 0; i < c->l_qseq; ++i) p[i] = str->s[i] - 33;
403 doff += c->l_qseq + (c->l_qseq+1)/2;
406 if (dret != '\n' && dret != '\r') { // aux
407 while (ks_getuntil(ks, KS_SEP_TAB, str, &dret) >= 0) {
408 uint8_t *s, type, key[2];
410 if (str->l < 6 || str->s[2] != ':' || str->s[4] != ':')
411 parse_error(fp->n_lines, "missing colon in auxiliary data");
412 key[0] = str->s[0]; key[1] = str->s[1];
414 s = alloc_data(b, doff + 3) + doff;
415 s[0] = key[0]; s[1] = key[1]; s += 2; doff += 2;
416 if (type == 'A' || type == 'a' || type == 'c' || type == 'C') { // c and C for backward compatibility
417 s = alloc_data(b, doff + 2) + doff;
418 *s++ = 'A'; *s = str->s[5];
420 } else if (type == 'I' || type == 'i') {
422 s = alloc_data(b, doff + 5) + doff;
423 x = (long long)atoll(str->s + 5);
426 *s++ = 'c'; *(int8_t*)s = (int8_t)x;
428 } else if (x >= -32767) {
429 *s++ = 's'; *(int16_t*)s = (int16_t)x;
432 *s++ = 'i'; *(int32_t*)s = (int32_t)x;
434 if (x < -2147483648ll)
435 fprintf(stderr, "Parse warning at line %lld: integer %lld is out of range.",
436 (long long)fp->n_lines, x);
440 *s++ = 'C'; *s++ = (uint8_t)x;
442 } else if (x <= 65535) {
443 *s++ = 'S'; *(uint16_t*)s = (uint16_t)x;
446 *s++ = 'I'; *(uint32_t*)s = (uint32_t)x;
448 if (x > 4294967295ll)
449 fprintf(stderr, "Parse warning at line %lld: integer %lld is out of range.",
450 (long long)fp->n_lines, x);
453 } else if (type == 'f') {
454 s = alloc_data(b, doff + 5) + doff;
456 *(float*)s = (float)atof(str->s + 5);
458 } else if (type == 'd') {
459 s = alloc_data(b, doff + 9) + doff;
461 *(float*)s = (float)atof(str->s + 9);
463 } else if (type == 'Z' || type == 'H') {
464 int size = 1 + (str->l - 5) + 1;
465 if (type == 'H') { // check whether the hex string is valid
467 if ((str->l - 5) % 2 == 1) parse_error(fp->n_lines, "length of the hex string not even");
468 for (i = 0; i < str->l - 5; ++i) {
469 int c = toupper(str->s[5 + i]);
470 if (!((c >= '0' && c <= '9') || (c >= 'A' && c <= 'F')))
471 parse_error(fp->n_lines, "invalid hex character");
474 s = alloc_data(b, doff + size) + doff;
476 memcpy(s, str->s + 5, str->l - 5);
479 } else parse_error(fp->n_lines, "unrecognized type");
480 if (dret == '\n' || dret == '\r') break;
483 b->l_aux = doff - doff0;
488 tamFile sam_open(const char *fn)
491 gzFile gzfp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
492 if (gzfp == 0) return 0;
493 fp = (tamFile)calloc(1, sizeof(struct __tamFile_t));
494 fp->str = (kstring_t*)calloc(1, sizeof(kstring_t));
496 fp->ks = ks_init(fp->fp);
500 void sam_close(tamFile fp)
505 free(fp->str->s); free(fp->str);