Browse code

embed samtools sources directly, since Rsamtools has moved to htslib

Michael Lawrence authored on 11/02/2019 20:33:07
Showing1 changed files
1 1
new file mode 100644
... ...
@@ -0,0 +1,491 @@
1
+#define _POSIX_C_SOURCE 200112L /* Rsamtools: c99 fileno */
2
+#define _SVID_SOURCE            /* Rsamtools: c99 strdup */
3
+#include <zlib.h>
4
+#include <stdio.h>
5
+#include <ctype.h>
6
+#include <string.h>
7
+#include <stdlib.h>
8
+#include <unistd.h>
9
+#include <assert.h>
10
+#ifdef _WIN32
11
+#include <fcntl.h>
12
+#endif
13
+#include "kstring.h"
14
+#include "bam.h"
15
+#include "sam_header.h"
16
+#include "kseq.h"
17
+#include "khash.h"
18
+
19
+KSTREAM_INIT(gzFile, gzread, 16384)
20
+KHASH_MAP_INIT_STR(ref, uint64_t)
21
+
22
+void bam_init_header_hash(bam_header_t *header);
23
+void bam_destroy_header_hash(bam_header_t *header);
24
+int32_t bam_get_tid(const bam_header_t *header, const char *seq_name);
25
+
26
+unsigned char bam_nt16_table[256] = {
27
+	15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
28
+	15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
29
+	15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
30
+	 1, 2, 4, 8, 15,15,15,15, 15,15,15,15, 15, 0 /*=*/,15,15,
31
+	15, 1,14, 2, 13,15,15, 4, 11,15,15,12, 15, 3,15,15,
32
+	15,15, 5, 6,  8,15, 7, 9, 15,10,15,15, 15,15,15,15,
33
+	15, 1,14, 2, 13,15,15, 4, 11,15,15,12, 15, 3,15,15,
34
+	15,15, 5, 6,  8,15, 7, 9, 15,10,15,15, 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,
40
+	15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
41
+	15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
42
+	15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15
43
+};
44
+
45
+unsigned short bam_char2flag_table[256] = {
46
+	0,0,0,0, 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
+	0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
49
+	0,BAM_FREAD1,BAM_FREAD2,0, 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
+	BAM_FPROPER_PAIR,0,BAM_FMREVERSE,0, 0,BAM_FMUNMAP,0,0, 0,0,0,0, 0,0,0,0,
52
+	0,0,0,0, BAM_FDUP,0,BAM_FQCFAIL,0, 0,0,0,0, 0,0,0,0,
53
+	BAM_FPAIRED,0,BAM_FREVERSE,BAM_FSECONDARY, 0,BAM_FUNMAP,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,
59
+	0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
60
+	0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0,
61
+	0,0,0,0, 0,0,0,0, 0,0,0,0, 0,0,0,0
62
+};
63
+
64
+char *bam_nt16_rev_table = "=ACMGRSVTWYHKDBN";
65
+
66
+struct __tamFile_t {
67
+	gzFile fp;
68
+	kstream_t *ks;
69
+	kstring_t *str;
70
+	uint64_t n_lines;
71
+	int is_first;
72
+};
73
+
74
+char **__bam_get_lines(const char *fn, int *_n) // for bam_plcmd.c only
75
+{
76
+	char **list = 0, *s;
77
+	int n = 0, dret, m = 0;
78
+	gzFile fp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
79
+	kstream_t *ks;
80
+	kstring_t *str;
81
+	str = (kstring_t*)calloc(1, sizeof(kstring_t));
82
+	ks = ks_init(fp);
83
+	while (ks_getuntil(ks, '\n', str, &dret) > 0) {
84
+		if (n == m) {
85
+			m = m? m << 1 : 16;
86
+			list = (char**)realloc(list, m * sizeof(char*));
87
+		}
88
+		if (str->s[str->l-1] == '\r')
89
+			str->s[--str->l] = '\0';
90
+		s = list[n++] = (char*)calloc(str->l + 1, 1);
91
+		strcpy(s, str->s);
92
+	}
93
+	ks_destroy(ks);
94
+	gzclose(fp);
95
+	free(str->s); free(str);
96
+	*_n = n;
97
+	return list;
98
+}
99
+
100
+static bam_header_t *hash2header(const kh_ref_t *hash)
101
+{
102
+	bam_header_t *header;
103
+	khiter_t k;
104
+	header = bam_header_init();
105
+	header->n_targets = kh_size(hash);
106
+	header->target_name = (char**)calloc(kh_size(hash), sizeof(char*));
107
+	header->target_len = (uint32_t*)calloc(kh_size(hash), 4);
108
+	for (k = kh_begin(hash); k != kh_end(hash); ++k) {
109
+		if (kh_exist(hash, k)) {
110
+			int i = (int)kh_value(hash, k);
111
+			header->target_name[i] = (char*)kh_key(hash, k);
112
+			header->target_len[i] = kh_value(hash, k)>>32;
113
+		}
114
+	}
115
+	bam_init_header_hash(header);
116
+	return header;
117
+}
118
+bam_header_t *sam_header_read2(const char *fn)
119
+{
120
+	bam_header_t *header;
121
+	int c, dret, ret, error = 0;
122
+	gzFile fp;
123
+	kstream_t *ks;
124
+	kstring_t *str;
125
+	kh_ref_t *hash;
126
+	khiter_t k;
127
+	if (fn == 0) return 0;
128
+	fp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "r") : gzopen(fn, "r");
129
+	if (fp == 0) return 0;
130
+	hash = kh_init(ref);
131
+	ks = ks_init(fp);
132
+	str = (kstring_t*)calloc(1, sizeof(kstring_t));
133
+	while (ks_getuntil(ks, 0, str, &dret) > 0) {
134
+		char *s = strdup(str->s);
135
+		int len, i;
136
+		i = kh_size(hash);
137
+		ks_getuntil(ks, 0, str, &dret);
138
+		len = atoi(str->s);
139
+		k = kh_put(ref, hash, s, &ret);
140
+		if (ret == 0) {
141
+			fprintf(stderr, "[sam_header_read2] duplicated sequence name: %s\n", s);
142
+			error = 1;
143
+		}
144
+		kh_value(hash, k) = (uint64_t)len<<32 | i;
145
+		if (dret != '\n')
146
+			while ((c = ks_getc(ks)) != '\n' && c != -1);
147
+	}
148
+	ks_destroy(ks);
149
+	gzclose(fp);
150
+	free(str->s); free(str);
151
+	fprintf(stderr, "[sam_header_read2] %d sequences loaded.\n", kh_size(hash));
152
+	if (error) return 0;
153
+	header = hash2header(hash);
154
+	kh_destroy(ref, hash);
155
+	return header;
156
+}
157
+static inline uint8_t *alloc_data(bam1_t *b, int size)
158
+{
159
+	if (b->m_data < size) {
160
+		b->m_data = size;
161
+		kroundup32(b->m_data);
162
+		b->data = (uint8_t*)realloc(b->data, b->m_data);
163
+	}
164
+	return b->data;
165
+}
166
+static inline void parse_error(int64_t n_lines, const char * __restrict msg)
167
+{
168
+	fprintf(stderr, "Parse error at line %lld: %s\n", (long long)n_lines, msg);
169
+	abort();
170
+}
171
+static inline void append_text(bam_header_t *header, kstring_t *str)
172
+{
173
+	size_t x = header->l_text, y = header->l_text + str->l + 2; // 2 = 1 byte dret + 1 byte null
174
+	kroundup32(x); kroundup32(y);
175
+	if (x < y) 
176
+    {
177
+        header->n_text = y;
178
+        header->text = (char*)realloc(header->text, y);
179
+        if ( !header->text ) 
180
+        {
181
+            fprintf(stderr,"realloc failed to alloc %ld bytes\n", y);
182
+            abort();
183
+        }
184
+    }
185
+    // Sanity check
186
+    if ( header->l_text+str->l+1 >= header->n_text )
187
+    {
188
+        fprintf(stderr,"append_text FIXME: %ld>=%ld, x=%ld,y=%ld\n",  header->l_text+str->l+1,(long)header->n_text,x,y);
189
+        abort();
190
+    }
191
+	strncpy(header->text + header->l_text, str->s, str->l+1); // we cannot use strcpy() here.
192
+	header->l_text += str->l + 1;
193
+	header->text[header->l_text] = 0;
194
+}
195
+
196
+int sam_header_parse(bam_header_t *h)
197
+{
198
+	char **tmp;
199
+	int i;
200
+	free(h->target_len); free(h->target_name);
201
+	h->n_targets = 0; h->target_len = 0; h->target_name = 0;
202
+	if (h->l_text < 3) return 0;
203
+	if (h->dict == 0) h->dict = sam_header_parse2(h->text);
204
+	tmp = sam_header2list(h->dict, "SQ", "SN", &h->n_targets);
205
+	if (h->n_targets == 0) return 0;
206
+	h->target_name = calloc(h->n_targets, sizeof(void*));
207
+	for (i = 0; i < h->n_targets; ++i)
208
+		h->target_name[i] = strdup(tmp[i]);
209
+	free(tmp);
210
+	tmp = sam_header2list(h->dict, "SQ", "LN", &h->n_targets);
211
+	h->target_len = calloc(h->n_targets, 4);
212
+	for (i = 0; i < h->n_targets; ++i)
213
+		h->target_len[i] = atoi(tmp[i]);
214
+	free(tmp);
215
+	return h->n_targets;
216
+}
217
+
218
+bam_header_t *sam_header_read(tamFile fp)
219
+{
220
+	int ret, dret;
221
+	bam_header_t *header = bam_header_init();
222
+	kstring_t *str = fp->str;
223
+	while ((ret = ks_getuntil(fp->ks, KS_SEP_TAB, str, &dret)) >= 0 && str->s[0] == '@') { // skip header
224
+		str->s[str->l] = dret; // note that str->s is NOT null terminated!!
225
+		append_text(header, str);
226
+		if (dret != '\n') {
227
+			ret = ks_getuntil(fp->ks, '\n', str, &dret);
228
+			str->s[str->l] = '\n'; // NOT null terminated!!
229
+			append_text(header, str);
230
+		}
231
+		++fp->n_lines;
232
+	}
233
+	sam_header_parse(header);
234
+	bam_init_header_hash(header);
235
+	fp->is_first = ret >= 0; /* MTM: not header-only SAM */
236
+	return header;
237
+}
238
+
239
+int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b)
240
+{
241
+	int ret, doff, doff0, dret, z = 0;
242
+	bam1_core_t *c = &b->core;
243
+	kstring_t *str = fp->str;
244
+	kstream_t *ks = fp->ks;
245
+
246
+	if (fp->is_first) {
247
+		fp->is_first = 0;
248
+		ret = str->l;
249
+	} else {
250
+		do { // special consideration for empty lines
251
+			ret = ks_getuntil(fp->ks, KS_SEP_TAB, str, &dret);
252
+			if (ret >= 0) z += str->l + 1;
253
+		} while (ret == 0);
254
+	}
255
+	if (ret < 0) return -1;
256
+	++fp->n_lines;
257
+	doff = 0;
258
+
259
+	{ // name
260
+		c->l_qname = strlen(str->s) + 1;
261
+		memcpy(alloc_data(b, doff + c->l_qname) + doff, str->s, c->l_qname);
262
+		doff += c->l_qname;
263
+	}
264
+	{ // flag
265
+		long flag;
266
+		char *s;
267
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
268
+		flag = strtol((char*)str->s, &s, 0);
269
+		if (*s) { // not the end of the string
270
+			flag = 0;
271
+			for (s = str->s; *s; ++s)
272
+				flag |= bam_char2flag_table[(int)*s];
273
+		}
274
+		c->flag = flag;
275
+	}
276
+	{ // tid, pos, qual
277
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->tid = bam_get_tid(header, str->s);
278
+		if (c->tid < 0 && strcmp(str->s, "*")) {
279
+			if (header->n_targets == 0) {
280
+				fprintf(stderr, "[sam_read1] missing header? Abort!\n");
281
+				exit(1);
282
+			} else fprintf(stderr, "[sam_read1] reference '%s' is recognized as '*'.\n", str->s);
283
+		}
284
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->pos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
285
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1; c->qual = isdigit(str->s[0])? atoi(str->s) : 0;
286
+		if (ret < 0) return -2;
287
+	}
288
+	{ // cigar
289
+		char *s, *t;
290
+		int i, op;
291
+		long x;
292
+		c->n_cigar = 0;
293
+		if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -3;
294
+		z += str->l + 1;
295
+		if (str->s[0] != '*') {
296
+			uint32_t *cigar;
297
+			for (s = str->s; *s; ++s) {
298
+				if ((isalpha(*s)) || (*s=='=')) ++c->n_cigar;
299
+				else if (!isdigit(*s)) parse_error(fp->n_lines, "invalid CIGAR character");
300
+			}
301
+			b->data = alloc_data(b, doff + c->n_cigar * 4);
302
+			cigar = bam1_cigar(b);
303
+			for (i = 0, s = str->s; i != c->n_cigar; ++i) {
304
+				x = strtol(s, &t, 10);
305
+				op = toupper(*t);
306
+				if (op == 'M') op = BAM_CMATCH;
307
+				else if (op == 'I') op = BAM_CINS;
308
+				else if (op == 'D') op = BAM_CDEL;
309
+				else if (op == 'N') op = BAM_CREF_SKIP;
310
+				else if (op == 'S') op = BAM_CSOFT_CLIP;
311
+				else if (op == 'H') op = BAM_CHARD_CLIP;
312
+				else if (op == 'P') op = BAM_CPAD;
313
+				else if (op == '=') op = BAM_CEQUAL;
314
+				else if (op == 'X') op = BAM_CDIFF;
315
+				else if (op == 'B') op = BAM_CBACK;
316
+				else parse_error(fp->n_lines, "invalid CIGAR operation");
317
+				s = t + 1;
318
+				cigar[i] = bam_cigar_gen(x, op);
319
+			}
320
+			if (*s) parse_error(fp->n_lines, "unmatched CIGAR operation");
321
+			c->bin = bam_reg2bin(c->pos, bam_calend(c, cigar));
322
+			doff += c->n_cigar * 4;
323
+		} else {
324
+			if (!(c->flag&BAM_FUNMAP)) {
325
+				fprintf(stderr, "Parse warning at line %lld: mapped sequence without CIGAR\n", (long long)fp->n_lines);
326
+				c->flag |= BAM_FUNMAP;
327
+			}
328
+			c->bin = bam_reg2bin(c->pos, c->pos + 1);
329
+		}
330
+	}
331
+	{ // mtid, mpos, isize
332
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
333
+		c->mtid = strcmp(str->s, "=")? bam_get_tid(header, str->s) : c->tid;
334
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
335
+		c->mpos = isdigit(str->s[0])? atoi(str->s) - 1 : -1;
336
+		ret = ks_getuntil(ks, KS_SEP_TAB, str, &dret); z += str->l + 1;
337
+		c->isize = (str->s[0] == '-' || isdigit(str->s[0]))? atoi(str->s) : 0;
338
+		if (ret < 0) return -4;
339
+	}
340
+	{ // seq and qual
341
+		int i;
342
+		uint8_t *p = 0;
343
+		if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -5; // seq
344
+		z += str->l + 1;
345
+		if (strcmp(str->s, "*")) {
346
+			c->l_qseq = strlen(str->s);
347
+			if (c->n_cigar && c->l_qseq != (int32_t)bam_cigar2qlen(c, bam1_cigar(b))) {
348
+				fprintf(stderr, "Line %ld, sequence length %i vs %i from CIGAR\n",
349
+						(long)fp->n_lines, c->l_qseq, (int32_t)bam_cigar2qlen(c, bam1_cigar(b)));
350
+				parse_error(fp->n_lines, "CIGAR and sequence length are inconsistent");
351
+			}
352
+			p = (uint8_t*)alloc_data(b, doff + c->l_qseq + (c->l_qseq+1)/2) + doff;
353
+			memset(p, 0, (c->l_qseq+1)/2);
354
+			for (i = 0; i < c->l_qseq; ++i)
355
+				p[i/2] |= bam_nt16_table[(int)str->s[i]] << 4*(1-i%2);
356
+		} else c->l_qseq = 0;
357
+		if (ks_getuntil(ks, KS_SEP_TAB, str, &dret) < 0) return -6; // qual
358
+		z += str->l + 1;
359
+		if (strcmp(str->s, "*") && c->l_qseq != strlen(str->s))
360
+			parse_error(fp->n_lines, "sequence and quality are inconsistent");
361
+		p += (c->l_qseq+1)/2;
362
+		if (strcmp(str->s, "*") == 0) for (i = 0; i < c->l_qseq; ++i) p[i] = 0xff;
363
+		else for (i = 0; i < c->l_qseq; ++i) p[i] = str->s[i] - 33;
364
+		doff += c->l_qseq + (c->l_qseq+1)/2;
365
+	}
366
+	doff0 = doff;
367
+	if (dret != '\n' && dret != '\r') { // aux
368
+		while (ks_getuntil(ks, KS_SEP_TAB, str, &dret) >= 0) {
369
+			uint8_t *s, type, key[2];
370
+			z += str->l + 1;
371
+			if (str->l < 6 || str->s[2] != ':' || str->s[4] != ':')
372
+				parse_error(fp->n_lines, "missing colon in auxiliary data");
373
+			key[0] = str->s[0]; key[1] = str->s[1];
374
+			type = str->s[3];
375
+			s = alloc_data(b, doff + 3) + doff;
376
+			s[0] = key[0]; s[1] = key[1]; s += 2; doff += 2;
377
+			if (type == 'A' || type == 'a' || type == 'c' || type == 'C') { // c and C for backward compatibility
378
+				s = alloc_data(b, doff + 2) + doff;
379
+				*s++ = 'A'; *s = str->s[5];
380
+				doff += 2;
381
+			} else if (type == 'I' || type == 'i') {
382
+				long long x;
383
+				s = alloc_data(b, doff + 5) + doff;
384
+				x = (long long)atoll(str->s + 5);
385
+				if (x < 0) {
386
+					if (x >= -127) {
387
+						*s++ = 'c'; *(int8_t*)s = (int8_t)x;
388
+						s += 1; doff += 2;
389
+					} else if (x >= -32767) {
390
+						*s++ = 's'; *(int16_t*)s = (int16_t)x;
391
+						s += 2; doff += 3;
392
+					} else {
393
+						*s++ = 'i'; *(int32_t*)s = (int32_t)x;
394
+						s += 4; doff += 5;
395
+						if (x < -2147483648ll)
396
+							fprintf(stderr, "Parse warning at line %lld: integer %lld is out of range.",
397
+									(long long)fp->n_lines, x);
398
+					}
399
+				} else {
400
+					if (x <= 255) {
401
+						*s++ = 'C'; *s++ = (uint8_t)x;
402
+						doff += 2;
403
+					} else if (x <= 65535) {
404
+						*s++ = 'S'; *(uint16_t*)s = (uint16_t)x;
405
+						s += 2; doff += 3;
406
+					} else {
407
+						*s++ = 'I'; *(uint32_t*)s = (uint32_t)x;
408
+						s += 4; doff += 5;
409
+						if (x > 4294967295ll)
410
+							fprintf(stderr, "Parse warning at line %lld: integer %lld is out of range.",
411
+									(long long)fp->n_lines, x);
412
+					}
413
+				}
414
+			} else if (type == 'f') {
415
+				s = alloc_data(b, doff + 5) + doff;
416
+				*s++ = 'f';
417
+				*(float*)s = (float)atof(str->s + 5);
418
+				s += 4; doff += 5;
419
+			} else if (type == 'd') {
420
+				s = alloc_data(b, doff + 9) + doff;
421
+				*s++ = 'd';
422
+				*(float*)s = (float)atof(str->s + 9);
423
+				s += 8; doff += 9;
424
+			} else if (type == 'Z' || type == 'H') {
425
+				int size = 1 + (str->l - 5) + 1;
426
+				if (type == 'H') { // check whether the hex string is valid
427
+					int i;
428
+					if ((str->l - 5) % 2 == 1) parse_error(fp->n_lines, "length of the hex string not even");
429
+					for (i = 0; i < str->l - 5; ++i) {
430
+						int c = toupper(str->s[5 + i]);
431
+						if (!((c >= '0' && c <= '9') || (c >= 'A' && c <= 'F')))
432
+							parse_error(fp->n_lines, "invalid hex character");
433
+					}
434
+				}
435
+				s = alloc_data(b, doff + size) + doff;
436
+				*s++ = type;
437
+				memcpy(s, str->s + 5, str->l - 5);
438
+				s[str->l - 5] = 0;
439
+				doff += size;
440
+			} else if (type == 'B') {
441
+				int32_t n = 0, Bsize, k = 0, size;
442
+				char *p;
443
+				if (str->l < 8) parse_error(fp->n_lines, "too few values in aux type B");
444
+				Bsize = bam_aux_type2size(str->s[5]); // the size of each element
445
+				for (p = (char*)str->s + 6; *p; ++p) // count the number of elements in the array
446
+					if (*p == ',') ++n;
447
+				p = str->s + 7; // now p points to the first number in the array
448
+				size = 6 + Bsize * n; // total number of bytes allocated to this tag
449
+				s = alloc_data(b, doff + 6 * Bsize * n) + doff; // allocate memory
450
+				*s++ = 'B'; *s++ = str->s[5];
451
+				memcpy(s, &n, 4); s += 4; // write the number of elements
452
+				if (str->s[5] == 'c')      while (p < str->s + str->l) ((int8_t*)s)[k++]   = (int8_t)strtol(p, &p, 0),   ++p;
453
+				else if (str->s[5] == 'C') while (p < str->s + str->l) ((uint8_t*)s)[k++]  = (uint8_t)strtol(p, &p, 0),  ++p;
454
+				else if (str->s[5] == 's') while (p < str->s + str->l) ((int16_t*)s)[k++]  = (int16_t)strtol(p, &p, 0),  ++p; // FIXME: avoid unaligned memory
455
+				else if (str->s[5] == 'S') while (p < str->s + str->l) ((uint16_t*)s)[k++] = (uint16_t)strtol(p, &p, 0), ++p;
456
+				else if (str->s[5] == 'i') while (p < str->s + str->l) ((int32_t*)s)[k++]  = (int32_t)strtol(p, &p, 0),  ++p;
457
+				else if (str->s[5] == 'I') while (p < str->s + str->l) ((uint32_t*)s)[k++] = (uint32_t)strtol(p, &p, 0), ++p;
458
+				else if (str->s[5] == 'f') while (p < str->s + str->l) ((float*)s)[k++]    = (float)strtod(p, &p),       ++p;
459
+				else parse_error(fp->n_lines, "unrecognized array type");
460
+				s += Bsize * n; doff += size;
461
+			} else parse_error(fp->n_lines, "unrecognized type");
462
+			if (dret == '\n' || dret == '\r') break;
463
+		}
464
+	}
465
+	b->l_aux = doff - doff0;
466
+	b->data_len = doff;
467
+	if (bam_no_B) bam_remove_B(b);
468
+	return z;
469
+}
470
+
471
+tamFile sam_open(const char *fn)
472
+{
473
+	tamFile fp;
474
+	gzFile gzfp = (strcmp(fn, "-") == 0)? gzdopen(fileno(stdin), "rb") : gzopen(fn, "rb");
475
+	if (gzfp == 0) return 0;
476
+	fp = (tamFile)calloc(1, sizeof(struct __tamFile_t));
477
+	fp->str = (kstring_t*)calloc(1, sizeof(kstring_t));
478
+	fp->fp = gzfp;
479
+	fp->ks = ks_init(fp->fp);
480
+	return fp;
481
+}
482
+
483
+void sam_close(tamFile fp)
484
+{
485
+	if (fp) {
486
+		ks_destroy(fp->ks);
487
+		gzclose(fp->fp);
488
+		free(fp->str->s); free(fp->str);
489
+		free(fp);
490
+	}
491
+}