]> git.donarmstrong.com Git - samtools.git/blob - sam.c
* samtools-0.1.5-12 (r415)
[samtools.git] / sam.c
1 #include <string.h>
2 #include "sam.h"
3
4 #define TYPE_BAM  1
5 #define TYPE_READ 2
6 #define TYPE_HEX  4
7
8 bam_header_t *bam_header_dup(const bam_header_t *h0)
9 {
10         bam_header_t *h;
11         int i;
12         h = bam_header_init();
13         *h = *h0;
14         h->hash = 0;
15         h->text = (char*)calloc(h->l_text + 1, 1);
16         memcpy(h->text, h0->text, h->l_text);
17         h->target_len = (uint32_t*)calloc(h->n_targets, 4);
18         h->target_name = (char**)calloc(h->n_targets, sizeof(void*));
19         for (i = 0; i < h->n_targets; ++i) {
20                 h->target_len[i] = h0->target_len[i];
21                 h->target_name[i] = strdup(h0->target_name[i]);
22         }
23         if (h0->rg2lib) h->rg2lib = bam_strmap_dup(h0->rg2lib);
24         return h;
25 }
26 static void append_header_text(bam_header_t *header, char* text, int len)
27 {
28         int x = header->l_text + 1;
29         int y = header->l_text + len + 1; // 1 byte null
30         if (text == 0) return;
31         kroundup32(x); 
32         kroundup32(y);
33         if (x < y) header->text = (char*)realloc(header->text, y);
34         strncpy(header->text + header->l_text, text, len); // we cannot use strcpy() here.
35         header->l_text += len;
36         header->text[header->l_text] = 0;
37 }
38
39 samfile_t *samopen(const char *fn, const char *mode, const void *aux)
40 {
41         samfile_t *fp;
42         fp = (samfile_t*)calloc(1, sizeof(samfile_t));
43         if (mode[0] == 'r') { // read
44                 fp->type |= TYPE_READ;
45                 if (mode[1] == 'b') { // binary
46                         fp->type |= TYPE_BAM;
47                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, "r") : bam_dopen(fileno(stdin), "r");
48                         if (fp->x.bam == 0) goto open_err_ret;
49                         fp->header = bam_header_read(fp->x.bam);
50                 } else { // text
51                         fp->x.tamr = sam_open(fn);
52                         if (fp->x.tamr == 0) goto open_err_ret;
53                         fp->header = sam_header_read(fp->x.tamr);
54                         if (fp->header->n_targets == 0) { // no @SQ fields
55                                 if (aux) { // check if aux is present
56                                         bam_header_t *textheader = fp->header;
57                                         fp->header = sam_header_read2((const char*)aux);
58                                         append_header_text(fp->header, textheader->text, textheader->l_text);
59                                         bam_header_destroy(textheader);
60                                 }
61                                 if (fp->header->n_targets == 0)
62                                         fprintf(stderr, "[samopen] no @SQ lines in the header.\n");
63                         } else fprintf(stderr, "[samopen] SAM header is present: %d sequences.\n", fp->header->n_targets);
64                 }
65                 sam_header_parse_rg(fp->header);
66         } else if (mode[0] == 'w') { // write
67                 fp->header = bam_header_dup((const bam_header_t*)aux);
68                 if (mode[1] == 'b') { // binary
69                         char bmode[3];
70                         bmode[0] = 'w'; bmode[1] = strstr(mode, "u")? 'u' : 0; bmode[2] = 0;
71                         fp->type |= TYPE_BAM;
72                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, bmode) : bam_dopen(fileno(stdout), bmode);
73                         if (fp->x.bam == 0) goto open_err_ret;
74                         bam_header_write(fp->x.bam, fp->header);
75                 } else { // text
76                         // open file
77                         fp->x.tamw = strcmp(fn, "-")? fopen(fn, "w") : stdout;
78                         if (fp->x.tamr == 0) goto open_err_ret;
79                         if (strstr(mode, "x")) fp->type |= TYPE_HEX;
80                         // write header
81                         if (strstr(mode, "h")) {
82                                 int i;
83                                 bam_header_t *alt;
84                                 // parse the header text 
85                                 alt = bam_header_init();
86                                 alt->l_text = fp->header->l_text; alt->text = fp->header->text;
87                                 sam_header_parse(alt);
88                                 alt->l_text = 0; alt->text = 0;
89                                 // check if there are @SQ lines in the header
90                                 fwrite(fp->header->text, 1, fp->header->l_text, fp->x.tamw);
91                                 if (alt->n_targets) { // then write the header text without dumping ->target_{name,len}
92                                         if (alt->n_targets != fp->header->n_targets)
93                                                 fprintf(stderr, "[samopen] inconsistent number of target sequences.\n");
94                                 } else { // then dump ->target_{name,len}
95                                         for (i = 0; i < fp->header->n_targets; ++i)
96                                                 fprintf(fp->x.tamw, "@SQ\tSN:%s\tLN:%d\n", fp->header->target_name[i], fp->header->target_len[i]);
97                                 }
98                                 bam_header_destroy(alt);
99                         }
100                 }
101         }
102         return fp;
103
104 open_err_ret:
105         free(fp);
106         return 0;
107 }
108
109 void samclose(samfile_t *fp)
110 {
111         if (fp == 0) return;
112         if (fp->header) bam_header_destroy(fp->header);
113         if (fp->type & TYPE_BAM) bam_close(fp->x.bam);
114         else if (fp->type & TYPE_READ) sam_close(fp->x.tamr);
115         else fclose(fp->x.tamw);
116         free(fp);
117 }
118
119 int samread(samfile_t *fp, bam1_t *b)
120 {
121         if (fp == 0 || !(fp->type & TYPE_READ)) return -1; // not open for reading
122         if (fp->type & TYPE_BAM) return bam_read1(fp->x.bam, b);
123         else return sam_read1(fp->x.tamr, fp->header, b);
124 }
125
126 int samwrite(samfile_t *fp, const bam1_t *b)
127 {
128         if (fp == 0 || (fp->type & TYPE_READ)) return -1; // not open for writing
129         if (fp->type & TYPE_BAM) return bam_write1(fp->x.bam, b);
130         else {
131                 char *s = bam_format1_core(fp->header, b, fp->type & TYPE_HEX);
132                 int l = strlen(s);
133                 fputs(s, fp->x.tamw); fputc('\n', fp->x.tamw);
134                 free(s);
135                 return l + 1;
136         }
137 }
138
139 int sampileup(samfile_t *fp, int mask, bam_pileup_f func, void *func_data)
140 {
141         bam_plbuf_t *buf;
142         int ret;
143         bam1_t *b;
144         b = bam_init1();
145         buf = bam_plbuf_init(func, func_data);
146         bam_plbuf_set_mask(buf, mask);
147         while ((ret = samread(fp, b)) >= 0)
148                 bam_plbuf_push(b, buf);
149         bam_plbuf_push(0, buf);
150         bam_plbuf_destroy(buf);
151         bam_destroy1(b);
152         return 0;
153 }