]> git.donarmstrong.com Git - samtools.git/blob - sam.c
* samtools-0.1.4-10 (r341)
[samtools.git] / sam.c
1 #include <string.h>
2 #include "sam.h"
3
4 #define TYPE_BAM  1
5 #define TYPE_READ 2
6
7 bam_header_t *bam_header_dup(const bam_header_t *h0)
8 {
9         bam_header_t *h;
10         int i;
11         h = bam_header_init();
12         *h = *h0;
13         h->hash = 0;
14         h->text = (char*)calloc(h->l_text + 1, 1);
15         memcpy(h->text, h0->text, h->l_text);
16         h->target_len = (uint32_t*)calloc(h->n_targets, 4);
17         h->target_name = (char**)calloc(h->n_targets, sizeof(void*));
18         for (i = 0; i < h->n_targets; ++i) {
19                 h->target_len[i] = h0->target_len[i];
20                 h->target_name[i] = strdup(h0->target_name[i]);
21         }
22         if (h0->rg2lib) h->rg2lib = bam_strmap_dup(h0->rg2lib);
23         return h;
24 }
25
26 samfile_t *samopen(const char *fn, const char *mode, const void *aux)
27 {
28         samfile_t *fp;
29         fp = (samfile_t*)calloc(1, sizeof(samfile_t));
30         if (mode[0] == 'r') { // read
31                 fp->type |= TYPE_READ;
32                 if (mode[1] == 'b') { // binary
33                         fp->type |= TYPE_BAM;
34                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, "r") : bam_dopen(fileno(stdin), "r");
35                         if (fp->x.bam == 0) goto open_err_ret;
36                         fp->header = bam_header_read(fp->x.bam);
37                 } else { // text
38                         fp->x.tamr = sam_open(fn);
39                         if (fp->x.tamr == 0) goto open_err_ret;
40                         fp->header = sam_header_read(fp->x.tamr);
41                         if (fp->header->n_targets == 0) { // no @SQ fields
42                                 if (aux) { // check if aux is present
43                                         bam_header_destroy(fp->header);
44                                         fp->header = sam_header_read2((const char*)aux);
45                                 }
46                                 if (fp->header->n_targets == 0)
47                                         fprintf(stderr, "[samopen] no @SQ lines in the header.\n");
48                         } else fprintf(stderr, "[samopen] SAM header is present: %d sequences.\n", fp->header->n_targets);
49                 }
50                 sam_header_parse_rg(fp->header);
51         } else if (mode[0] == 'w') { // write
52                 fp->header = bam_header_dup((const bam_header_t*)aux);
53                 if (mode[1] == 'b') { // binary
54                         char bmode[3];
55                         bmode[0] = 'w'; bmode[1] = strstr(mode, "u")? 'u' : 0; bmode[2] = 0;
56                         fp->type |= TYPE_BAM;
57                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, bmode) : bam_dopen(fileno(stdout), bmode);
58                         if (fp->x.bam == 0) goto open_err_ret;
59                         bam_header_write(fp->x.bam, fp->header);
60                 } else { // text
61                         // open file
62                         fp->x.tamw = strcmp(fn, "-")? fopen(fn, "w") : stdout;
63                         if (fp->x.tamr == 0) goto open_err_ret;
64                         // write header
65                         if (strstr(mode, "h")) {
66                                 int i;
67                                 bam_header_t *alt;
68                                 // parse the header text 
69                                 alt = bam_header_init();
70                                 alt->l_text = fp->header->l_text; alt->text = fp->header->text;
71                                 sam_header_parse(alt);
72                                 alt->l_text = 0; alt->text = 0;
73                                 // check if there are @SQ lines in the header
74                                 fwrite(fp->header->text, 1, fp->header->l_text, fp->x.tamw);
75                                 if (alt->n_targets) { // then write the header text without dumping ->target_{name,len}
76                                         if (alt->n_targets != fp->header->n_targets)
77                                                 fprintf(stderr, "[samopen] inconsistent number of target sequences.\n");
78                                 } else { // then dump ->target_{name,len}
79                                         for (i = 0; i < fp->header->n_targets; ++i)
80                                                 fprintf(fp->x.tamw, "@SQ\tSN:%s\tLN:%d\n", fp->header->target_name[i], fp->header->target_len[i]);
81                                 }
82                                 bam_header_destroy(alt);
83                         }
84                 }
85         }
86         return fp;
87
88 open_err_ret:
89         free(fp);
90         return 0;
91 }
92
93 void samclose(samfile_t *fp)
94 {
95         if (fp == 0) return;
96         if (fp->header) bam_header_destroy(fp->header);
97         if (fp->type & TYPE_BAM) bam_close(fp->x.bam);
98         else if (fp->type & TYPE_READ) sam_close(fp->x.tamr);
99         else fclose(fp->x.tamw);
100         free(fp);
101 }
102
103 int samread(samfile_t *fp, bam1_t *b)
104 {
105         if (fp == 0 || !(fp->type & TYPE_READ)) return -1; // not open for reading
106         if (fp->type & TYPE_BAM) return bam_read1(fp->x.bam, b);
107         else return sam_read1(fp->x.tamr, fp->header, b);
108 }
109
110 int samwrite(samfile_t *fp, const bam1_t *b)
111 {
112         if (fp == 0 || (fp->type & TYPE_READ)) return -1; // not open for writing
113         if (fp->type & TYPE_BAM) return bam_write1(fp->x.bam, b);
114         else {
115                 char *s = bam_format1(fp->header, b);
116                 int l = strlen(s);
117                 fputs(s, fp->x.tamw); fputc('\n', fp->x.tamw);
118                 free(s);
119                 return l + 1;
120         }
121 }
122
123 int sampileup(samfile_t *fp, int mask, bam_pileup_f func, void *func_data)
124 {
125         bam_plbuf_t *buf;
126         int ret;
127         bam1_t *b;
128         b = bam_init1();
129         buf = bam_plbuf_init(func, func_data);
130         bam_plbuf_set_mask(buf, mask);
131         while ((ret = samread(fp, b)) >= 0)
132                 bam_plbuf_push(b, buf);
133         bam_plbuf_push(0, buf);
134         bam_plbuf_destroy(buf);
135         bam_destroy1(b);
136         return 0;
137 }