]> git.donarmstrong.com Git - samtools.git/blob - sam.c
* samtools-0.1.4-23 (r372)
[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 static void append_header_text(bam_header_t *header, char* text, int len)
26 {
27         int x = header->l_text + 1;
28         int y = header->l_text + len + 1; // 1 byte null
29         kroundup32(x); 
30         kroundup32(y);
31         if (x < y) header->text = (char*)realloc(header->text, y);
32         strncpy(header->text + header->l_text, text, len); // we cannot use strcpy() here.
33         header->l_text += len;
34         header->text[header->l_text] = 0;
35 }
36
37 samfile_t *samopen(const char *fn, const char *mode, const void *aux)
38 {
39         samfile_t *fp;
40         fp = (samfile_t*)calloc(1, sizeof(samfile_t));
41         if (mode[0] == 'r') { // read
42                 fp->type |= TYPE_READ;
43                 if (mode[1] == 'b') { // binary
44                         fp->type |= TYPE_BAM;
45                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, "r") : bam_dopen(fileno(stdin), "r");
46                         if (fp->x.bam == 0) goto open_err_ret;
47                         fp->header = bam_header_read(fp->x.bam);
48                 } else { // text
49                         fp->x.tamr = sam_open(fn);
50                         if (fp->x.tamr == 0) goto open_err_ret;
51                         fp->header = sam_header_read(fp->x.tamr);
52                         if (fp->header->n_targets == 0) { // no @SQ fields
53                                 if (aux) { // check if aux is present
54                                         bam_header_t *textheader = fp->header;
55                                         fp->header = sam_header_read2((const char*)aux);
56                                         append_header_text(fp->header, textheader->text, textheader->l_text);
57                                         bam_header_destroy(textheader);
58                                 }
59                                 if (fp->header->n_targets == 0)
60                                         fprintf(stderr, "[samopen] no @SQ lines in the header.\n");
61                         } else fprintf(stderr, "[samopen] SAM header is present: %d sequences.\n", fp->header->n_targets);
62                 }
63                 sam_header_parse_rg(fp->header);
64         } else if (mode[0] == 'w') { // write
65                 fp->header = bam_header_dup((const bam_header_t*)aux);
66                 if (mode[1] == 'b') { // binary
67                         char bmode[3];
68                         bmode[0] = 'w'; bmode[1] = strstr(mode, "u")? 'u' : 0; bmode[2] = 0;
69                         fp->type |= TYPE_BAM;
70                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, bmode) : bam_dopen(fileno(stdout), bmode);
71                         if (fp->x.bam == 0) goto open_err_ret;
72                         bam_header_write(fp->x.bam, fp->header);
73                 } else { // text
74                         // open file
75                         fp->x.tamw = strcmp(fn, "-")? fopen(fn, "w") : stdout;
76                         if (fp->x.tamr == 0) goto open_err_ret;
77                         // write header
78                         if (strstr(mode, "h")) {
79                                 int i;
80                                 bam_header_t *alt;
81                                 // parse the header text 
82                                 alt = bam_header_init();
83                                 alt->l_text = fp->header->l_text; alt->text = fp->header->text;
84                                 sam_header_parse(alt);
85                                 alt->l_text = 0; alt->text = 0;
86                                 // check if there are @SQ lines in the header
87                                 fwrite(fp->header->text, 1, fp->header->l_text, fp->x.tamw);
88                                 if (alt->n_targets) { // then write the header text without dumping ->target_{name,len}
89                                         if (alt->n_targets != fp->header->n_targets)
90                                                 fprintf(stderr, "[samopen] inconsistent number of target sequences.\n");
91                                 } else { // then dump ->target_{name,len}
92                                         for (i = 0; i < fp->header->n_targets; ++i)
93                                                 fprintf(fp->x.tamw, "@SQ\tSN:%s\tLN:%d\n", fp->header->target_name[i], fp->header->target_len[i]);
94                                 }
95                                 bam_header_destroy(alt);
96                         }
97                 }
98         }
99         return fp;
100
101 open_err_ret:
102         free(fp);
103         return 0;
104 }
105
106 void samclose(samfile_t *fp)
107 {
108         if (fp == 0) return;
109         if (fp->header) bam_header_destroy(fp->header);
110         if (fp->type & TYPE_BAM) bam_close(fp->x.bam);
111         else if (fp->type & TYPE_READ) sam_close(fp->x.tamr);
112         else fclose(fp->x.tamw);
113         free(fp);
114 }
115
116 int samread(samfile_t *fp, bam1_t *b)
117 {
118         if (fp == 0 || !(fp->type & TYPE_READ)) return -1; // not open for reading
119         if (fp->type & TYPE_BAM) return bam_read1(fp->x.bam, b);
120         else return sam_read1(fp->x.tamr, fp->header, b);
121 }
122
123 int samwrite(samfile_t *fp, const bam1_t *b)
124 {
125         if (fp == 0 || (fp->type & TYPE_READ)) return -1; // not open for writing
126         if (fp->type & TYPE_BAM) return bam_write1(fp->x.bam, b);
127         else {
128                 char *s = bam_format1(fp->header, b);
129                 int l = strlen(s);
130                 fputs(s, fp->x.tamw); fputc('\n', fp->x.tamw);
131                 free(s);
132                 return l + 1;
133         }
134 }
135
136 int sampileup(samfile_t *fp, int mask, bam_pileup_f func, void *func_data)
137 {
138         bam_plbuf_t *buf;
139         int ret;
140         bam1_t *b;
141         b = bam_init1();
142         buf = bam_plbuf_init(func, func_data);
143         bam_plbuf_set_mask(buf, mask);
144         while ((ret = samread(fp, b)) >= 0)
145                 bam_plbuf_push(b, buf);
146         bam_plbuf_push(0, buf);
147         bam_plbuf_destroy(buf);
148         bam_destroy1(b);
149         return 0;
150 }