sam.c

   1 #include <string.h>
   2 #include <unistd.h>
   3 #include "faidx.h"
   4 #include "sam.h"
   5
   6 #define TYPE_BAM  1
   7 #define TYPE_READ 2
   8
   9 bam_header_t *bam_header_dup(const bam_header_t *h0)
  10 {
  11         bam_header_t *h;
  12         int i;
  13         h = bam_header_init();
  14         *h = *h0;
  15         h->hash = 0;
  16         h->text = (char*)calloc(h->l_text + 1, 1);
  17         memcpy(h->text, h0->text, h->l_text);
  18         h->target_len = (uint32_t*)calloc(h->n_targets, 4);
  19         h->target_name = (char**)calloc(h->n_targets, sizeof(void*));
  20         for (i = 0; i < h->n_targets; ++i) {
  21                 h->target_len[i] = h0->target_len[i];
  22                 h->target_name[i] = strdup(h0->target_name[i]);
  23         }
  24         if (h0->rg2lib) h->rg2lib = bam_strmap_dup(h0->rg2lib);
  25         return h;
  26 }
  27 static void append_header_text(bam_header_t *header, char* text, int len)
  28 {
  29         int x = header->l_text + 1;
  30         int y = header->l_text + len + 1; // 1 byte null
  31         if (text == 0) return;
  32         kroundup32(x);
  33         kroundup32(y);
  34         if (x < y) header->text = (char*)realloc(header->text, y);
  35         strncpy(header->text + header->l_text, text, len); // we cannot use strcpy() here.
  36         header->l_text += len;
  37         header->text[header->l_text] = 0;
  38 }
  39
  40 samfile_t *samopen(const char *fn, const char *mode, const void *aux)
  41 {
  42         samfile_t *fp;
  43         fp = (samfile_t*)calloc(1, sizeof(samfile_t));
  44         if (mode[0] == 'r') { // read
  45                 fp->type |= TYPE_READ;
  46                 if (mode[1] == 'b') { // binary
  47                         fp->type |= TYPE_BAM;
  48                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, "r") : bam_dopen(fileno(stdin), "r");
  49                         if (fp->x.bam == 0) goto open_err_ret;
  50                         fp->header = bam_header_read(fp->x.bam);
  51                 } else { // text
  52                         fp->x.tamr = sam_open(fn);
  53                         if (fp->x.tamr == 0) goto open_err_ret;
  54                         fp->header = sam_header_read(fp->x.tamr);
  55                         if (fp->header->n_targets == 0) { // no @SQ fields
  56                                 if (aux) { // check if aux is present
  57                                         bam_header_t *textheader = fp->header;
  58                                         fp->header = sam_header_read2((const char*)aux);
  59                                         append_header_text(fp->header, textheader->text, textheader->l_text);
  60                                         bam_header_destroy(textheader);
  61                                 }
  62                                 if (fp->header->n_targets == 0)
  63                                         fprintf(stderr, "[samopen] no @SQ lines in the header.\n");
  64                         } else fprintf(stderr, "[samopen] SAM header is present: %d sequences.\n", fp->header->n_targets);
  65                 }
  66                 sam_header_parse_rg(fp->header);
  67         } else if (mode[0] == 'w') { // write
  68                 fp->header = bam_header_dup((const bam_header_t*)aux);
  69                 if (mode[1] == 'b') { // binary
  70                         char bmode[3];
  71                         bmode[0] = 'w'; bmode[1] = strstr(mode, "u")? 'u' : 0; bmode[2] = 0;
  72                         fp->type |= TYPE_BAM;
  73                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, bmode) : bam_dopen(fileno(stdout), bmode);
  74                         if (fp->x.bam == 0) goto open_err_ret;
  75                         bam_header_write(fp->x.bam, fp->header);
  76                 } else { // text
  77                         // open file
  78                         fp->x.tamw = strcmp(fn, "-")? fopen(fn, "w") : stdout;
  79                         if (fp->x.tamr == 0) goto open_err_ret;
  80                         if (strstr(mode, "X")) fp->type |= BAM_OFSTR<<2;
  81                         else if (strstr(mode, "x")) fp->type |= BAM_OFHEX<<2;
  82                         else fp->type |= BAM_OFDEC<<2;
  83                         // write header
  84                         if (strstr(mode, "h")) {
  85                                 int i;
  86                                 bam_header_t *alt;
  87                                 // parse the header text
  88                                 alt = bam_header_init();
  89                                 alt->l_text = fp->header->l_text; alt->text = fp->header->text;
  90                                 sam_header_parse(alt);
  91                                 alt->l_text = 0; alt->text = 0;
  92                                 // check if there are @SQ lines in the header
  93                                 fwrite(fp->header->text, 1, fp->header->l_text, fp->x.tamw);
  94                                 if (alt->n_targets) { // then write the header text without dumping ->target_{name,len}
  95                                         if (alt->n_targets != fp->header->n_targets)
  96                                                 fprintf(stderr, "[samopen] inconsistent number of target sequences.\n");
  97                                 } else { // then dump ->target_{name,len}
  98                                         for (i = 0; i < fp->header->n_targets; ++i)
  99                                                 fprintf(fp->x.tamw, "@SQ\tSN:%s\tLN:%d\n", fp->header->target_name[i], fp->header->target_len[i]);
 100                                 }
 101                                 bam_header_destroy(alt);
 102                         }
 103                 }
 104         }
 105         return fp;
 106
 107 open_err_ret:
 108         free(fp);
 109         return 0;
 110 }
 111
 112 void samclose(samfile_t *fp)
 113 {
 114         if (fp == 0) return;
 115         if (fp->header) bam_header_destroy(fp->header);
 116         if (fp->type & TYPE_BAM) bam_close(fp->x.bam);
 117         else if (fp->type & TYPE_READ) sam_close(fp->x.tamr);
 118         else fclose(fp->x.tamw);
 119         free(fp);
 120 }
 121
 122 int samread(samfile_t *fp, bam1_t *b)
 123 {
 124         if (fp == 0 || !(fp->type & TYPE_READ)) return -1; // not open for reading
 125         if (fp->type & TYPE_BAM) return bam_read1(fp->x.bam, b);
 126         else return sam_read1(fp->x.tamr, fp->header, b);
 127 }
 128
 129 int samwrite(samfile_t *fp, const bam1_t *b)
 130 {
 131         if (fp == 0 || (fp->type & TYPE_READ)) return -1; // not open for writing
 132         if (fp->type & TYPE_BAM) return bam_write1(fp->x.bam, b);
 133         else {
 134                 char *s = bam_format1_core(fp->header, b, fp->type>>2&3);
 135                 int l = strlen(s);
 136                 fputs(s, fp->x.tamw); fputc('\n', fp->x.tamw);
 137                 free(s);
 138                 return l + 1;
 139         }
 140 }
 141
 142 int sampileup(samfile_t *fp, int mask, bam_pileup_f func, void *func_data)
 143 {
 144         bam_plbuf_t *buf;
 145         int ret;
 146         bam1_t *b;
 147         b = bam_init1();
 148         buf = bam_plbuf_init(func, func_data);
 149         bam_plbuf_set_mask(buf, mask);
 150         while ((ret = samread(fp, b)) >= 0)
 151                 bam_plbuf_push(b, buf);
 152         bam_plbuf_push(0, buf);
 153         bam_plbuf_destroy(buf);
 154         bam_destroy1(b);
 155         return 0;
 156 }
 157
 158 char *samfaipath(const char *fn_ref)
 159 {
 160         char *fn_list = 0;
 161         if (fn_ref == 0) return 0;
 162         fn_list = calloc(strlen(fn_ref) + 5, 1);
 163         strcat(strcpy(fn_list, fn_ref), ".fai");
 164         if (access(fn_list, R_OK) == -1) { // fn_list is unreadable
 165                 if (access(fn_ref, R_OK) == -1) {
 166                         fprintf(stderr, "[samfaipath] fail to read file %s.\n", fn_ref);
 167                 } else {
 168                         fprintf(stderr, "[samfaipath] build FASTA index...\n");
 169                         if (fai_build(fn_ref) == -1) {
 170                                 fprintf(stderr, "[samfaipath] fail to build FASTA index.\n");
 171                                 free(fn_list); fn_list = 0;
 172                         }
 173                 }
 174         }
 175         return fn_list;
 176 }