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 = h->dict = h->rg2lib = 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         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 (strchr(mode, 'r')) { // read
  44                 fp->type |= TYPE_READ;
  45                 if (strchr(mode, '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                                         if (fp->header == 0) goto open_err_ret;
  59                                         append_header_text(fp->header, textheader->text, textheader->l_text);
  60                                         bam_header_destroy(textheader);
  61                                 }
  62                                 if (fp->header->n_targets == 0 && bam_verbose >= 1)
  63                                         fprintf(stderr, "[samopen] no @SQ lines in the header.\n");
  64                         } else if (bam_verbose >= 2) fprintf(stderr, "[samopen] SAM header is present: %d sequences.\n", fp->header->n_targets);
  65                 }
  66         } else if (strchr(mode, 'w')) { // write
  67                 fp->header = bam_header_dup((const bam_header_t*)aux);
  68                 if (strchr(mode, 'b')) { // binary
  69                         char bmode[3];
  70                         int i, compress_level = -1;
  71                         for (i = 0; mode[i]; ++i) if (mode[i] >= '0' && mode[i] <= '9') break;
  72                         if (mode[i]) compress_level = mode[i] - '0';
  73                         if (strchr(mode, 'u')) compress_level = 0;
  74                         bmode[0] = 'w'; bmode[1] = compress_level < 0? 0 : compress_level + '0'; bmode[2] = 0;
  75                         fp->type |= TYPE_BAM;
  76                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, bmode) : bam_dopen(fileno(stdout), bmode);
  77                         if (fp->x.bam == 0) goto open_err_ret;
  78                         bam_header_write(fp->x.bam, fp->header);
  79                 } else { // text
  80                         // open file
  81                         fp->x.tamw = strcmp(fn, "-")? fopen(fn, "w") : stdout;
  82                         if (fp->x.tamr == 0) goto open_err_ret;
  83                         if (strchr(mode, 'X')) fp->type |= BAM_OFSTR<<2;
  84                         else if (strchr(mode, 'x')) fp->type |= BAM_OFHEX<<2;
  85                         else fp->type |= BAM_OFDEC<<2;
  86                         // write header
  87                         if (strchr(mode, 'h')) {
  88                                 int i;
  89                                 bam_header_t *alt;
  90                                 // parse the header text
  91                                 alt = bam_header_init();
  92                                 alt->l_text = fp->header->l_text; alt->text = fp->header->text;
  93                                 sam_header_parse(alt);
  94                                 alt->l_text = 0; alt->text = 0;
  95                                 // check if there are @SQ lines in the header
  96                                 fwrite(fp->header->text, 1, fp->header->l_text, fp->x.tamw); // FIXME: better to skip the trailing NULL
  97                                 if (alt->n_targets) { // then write the header text without dumping ->target_{name,len}
  98                                         if (alt->n_targets != fp->header->n_targets && bam_verbose >= 1)
  99                                                 fprintf(stderr, "[samopen] inconsistent number of target sequences. Output the text header.\n");
 100                                 } else { // then dump ->target_{name,len}
 101                                         for (i = 0; i < fp->header->n_targets; ++i)
 102                                                 fprintf(fp->x.tamw, "@SQ\tSN:%s\tLN:%d\n", fp->header->target_name[i], fp->header->target_len[i]);
 103                                 }
 104                                 bam_header_destroy(alt);
 105                         }
 106                 }
 107         }
 108         return fp;
 109
 110 open_err_ret:
 111         free(fp);
 112         return 0;
 113 }
 114
 115 void samclose(samfile_t *fp)
 116 {
 117         if (fp == 0) return;
 118         if (fp->header) bam_header_destroy(fp->header);
 119         if (fp->type & TYPE_BAM) bam_close(fp->x.bam);
 120         else if (fp->type & TYPE_READ) sam_close(fp->x.tamr);
 121         else fclose(fp->x.tamw);
 122         free(fp);
 123 }
 124
 125 int samread(samfile_t *fp, bam1_t *b)
 126 {
 127         if (fp == 0 || !(fp->type & TYPE_READ)) return -1; // not open for reading
 128         if (fp->type & TYPE_BAM) return bam_read1(fp->x.bam, b);
 129         else return sam_read1(fp->x.tamr, fp->header, b);
 130 }
 131
 132 int samwrite(samfile_t *fp, const bam1_t *b)
 133 {
 134         if (fp == 0 || (fp->type & TYPE_READ)) return -1; // not open for writing
 135         if (fp->type & TYPE_BAM) return bam_write1(fp->x.bam, b);
 136         else {
 137                 char *s = bam_format1_core(fp->header, b, fp->type>>2&3);
 138                 int l = strlen(s);
 139                 fputs(s, fp->x.tamw); fputc('\n', fp->x.tamw);
 140                 free(s);
 141                 return l + 1;
 142         }
 143 }
 144
 145 int sampileup(samfile_t *fp, int mask, bam_pileup_f func, void *func_data)
 146 {
 147         bam_plbuf_t *buf;
 148         int ret;
 149         bam1_t *b;
 150         b = bam_init1();
 151         buf = bam_plbuf_init(func, func_data);
 152         bam_plbuf_set_mask(buf, mask);
 153         while ((ret = samread(fp, b)) >= 0)
 154                 bam_plbuf_push(b, buf);
 155         bam_plbuf_push(0, buf);
 156         bam_plbuf_destroy(buf);
 157         bam_destroy1(b);
 158         return 0;
 159 }
 160
 161 char *samfaipath(const char *fn_ref)
 162 {
 163         char *fn_list = 0;
 164         if (fn_ref == 0) return 0;
 165         fn_list = calloc(strlen(fn_ref) + 5, 1);
 166         strcat(strcpy(fn_list, fn_ref), ".fai");
 167         if (access(fn_list, R_OK) == -1) { // fn_list is unreadable
 168                 if (access(fn_ref, R_OK) == -1) {
 169                         fprintf(stderr, "[samfaipath] fail to read file %s.\n", fn_ref);
 170                 } else {
 171                         if (bam_verbose >= 3) fprintf(stderr, "[samfaipath] build FASTA index...\n");
 172                         if (fai_build(fn_ref) == -1) {
 173                                 fprintf(stderr, "[samfaipath] fail to build FASTA index.\n");
 174                                 free(fn_list); fn_list = 0;
 175                         }
 176                 }
 177         }
 178         return fn_list;
 179 }