samtools/sam.c.pysam.c

   1 #include "pysam.h"
   2
   3 #include <string.h>
   4 #include <unistd.h>
   5 #include "faidx.h"
   6 #include "sam.h"
   7
   8 #define TYPE_BAM  1
   9 #define TYPE_READ 2
  10
  11 bam_header_t *bam_header_dup(const bam_header_t *h0)
  12 {
  13         bam_header_t *h;
  14         int i;
  15         h = bam_header_init();
  16         *h = *h0;
  17         h->hash = h->dict = h->rg2lib = 0;
  18         h->text = (char*)calloc(h->l_text + 1, 1);
  19         memcpy(h->text, h0->text, h->l_text);
  20         h->target_len = (uint32_t*)calloc(h->n_targets, 4);
  21         h->target_name = (char**)calloc(h->n_targets, sizeof(void*));
  22         for (i = 0; i < h->n_targets; ++i) {
  23                 h->target_len[i] = h0->target_len[i];
  24                 h->target_name[i] = strdup(h0->target_name[i]);
  25         }
  26         return h;
  27 }
  28 static void append_header_text(bam_header_t *header, char* text, int len)
  29 {
  30         int x = header->l_text + 1;
  31         int y = header->l_text + len + 1; // 1 byte null
  32         if (text == 0) return;
  33         kroundup32(x);
  34         kroundup32(y);
  35         if (x < y) header->text = (char*)realloc(header->text, y);
  36         strncpy(header->text + header->l_text, text, len); // we cannot use strcpy() here.
  37         header->l_text += len;
  38         header->text[header->l_text] = 0;
  39 }
  40
  41 samfile_t *samopen(const char *fn, const char *mode, const void *aux)
  42 {
  43         samfile_t *fp;
  44         fp = (samfile_t*)calloc(1, sizeof(samfile_t));
  45         if (strchr(mode, 'r')) { // read
  46                 fp->type |= TYPE_READ;
  47                 if (strchr(mode, 'b')) { // binary
  48                         fp->type |= TYPE_BAM;
  49                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, "r") : bam_dopen(fileno(stdin), "r");
  50                         if (fp->x.bam == 0) goto open_err_ret;
  51                         fp->header = bam_header_read(fp->x.bam);
  52                 } else { // text
  53                         fp->x.tamr = sam_open(fn);
  54                         if (fp->x.tamr == 0) goto open_err_ret;
  55                         fp->header = sam_header_read(fp->x.tamr);
  56                         if (fp->header->n_targets == 0) { // no @SQ fields
  57                                 if (aux) { // check if aux is present
  58                                         bam_header_t *textheader = fp->header;
  59                                         fp->header = sam_header_read2((const char*)aux);
  60                                         if (fp->header == 0) goto open_err_ret;
  61                                         append_header_text(fp->header, textheader->text, textheader->l_text);
  62                                         bam_header_destroy(textheader);
  63                                 }
  64                                 if (fp->header->n_targets == 0 && bam_verbose >= 1)
  65                                         fprintf(pysamerr, "[samopen] no @SQ lines in the header.\n");
  66                         } else if (bam_verbose >= 2) fprintf(pysamerr, "[samopen] SAM header is present: %d sequences.\n", fp->header->n_targets);
  67                 }
  68         } else if (strchr(mode, 'w')) { // write
  69                 fp->header = bam_header_dup((const bam_header_t*)aux);
  70                 if (strchr(mode, 'b')) { // binary
  71                         char bmode[3];
  72                         int i, compress_level = -1;
  73                         for (i = 0; mode[i]; ++i) if (mode[i] >= '0' && mode[i] <= '9') break;
  74                         if (mode[i]) compress_level = mode[i] - '0';
  75                         if (strchr(mode, 'u')) compress_level = 0;
  76                         bmode[0] = 'w'; bmode[1] = compress_level < 0? 0 : compress_level + '0'; bmode[2] = 0;
  77                         fp->type |= TYPE_BAM;
  78                         fp->x.bam = strcmp(fn, "-")? bam_open(fn, bmode) : bam_dopen(fileno(stdout), bmode);
  79                         if (fp->x.bam == 0) goto open_err_ret;
  80                         bam_header_write(fp->x.bam, fp->header);
  81                 } else { // text
  82                         // open file
  83                         fp->x.tamw = strcmp(fn, "-")? fopen(fn, "w") : stdout;
  84                         if (fp->x.tamr == 0) goto open_err_ret;
  85                         if (strchr(mode, 'X')) fp->type |= BAM_OFSTR<<2;
  86                         else if (strchr(mode, 'x')) fp->type |= BAM_OFHEX<<2;
  87                         else fp->type |= BAM_OFDEC<<2;
  88                         // write header
  89                         if (strchr(mode, 'h')) {
  90                                 int i;
  91                                 bam_header_t *alt;
  92                                 // parse the header text
  93                                 alt = bam_header_init();
  94                                 alt->l_text = fp->header->l_text; alt->text = fp->header->text;
  95                                 sam_header_parse(alt);
  96                                 alt->l_text = 0; alt->text = 0;
  97                                 // check if there are @SQ lines in the header
  98                                 fwrite(fp->header->text, 1, fp->header->l_text, fp->x.tamw); // FIXME: better to skip the trailing NULL
  99                                 if (alt->n_targets) { // then write the header text without dumping ->target_{name,len}
 100                                         if (alt->n_targets != fp->header->n_targets && bam_verbose >= 1)
 101                                                 fprintf(pysamerr, "[samopen] inconsistent number of target sequences. Output the text header.\n");
 102                                 } else { // then dump ->target_{name,len}
 103                                         for (i = 0; i < fp->header->n_targets; ++i)
 104                                                 fprintf(fp->x.tamw, "@SQ\tSN:%s\tLN:%d\n", fp->header->target_name[i], fp->header->target_len[i]);
 105                                 }
 106                                 bam_header_destroy(alt);
 107                         }
 108                 }
 109         }
 110         return fp;
 111
 112 open_err_ret:
 113         free(fp);
 114         return 0;
 115 }
 116
 117 void samclose(samfile_t *fp)
 118 {
 119         if (fp == 0) return;
 120         if (fp->header) bam_header_destroy(fp->header);
 121         if (fp->type & TYPE_BAM) bam_close(fp->x.bam);
 122         else if (fp->type & TYPE_READ) sam_close(fp->x.tamr);
 123         else fclose(fp->x.tamw);
 124         free(fp);
 125 }
 126
 127 int samread(samfile_t *fp, bam1_t *b)
 128 {
 129         if (fp == 0 || !(fp->type & TYPE_READ)) return -1; // not open for reading
 130         if (fp->type & TYPE_BAM) return bam_read1(fp->x.bam, b);
 131         else return sam_read1(fp->x.tamr, fp->header, b);
 132 }
 133
 134 int samwrite(samfile_t *fp, const bam1_t *b)
 135 {
 136         if (fp == 0 || (fp->type & TYPE_READ)) return -1; // not open for writing
 137         if (fp->type & TYPE_BAM) return bam_write1(fp->x.bam, b);
 138         else {
 139                 char *s = bam_format1_core(fp->header, b, fp->type>>2&3);
 140                 int l = strlen(s);
 141                 fputs(s, fp->x.tamw); fputc('\n', fp->x.tamw);
 142                 free(s);
 143                 return l + 1;
 144         }
 145 }
 146
 147 int sampileup(samfile_t *fp, int mask, bam_pileup_f func, void *func_data)
 148 {
 149         bam_plbuf_t *buf;
 150         int ret;
 151         bam1_t *b;
 152         b = bam_init1();
 153         buf = bam_plbuf_init(func, func_data);
 154         bam_plbuf_set_mask(buf, mask);
 155         while ((ret = samread(fp, b)) >= 0)
 156                 bam_plbuf_push(b, buf);
 157         bam_plbuf_push(0, buf);
 158         bam_plbuf_destroy(buf);
 159         bam_destroy1(b);
 160         return 0;
 161 }
 162
 163 char *samfaipath(const char *fn_ref)
 164 {
 165         char *fn_list = 0;
 166         if (fn_ref == 0) return 0;
 167         fn_list = calloc(strlen(fn_ref) + 5, 1);
 168         strcat(strcpy(fn_list, fn_ref), ".fai");
 169         if (access(fn_list, R_OK) == -1) { // fn_list is unreadable
 170                 if (access(fn_ref, R_OK) == -1) {
 171                         fprintf(pysamerr, "[samfaipath] fail to read file %s.\n", fn_ref);
 172                 } else {
 173                         if (bam_verbose >= 3) fprintf(pysamerr, "[samfaipath] build FASTA index...\n");
 174                         if (fai_build(fn_ref) == -1) {
 175                                 fprintf(pysamerr, "[samfaipath] fail to build FASTA index.\n");
 176                                 free(fn_list); fn_list = 0;
 177                         }
 178                 }
 179         }
 180         return fn_list;
 181 }