Speed-up reading of large header files

src/bwa.cpp

@@ -623,3 +623,36 @@ char *bwa_insert_header(const char *s, char *hdr)
  bwa_escape(hdr + len);
  return hdr;
 }
+
+char *bwa_insert_header_file(const FILE *fp, char *hdr)
+{
+ // Copy all lines beginning with @ from file to hdr at the same time
+ //
+ // This is more efficient than calling bwa_insert_header for every
+ // single line since bwa_insert_header computes strlen of existing hdr
+ // and reallocs it every time the function is called resulting in quadratic
+ // complexity.
+ char *buf, *p;
+ int file_size = 0, i;
+
+ fseek(fp, 0L, SEEK_END);
+ file_size = ftell(fp);
+ rewind(fp);
+
+ buf = p = (char *) calloc(1, file_size);
+ assert(buf != NULL);
+ while (fgets(p, 0xffff, fp))
+ {
+ if (p[0] == '@') {
+ i = strlen(p);
+ assert(p[i-1] == '\n');
+ p += i;
+ }
+ }
+ if (p != buf)
+ // bwa_insert_header expects a string without a trailing \n
+ p[-1] = 0;
+ hdr = bwa_insert_header(buf, hdr);
+ free(buf);
+ return hdr;
+}

src/bwa.h

@@ -104,6 +104,7 @@ extern "C" {
  void bwa_print_sam_hdr(const bntseq_t *bns, const char *hdr_line, FILE *fp);
  char *bwa_set_rg(const char *s);
  char *bwa_insert_header(const char *s, char *hdr);
+ char *bwa_insert_header_file(const FILE *fp, char *hdr);
 #ifdef __cplusplus
 }
 #endif

src/fastmap.cpp

@@ -739,17 +739,7 @@ int main_mem(int argc, char *argv[])
  FILE *fp;
  if ((fp = fopen(optarg, "r")) != 0)
  {
- char *buf;
- buf = (char *) calloc(1, 0x10000);
- assert(buf != NULL);
- while (fgets(buf, 0xffff, fp))
- {
- i = strlen(buf);
- assert(buf[i-1] == '\n');
- buf[i-1] = 0;
- hdr_line = bwa_insert_header(buf, hdr_line);
- }
- free(buf);
+ bwa_insert_header_file(fp, hdr_line);
  fclose(fp);
  }
  } else hdr_line = bwa_insert_header(optarg, hdr_line);