/* read a file containing floating vectors in various formats. Check the NaN */
float * my_fvec_read (const char * fname, int fmt, int verbose,
int * nptr, int * dptr)
{
int ret = -1;
float * v = NULL;
int n = *nptr;
int d = *dptr;
/* read the vectors from the input file, and sanitize them if needed */
switch (fmt) {
case FMT_FVEC:
if (d == 0 || n == 0) {/* automatically read the dimension */
ret = fvecs_fsize (fname, &d, &n);
fprintf (stderr, "File %s contains (%d bytes) %d vectors of dimension %d\n",
fname, ret, n, d);
assert (ret);
}
v = fvec_new (n * d);
ret = fvecs_read (fname, d, n, v);
break;
case FMT_TEXT:
assert (n !=0 && d != 0);
v = fvec_new (n * d);
ret = fvecs_read_txt (fname, d, n, v);
break;
case FMT_BVEC:
if (d == 0 || n == 0) {/* automatically read the dimension */
ret = bvecs_fsize (fname, &d, &n);
fprintf (stderr, "File %s contains (%d bytes) %d vectors of dimension %d\n",
fname, ret, n, d);
assert (ret);
}
v = fvec_new (n * d);
ret = b2fvecs_read (fname, d, n, v);
break;
default: assert (1 || "Unknown input format\n");
}
assert (ret >= n);
/* Remove the Nan values */
int nNaN = fvec_purge_nans (v, n * d, 2);
if (nNaN > 0 && verbose >= 1)
fprintf (stderr, "found %d NaN values in what read from %s\n",
nNaN, fname);
*dptr = d;
*nptr = n;
return v;
}
int b2fvecs_read (const char *fname, int d, int n, float *v)
{
int n_new;
int d_new;
bvecs_fsize (fname, &d_new, &n_new);
assert (d_new == d);
assert (n <= n_new);
FILE * f = fopen (fname, "r");
assert (f || "b2fvecs_read: Unable to open the file");
b2fvecs_fread (f, v, n);
fclose (f);
return n;
}
int b2fvecs_new_read (const char *fname, int *d_out, float **v_out)
{
int n;
int d;
bvecs_fsize (fname, &d, &n);
float * v = fvec_new ((long) n * (long) d);
FILE * f = fopen (fname, "r");
assert (f || "bvecs_new_read: Unable to open the file");
b2fvecs_fread (f, v, n);
fclose (f);
v_out = &v;
return n;
}
