amdym.c

ADaM is a data mining and image processing toolkit

    if (val == 0.0) ivec_set( iv, i, 0);
    else if (val == 1.0) ivec_set( iv, i, 1);
    else {
      my_errorf( "mk_ivec_from_binary_dym_column: row=%d, col=%d, val=%f, "
                 "not binary.", i, col, val);
    }
  }

  return iv;
}

void fprintf_dym(FILE *s, const char *m1, const dym *d, const char *m2)
{
  if ( d == NULL ){
    fprintf(s,"%s = (dym *)NULL%s",m1,m2);
  }
  else if ( d->rows <= 0 || d->cols <= 0 )
    fprintf(s,"%s = <Dym with %d row%s and %d column%s>%s",
            m1,d->rows,(d->rows==-1)?"":"s",
            d->cols,(d->cols==-1)?"":"s",m2
           );
  else
  {
    int i;
    buftab bt;

    init_buftab(&bt,d->rows,d->cols + 4);

    for ( i = 0 ; i < d->rows ; i++ )
    {
      int j;
      set_buftab(&bt,i,0,(i == (d->rows-1)/2) ? m1 : "");
      set_buftab(&bt,i,1,(i == (d->rows-1)/2) ? "=" : "");
      set_buftab(&bt,i,2,"[");

      for ( j = 0 ; j < d -> cols ; j++ )
      {
        char buff[100];
        sprintf(buff," %g ",d->tdarr[i][j]);
        set_buftab(&bt,i,3+j,buff);
      }

      set_buftab(&bt,i,3+d->cols,"]");
    }

    fprint_buftab(s,&bt);
    free_buftab_contents(&bt);
  }
  fprintf(s,"\n");
}


 /** Read the values from the file and set them in dym 
 *  fp: File pointer at the beginning of the file
 *  d : allocated dym
 */
int assign_dym_values(PFILE* fp, dym* d)
{
  int rc, j, nrows, ncols, rowno, lineno;
  double v;
  char *line, *s, *endptr;
  string_array *sa;

  rc = 0;

  if (fp == NULL) my_error( "assign_dym_values: FILE *fp=NULL.");
  if (d == NULL) my_error( "assign_dym_values: dym *d=NULL.");

  nrows = dym_rows( d);
  ncols = dym_cols( d);
  lineno = 0;
  rowno = -1;

  while (!pfeof(fp)) {
    /* Get line. */
    line = mk_readline( fp);

    /* Check it is defined. */
    if (line==NULL) {
      if (pfeof(fp)) break; /* Read eof. */
      else my_error( "assign_dym_values: mk_readline() returned NULL.");
    }

    /* Line is happy. */
    lineno += 1;

    /* Skip comments. */
    if (is_comment(line)) {
      free( line);
      continue;
    }

    /* Line is not a comment. */
    rowno += 1;
    sa = mk_broken_string_using_seppers( line, (const char*) delim);
    if (string_array_size(sa) != ncols) {
      printf( "assign_dym_values: bad line:\n%s\n", line);
      my_errorf( "assign_dym_values: line %d has %d atts, but the data file\n"
                 "appears to have %d atts.",
                 lineno, string_array_size(sa), ncols);
    }

    /* Parse numbers from line. */
    for (j=0; j<ncols; ++j) {
      s = string_array_ref(sa,j);
      v = strtod( string_array_ref(sa,j), &endptr);
      if(errno == ERANGE) {
        printf( "assign_dym_values: bad line:\n%s\n", line);
        my_errorf( "assign_dym_values: value '%s' cause over- or underflow.\n"
                   "line=%d, att=%d\n", s, lineno, j+1);
      }
      if (endptr == s) {
        printf( "assign_dym_values: bad line:\n%s\n", line);
        my_errorf( "assign_dym_values: failed to convert '%s' to a number\n"
                   "line=%d, att=%d\n", s, lineno, j+1);
      }

      /* Set dym value. */
      v =d->tdarr[rowno][j] = v;
    }

    /* Done with line and tokens. */
    free_string_array( sa);
    free( line);
  }

  if (rowno+1 != nrows) {
    my_errorf( "assign_dym_values: read %d lines of data, but expected %d from"
               " first pass.\n",
               rowno+1, nrows);
  }

  return rc;
}

 /** Read the values from the file and set them in dym 
 *  fp: File pointer at the beginning of the file
 *  factors: allocated dym
 *  outputs: allocated dyv
 */
int assign_dym_values_for_csv( PFILE* fp, dym *factors, dyv *outputs)
{
  int rc, j, nrows, ncols, rowno, lineno;
  double v;
  char *line, *s, *endptr;
  string_array *sa;

  rc = 0;

  if (fp == NULL) my_error( "assign_dym_values_for_csv: FILE *fp=NULL.");
  if (factors == NULL) my_error( "assign_dym_values_for_csv: dym *d=NULL.");
  if (outputs == NULL) my_error( "assign_dym_values_for_csv: dym *d=NULL.");

  nrows = dym_rows( factors);
  ncols = dym_cols( factors);
  lineno = 0;
  rowno = -1;

  while (!pfeof(fp)) {
    /* Get line. */
    line = mk_readline( fp);

    /* Check it is defined. */
    if (line==NULL) {
      if (pfeof(fp)) break; /* Read eof. */
      else my_error( "assign_dym_values_for_csv: mk_readline() returned "
                     "NULL.");
    }

    /* Line is happy. */
    lineno += 1;

    /* Skip comments. */
    if (is_comment(line)) {
      free( line);
      continue;
    }

    /* Line is not a comment. */
    rowno += 1;
    sa = mk_broken_string_using_seppers( line, (const char*) delim);
    if (string_array_size(sa) != ncols+1) {
      printf( "assign_dym_values_for_csv: bad line:\n%s\n", line);
      my_errorf( "assign_dym_values_for_csv: line %d has %d values, but we\n"
                 "thought the data file had %d values per line.",
                 lineno, string_array_size(sa), ncols+1);
    }

    /* Parse numbers from line. */
    for (j=0; j < (ncols+1); ++j) {
      s = string_array_ref(sa,j);
      v = strtod( string_array_ref(sa,j), &endptr);
      if(errno == ERANGE) {
        printf( "assign_dym_values_for_csv: bad line:\n%s\n", line);
        my_errorf( "assign_dym_values_for_csv: value '%s' cause over- or "
                   "underflow.\n"
                   "line=%d, att=%d\n", s, lineno, j+1);
      }
      if (endptr == s) {
        printf( "assign_dym_values_for_csv: bad line:\n%s\n", line);
        my_errorf( "assign_dym_values_for_csv: failed to convert '%s' to a "
                   "number\n"
                   "line=%d, att=%d\n", s, lineno, j+1);
      }

      /* Set value. */
      if (j < ncols) dym_set( factors, rowno, j, v);
      else dyv_set( outputs, rowno, v);
    }

    /* Done with line and tokens. */
    free_string_array( sa);
    free( line);
  }

  if (rowno+1 != nrows) {
    my_errorf( "assign_dym_values_for_csv: read %d lines of data, but "
               "expected %d from first pass.\n",
               rowno+1, nrows);
  }

  return rc;
}

char *mk_readline(PFILE* fp )
{
   char *cptr, *in;

   in = (char *) malloc(MAX_LINE_SIZE * sizeof(char));

   if(in==NULL) {
     cptr = NULL;
     my_error( "mk_readline: Failed to allocate memory");
   }
   else {
     cptr = pfgets(in, MAX_LINE_SIZE, fp);
     if (cptr == NULL) free( in);
   }

   return cptr;
}


int find_number_of_columns(const char* line)
{
  int ncols = 0;

  if( line ){
    string_array *sa = mk_broken_string_using_seppers(line,(const char*)delim);
    ncols = string_array_size(sa);
    free_string_array(sa);
  }

  return ncols;

}

int find_number_of_rows_and_columns(PFILE* fp , int *nrows , int *ncols )
{
  int cols;
  char *line;
  *ncols = 0;
  *nrows = 0;

  while( ! pfeof(fp) ){
    line = mk_readline(fp);
    if(line) {
      if (!is_comment(line)) {
        cols = find_number_of_columns(line);
        if( *ncols < cols) *ncols = cols;
        ++(*nrows);
      }
      free(line);
    }
  }

  return 0;

}

int is_comment(const char* line)
{
  /* Check if first non-whitespace char is '#'. */
  size_t pos;

  /* First non-whitespace char with C99 function strspn. */
  pos = strspn( line, " \t");

  if (pos == strlen(line)-1) {
    /* First non-whitespace char is \n.  Treat as comment. */
    return 1;
  }
  else if (line[pos] == '#') {
    /* Comment. */
    return 1;
  }
  else {
    /* Not a comment. */
    return 0;
  }
}



/*************************************************************************/
/* TESTING                                                               */
/*************************************************************************/

#if 0
int assign_value(dym* d)
{

  if( d ){
    int i = 0;
    int j = 0;

    for(i =0 ; i < d->rows; i++){
      for(j=0 ; j < d->cols ; j++){
	d->tdarr[i][j] = 1.5;
      }
    }
  }

  return 0;
}

int display_dym(dym* d)
{

  if( d ){
    int i = 0;
    int j = 0;

    for(i =0 ; i < d->rows; i++){
      for(j=0 ; j < d->cols ; j++){
	printf("%g ",d->tdarr[i][j] );
      }
      printf("\n");
    }
  }

  return 0;
}



int display(PFILE *fp)
{
  int r , c;

  find_number_of_rows_and_columns(fp , &r ,&c);

  printf(" display %d %d\n",r,c);

  while (! pfeof(fp) ){
    char* line = mk_readline(fp);
    if(line){
      printf("line %s\n",line);
      printf("ncols %d\n",find_number_of_columns(line));
      free(line);
    }
  }

  return 0;
}

int main(int argc, char** argv)
{
  dym* d = mk_read_dym("dym.out");
  if( d ) {
    display_dym(d);
    free_dym(d);
  }

  return 0;

}

#endif