functions in rdcols.i -

             cols = rdcols(f)  
         or cols = rdcols(f, ncols)  
            eq_nocopy, col1, *cols(1)  
            eq_nocopy, col2, *cols(2)  
            ...  
            eq_nocopy, colN, *cols(ncols)  
 
    
   cracks ascii file F into NCOLS columns, returning an array of  
   pointers to the columns.  A "column" is defined by either its  
   width in characters, or by the appearance of a delimiting  
   character (such as a blank, a comma, or a tab).  Multiple  
   delimiting characters may either be skipped (as usual if the  
   character is a blank), or may represent a number of empty  
   columns (as for reading a typical text file exported from a  
   spreadsheet or database program).  
    
   rdcols returns an array of pointers COLS, with *COLS(i) the  
   contents of the i-th column.  The NCOLS parameter may be omitted  
   in which case rdcols will guess the number of columns.  
    
   F may be either a file handle or a filename.  
    
   nskip=m       skip the first M lines of the file (M=0 by default)  
   nlines=n      read only the first M+N lines of the file  
   comment=text  interpret lines beginning with TEXT (possibly preceded  
                 by blanks) as comments (blank lines are always comments)  
   width=w       column width, or list of widths (w=0 means any width)  
   delim=d       column delimiters (as for strtok), or list of delimiters  
   type=t        column type, or list of types  
     t=0 means guess, t=1 means string, t=2 means integer, t=3 means real,  
     t=4 means either integer or real  
   marker=k      marker column delimiters (as for strtok)  
   missing=m     missing value, or list of values  
                 for empty numeric columns  
    
   The width=, delim=, type=, and missing= keywords may either be scalar  
   values to apply to every column, or lists to apply to successive  
   columns.  The marker= keyword is similar to the delim= keyword,  
   except that multiple consecutive occurrences indicate empty  
   columns.  If any list is shorter than NCOLS, its final value  
   applies to all remaining columns.  If any list is longer than  
   NCOLS, its trailing values are silently ignored.  
    
   By default, width=0.  Non-zero width is very tricky, since a  
   column begins wherever the previous column ended; if the previous  
   column had width=0, its delimiter is treated as a part of it, so  
   it is not included in the width count.  Consequently, the width=  
   keyword is best suited to situations in which only the final  
   width value is 0.  
    
   By default, marker=[].  If present, then every column is delimited  
   by a character in marker, and every occurrence of such a character  
   indicates a new column.  Use marker="\t" to read tab-delimited text  
   files exported from a spreadsheet or database program.  Non-zero  
   width= values supersede marker=, but delim= is ignored if marker  
   is present.  
    
   By default, delim=" \t",  and type=0, so that blanks and tabs delimit  
   columns, and rdcols guesses whether a column contains string or  
   numeric data.  A numeric column ends at the first character that  
   cannot be interpreted as part of the number, whether or not that is  
   a character in delim.  If you use type= to force a numeric column,  
   empty rows or rows which are not numbers get the value 0.0 by  
   default, or the value specified by the missing= keyword.  Missing  
   values in string columns always get the value "".  
    
   If you need finer control of the returned types than provided by  
   the type= keyword, see rdconvert.  
    

             rdconvert, cols, type1, type2, type3, ...  
         or cols = rdconvert(cols, type1, type2, type3, ...)  
 
    
   converts the data types of the COLS, so that COLS(1) becomes  
   TYPE1, COLS(2) TYPE2, COLS(3) TYPE3, and so on.  The COLS  
   array is an array of pointers as returned by rdcols.  If any  
   of the TYPEi is nil, that column is not altered.  This function  
   is only useful if you are not satisfied with either the long  
   or double types returned by rdcols, e.g.-  
      cols = rdcols("datafile", 5);  
      rdconvert, cols, , short, float, , int;