/* oct27.ex1                            */
 /* apply first approach to fed paper    */
  options ls=80 ;
  data a ;
    keep gword ;                         * just words ;
    length line line2 $1000 word $100 ;  * lengths of char variables ;
    length gword $30 ;                   * good words can be shorter ;
    infile 'w6y2009' pad lrecl=1000 ;
    retain go 1 ;                        * flag for no html ;
    input line 1-1000 ;
    line2 = tranwrd(line,'<',' <') ;     * space in front of < ;
    line2 = tranwrd(line2,'>','> ') ;    * space in back of > ;
    num = countw(line2,' ') ;
     do i = 1 to num ;                   * start loop *******;
    word = scan(line2,i,' ') ;
     lw = length(word) ;
     if( substr(word,1,1) eq '<' ) then go = 0 ;  * leading < ;
     if( go eq 1 ) then do ;
                        gword = compress(word,'.,-/;:','s') ;
                        lword = lengthn(gword);
           if(lword > 0 ) then  output ;         * only good ones ;
                        end ;            * ********** loop ;
     if( substr(word,lw,1) eq '>') then go = 1 ;  * trailing > ;
     end ;
  run ;
 /* print out results                    */
  proc print data=a (firstobs=200) ;
    var gword ;
    title 'still first approach' ;
  run ;