#!/usr/bin/perl

$file="testfile1";
#...... grab the file via curl, only do once
#  system("curl 'http://squirrel.sr.unh.edu/~jraeder/PHYS-795-895-F2017/uploads/testfile1' > $file"); exit;

#...... simplest way to read entire file
$_=`cat $file`;
#print $_;  exit;


#...... get rid of unwanted characters
s/\"/ /g; s/\?/ /g; s/\./ /g; s/\,/ /g; s/\*/ /g; s/\'/ /g; 
$text=$_;
$text=~s/[^a-zA-Z]/ /g;  #..... convert all non-letters to blanks

#  print $_; exit;  #...... see what we got

#...... loop over lines and print with numbers
  $k=0; foreach(split('\n',$text)){ $kk=sprintf("%5.5d",$k++); print "$kk $_\n"; }

#...... loop over lines and count words
%W=();
foreach(split('\n',$text)){ @a=split(' ',$_);
    foreach $w (@a){ $ww=lc($w); $W{$ww}++; }  }
#..... find max
$m=0; foreach(keys(%W)){ $x=$W{$_}; if($x>$m){ $m=$x; } }  print "max count:  $m\n";
for($i=$m;$i>5;$i--){ 
    $line=''; $n=0;
    foreach(keys(%W)){ $x=$W{$_}; if($x == $i){ $line="$line $_"; $n++; } }
    if(length($line)>0){ print " $i occurrences ($n words):  $line\n"; }
}