#
# FORM{'comments'} contains the original message.
#
sub hudlomer{
#
# Determines so-called "functional style" of a text sample.
# A text size should be larger than 100 words.
#
$maxthreshold=18; # Maximal acceptable word length. Longer words should be ignored.
$ss=$FORM{'comments'}; # buffer
$ss=~s/[\.,\-"\!\?()\[\]]/ /g; # Delete punctuation signs
$ss=~s/[^áâ÷çäå³öúéêëìíîïðòóôõæèãþûýÿùøüàñÁÂ×ÇÄÅ£ÖÚÉÊËÌÍÎÏÐÒÓÔÕÆÈÃÞÛÝßÙØÜÀÑ ]//g; #capitalize
@words=split(/\s+/,$ss); # split the message into an array
($totalwords,$wordlensum,$wordlensumsq)=();
foreach $word( @words){ # Math
$l=length($word);
if( $l > 0 ){
if( $l > $maxthreshold ){$next}; # dicard long words
$wordlensum+=$l; # total length
$wordlensumsq+=$l*$l; # squared sum of lengths
$totalwords++; #total number of words
}
}
if( $totalwords > 0 ){
$averl=$wordlensum/$totalwords; # average length
$varl=$wordlensumsq/$totalwords-$averl*$averl; # variance
$varlteor=3.89*$averl-10.67; # theoretical variance. Used to compute the limits for 'natural' texts
}else{
$averl=0;
$varl=0;
}
if( $totalwords < 100 ){# Text is too short. Hudlomer is useless.
return 2;
}
if( !$FORM{'hudlomer'}){ # hudlomer is turned off. The average length is still logged by some other subroutine
return 3;
}
if( $varlteor - $varl > 3.89 ){ # Check whether this text is 'natural'
$termcolor="#00FF00"; #green == too small dispersion
}elsif( ($varlteor - $varl < -3.89)|| ($averl<3.9) ){
$termcolor="#5555FF"; #blue == too large dispersion. (too short words)
}else{
$termcolor="#FF0000"; # normal dispersion
}
if( $averl < 3.9 ){ $averl=3.9}; # too short average word length. Limit by 3.9
if( $averl > 7.9 ){ $averl=7.9}; # too long average word length. Limit by 7.9
$width=int(25*($averl-3.9)); # width of the HTML-table
# output the table
print GUEST<<__E1__;
| ÕÓÄËÎÌÅÐ Ñ×ÈÒÀÅÒ, ×ÒÎ ÝÒÎÒ ÎÒÇÛÂ: |
|---|
__E1__
$c1=$c2=$c3=$c4="";
# Choose a cell to highlight
if( $width < 25 ){
$c1=" BGCOLOR=$termcolor ";
}elsif( $width < 50 ){
$c2=" BGCOLOR=$termcolor ";
}elsif( $width < 75 ){
$c3=" BGCOLOR=$termcolor ";
}else{
$c4=" BGCOLOR=$termcolor ";
}
# draw a scale
print GUEST "| ÐÀÇÃÎÂÎÐÍÀß ÐÅ×Ü | ";
print GUEST "ÕÓÄËÎ | ";
print GUEST "ÃÀÇÅÒÍÀß ÑÒÀÒÜß | ";
print GUEST "ÍÀÓ×ÍÀß ÑÒÀÒÜß | ";
print GUEST<<__E2__;
__E2__
if( $width > 0){
print GUEST<<__E22__;
__E22__
}
return 1;
}