[popfile-commit] engine/Classifier Bayes.pm,1.99,1.100

[<jgr...@us...>]

Update of /cvsroot/popfile/engine/Classifier
In directory sc8-pr-cvs1:/tmp/cvs-serv31512/Classifier

Modified Files:
	Bayes.pm 
Log Message:
Merge helphand patch that prevents various unintialized variable warnings, greatly improves URL encoding and makes loading of the corpus safer so that corrupted information in the corpus is discarded

Index: Bayes.pm
===================================================================
RCS file: /cvsroot/popfile/engine/Classifier/Bayes.pm,v
retrieving revision 1.99
retrieving revision 1.100
diff -C2 -d -r1.99 -r1.100
*** Bayes.pm	19 Feb 2003 19:51:42 -0000	1.99
--- Bayes.pm	20 Feb 2003 16:25:53 -0000	1.100
***************
*** 464,468 ****
              }
  
!             if ( /(.+) (.+)/ ) {
                  my $word = $self->{mangler}->mangle($1,1);
                  my $value = $2;
--- 464,468 ----
              }
  
!             if ( /([^\s]+) (\d+)/ ) {
                  my $word = $self->{mangler}->mangle($1,1);
                  my $value = $2;
***************
*** 538,555 ****
          for my $type (sort keys %{$self->{magnets}{$bucket}}) {
  
! 			# You cannot use @ or $ inside a \Q\E regular expression and hence
! 			# we have to change the $magnet and the text we are comparing against
! 			# by changing the $ and @ signs to .
  
!         	my $noattype;
  
!         	$noattype = $self->{parser}->{$type};
!         	$noattype =~ s/[@\$]/\./g;
  
! 			for my $magnet (sort keys %{$self->{magnets}{$bucket}{$type}}) {
!             	my $regex;
  
!             	$regex = $magnet;
!             	$regex =~ s/[@\$]/\./g;
  
                  if ( $noattype =~ m/\Q$regex\E/i ) {
--- 538,555 ----
          for my $type (sort keys %{$self->{magnets}{$bucket}}) {
  
! 	    # You cannot use @ or $ inside a \Q\E regular expression and hence
! 	    # we have to change the $magnet and the text we are comparing against
! 	    # by changing the $ and @ signs to .
  
!             my $noattype;
  
!             $noattype = $self->{parser}->{$type};
!             $noattype =~ s/[@\$]/\./g;
  
! 	    for my $magnet (sort keys %{$self->{magnets}{$bucket}{$type}}) {
!                 my $regex;
  
!                 $regex = $magnet;
!                 $regex =~ s/[@\$]/\./g;
  
                  if ( $noattype =~ m/\Q$regex\E/i ) {