[popfile-commit] engine/Classifier Bayes.pm,1.212,1.213

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

Update of /cvsroot/popfile/engine/Classifier
In directory sc8-pr-cvs1:/tmp/cvs-serv25504/Classifier

Modified Files:
	Bayes.pm 
Log Message:
Entries in the table files with a count of 0 screw up the upgrade process, fix that and add a test case

Index: Bayes.pm
===================================================================
RCS file: /cvsroot/popfile/engine/Classifier/Bayes.pm,v
retrieving revision 1.212
retrieving revision 1.213
diff -C2 -d -r1.212 -r1.213
*** Bayes.pm	16 Oct 2003 17:55:12 -0000	1.212
--- Bayes.pm	17 Oct 2003 06:12:03 -0000	1.213
***************
*** 714,718 ****
  
                          if ( /^([^\s]+) (\d+)$/ ) {
!                             $self->set_value_( $bucket, $1, $2 );
                          } else {
                              $self->log_( "Found entry in corpus for $bucket that looks wrong: \"$_\" (ignoring)" );
--- 714,720 ----
  
                          if ( /^([^\s]+) (\d+)$/ ) {
!   			    if ( $2 != 0 ) {
!                                 $self->set_value_( $bucket, $1, $2 );
! 			    }
                          } else {
                              $self->log_( "Found entry in corpus for $bucket that looks wrong: \"$_\" (ignoring)" );
***************
*** 757,767 ****
  
                          if ( /^([^\s]+) (\d+)$/ ) {
! 			    if ( $self->get_base_value_( $bucket, $1 ) != $2 ) {
!                                 print "\nUpgrade error for word $1 in bucket $bucket.\nShutdown POPFile and rerun.\n";
!                                 $upgrade_failed = 1;
!                                 last;
  			    }
-                             $bucket_total  += $2;
-                             $bucket_unique += 1;
                          } else {
                              $self->log_( "Found entry in corpus for $bucket that looks wrong: \"$_\" (ignoring)" );
--- 759,771 ----
  
                          if ( /^([^\s]+) (\d+)$/ ) {
!   			    if ( $2 != 0 ) {
!     			        if ( $self->get_base_value_( $bucket, $1 ) != $2 ) {
!                                     print "\nUpgrade error for word $1 in bucket $bucket.\nShutdown POPFile and rerun.\n";
!                                     $upgrade_failed = 1;
!                                     last;
! 			        }
!                                 $bucket_total  += $2;
!                                 $bucket_unique += 1;
  			    }
                          } else {
                              $self->log_( "Found entry in corpus for $bucket that looks wrong: \"$_\" (ignoring)" );