popfile-commit

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

Update of /cvsroot/popfile/engine/Classifier
In directory sc8-pr-cvs1:/tmp/cvs-serv16302/Classifier

Modified Files:
	MailParse.pm 
Log Message:
Improve the handling of the words found in HTML once HTML tags are removed

Index: MailParse.pm
===================================================================
RCS file: /cvsroot/popfile/engine/Classifier/MailParse.pm,v
retrieving revision 1.162
retrieving revision 1.163
diff -C2 -d -r1.162 -r1.163
*** MailParse.pm	22 Sep 2003 13:27:21 -0000	1.162
--- MailParse.pm	24 Sep 2003 15:50:58 -0000	1.163
***************
*** 978,981 ****
--- 978,982 ----
  
      my $found = 1;
+     my $non_html = '';
  
      $line =~ s/[\r\n]+//gm;
***************
*** 994,998 ****
      # recognized.
  
!     # FIXME: This also removes tags in plain text emails so a sentence
      # such as 'To run the program type "program <filename>".' is also
      # effected.  The correct fix seams to be to look at the
--- 995,999 ----
      # recognized.
  
!     # TODO: This also removes tags in plain text emails so a sentence
      # such as 'To run the program type "program <filename>".' is also
      # effected.  The correct fix seams to be to look at the
***************
*** 1006,1011 ****
      # Remove pairs of non-spacing tags without content such as <b></b>
      # and also <b><i></i></b>.
!     
!     # FIXME: What about combined open and close tags such as <b />?
  
      while ( $line =~s/(<($non_spacing_tags)(?:\s+[^>]*?)?><\/\2>)//io ) {
--- 1007,1012 ----
      # Remove pairs of non-spacing tags without content such as <b></b>
      # and also <b><i></i></b>.
! 
!     # TODO: What about combined open and close tags such as <b />?
  
      while ( $line =~s/(<($non_spacing_tags)(?:\s+[^>]*?)?><\/\2>)//io ) {
***************
*** 1033,1036 ****
--- 1034,1038 ----
              } else {
                  $self->{html_arg__} .= $line;
+                 $self->add_line( $non_html, $encoded, '' );
                  return 1;
              }
***************
*** 1055,1058 ****
--- 1057,1061 ----
              $self->{html_arg__}    = $3;
              $self->{in_html_tag__} = 1;
+             $self->add_line( $non_html, $encoded, '' );
              return 1;
          }
***************
*** 1064,1070 ****
          if ( $line =~ s/^([^<]+)(<|$)/$2/ ) {
              $found = 1;
!             add_line( $self, $1, $encoded, '' );
          }
      }
  
      return 0;
--- 1067,1075 ----
          if ( $line =~ s/^([^<]+)(<|$)/$2/ ) {
              $found = 1;
!             $non_html .= $1;
          }
      }
+ 
+     $self->add_line( $non_html, $encoded, '' );
  
      return 0;