[popfile-commit] engine/Classifier MailParse.pm,1.118,1.119

[<ssc...@us...>]

Update of /cvsroot/popfile/engine/Classifier
In directory sc8-pr-cvs1:/tmp/cvs-serv21009

Modified Files:
	MailParse.pm 
Log Message:
protect CID: URL's from identification as off-site images

Index: MailParse.pm
===================================================================
RCS file: /cvsroot/popfile/engine/Classifier/MailParse.pm,v
retrieving revision 1.118
retrieving revision 1.119
diff -C2 -d -r1.118 -r1.119
*** MailParse.pm	27 Apr 2003 00:20:51 -0000	1.118
--- MailParse.pm	27 Apr 2003 02:57:12 -0000	1.119
***************
*** 408,422 ****
               ( ( $tag =~ /^img|frame|iframe$/i )
                 || ( $tag =~ /^script$/i && $parse_script_uri ) ) ) {
!             my $host = add_url( $self, $value, $encoded, $quote, $end_quote, '' );
  
! 	    # If the host name is not blank (i.e. there was a hostname in the url
! 	    # and it was an image, then if the host was not this host then report
! 	    # an off machine image
  
!             if ( ( $host ne '' ) && ( $tag =~ /^img$/i ) ) {
! 	        if ( $host ne 'localhost' ) {
!                     $self->update_pseudoword( 'html', 'imgremotesrc' );
! 	        }
! 	    }
  
              next;
--- 408,435 ----
               ( ( $tag =~ /^img|frame|iframe$/i )
                 || ( $tag =~ /^script$/i && $parse_script_uri ) ) ) {
!                 
!             # "CID:" links refer to an origin-controlled attachment to a html email.
!             # Adding strings from these, even if they appear to be hostnames, may or
!             # may not be beneficial
  
!             if ($value =~ /^cid\:/i )
!             {
!                 # TODO: Decide what to do here, ignoring CID's for now
!                 
!             } else {
  
!                 my $host = add_url( $self, $value, $encoded, $quote, $end_quote, '' );
! 
!                 # If the host name is not blank (i.e. there was a hostname in the url
!                 # and it was an image, then if the host was not this host then report
!                 # an off machine image
! 
!                 if ( ( $host ne '' ) && ( $tag =~ /^img$/i ) ) {
!         	        if ( $host ne 'localhost' ) {
!                         $self->update_pseudoword( 'html', 'imgremotesrc' );
!         	        }
!                 }
!             }
!             
  
              next;
***************
*** 485,489 ****
              $attribute = lc( $attribute );
              $self->update_pseudoword( 'html', "img$attribute$value" );
! 	}
  
          # Font sizes
--- 498,502 ----
              $attribute = lc( $attribute );
              $self->update_pseudoword( 'html', "img$attribute$value" );
!         }
  
          # Font sizes