[popfile-commit] engine/Classifier MailParse.pm,1.91.2.1,1.91.2.2

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

Update of /cvsroot/popfile/engine/Classifier
In directory sc8-pr-cvs1:/tmp/cvs-serv3846

Modified Files:
      Tag: v0/18/1
	MailParse.pm 
Log Message:
fix bug with protocol-less href's

Index: MailParse.pm
===================================================================
RCS file: /cvsroot/popfile/engine/Classifier/MailParse.pm,v
retrieving revision 1.91.2.1
retrieving revision 1.91.2.2
diff -C2 -d -r1.91.2.1 -r1.91.2.2
*** MailParse.pm	11 Mar 2003 08:08:00 -0000	1.91.2.1
--- MailParse.pm	10 Apr 2003 22:11:23 -0000	1.91.2.2
***************
*** 379,399 ****
          
          # Tags with href attributes
! 
          if ( $attribute =~ /^href$/i && $tag =~ /^(a|link|base|area)$/i )  {
  
!             # ftp, http, https
  
!             if ( $value =~ /^(ftp|http|https):\/\//i ) {
!                 add_url($self, $value, $encoded, $quote, $end_quote, '');
!                 next;
              }
-     
-             # The less common mailto: goes second, and we only care if this is in an anchor
  
!             if ( $tag =~ /^a$/ && $value =~ /^mailto:([[:alpha:]0-9\-_\.]+?@([[:alpha:]0-9\-_\.]+?))([>\&\?\:\/]|$)/i )  {
!                update_word( $self, $1, $encoded, 'mailto:', ($3?'[\\\>\&\?\:\/]':$end_quote), '' );
!                add_url( $self, $2, $encoded, '@', ($3?'[\\\&\?\:\/]':$end_quote), '' );
!             }
!             next;
          }
          
--- 379,399 ----
          
          # Tags with href attributes
!         
          if ( $attribute =~ /^href$/i && $tag =~ /^(a|link|base|area)$/i )  {
  
!             # Look for mailto:'s
  
!             if ($value =~ /^mailto:/i) {
!                 if ( $tag =~ /^a$/ && $value =~ /^mailto:([[:alpha:]0-9\-_\.]+?@([[:alpha:]0-9\-_\.]+?))([>\&\?\:\/]|$)/i )  {
!                    update_word( $self, $1, $encoded, 'mailto:', ($3?'[\\\>\&\?\:\/]':$end_quote), '' );
!                    add_url( $self, $2, $encoded, '@', ($3?'[\\\&\?\:\/]':$end_quote), '' );
!                 }
!             } else {
!                 # Anything that isn't a mailto is probably an URL
!                 
!                 $self->add_url($value, $encoded, $quote, $end_quote, '');
              }
  
!             next;            
          }