view release on metacpan or  search on metacpan

lib/XML/Smart/HTMLParser.pm  view on Meta::CPAN

#############################################################################
## Name:        HTMLParser.pm
## Purpose:     XML::Smart::HTMLParser
## Author:      Graciliano M. P.
## Modified by: Harish Madabushi
## Created:     29/05/2003
## RCS-ID:      
## Copyright:   (c) 2003 Graciliano M. P.
## Licence:     This program is free software; you can redistribute it and/or
##              modify it under the same terms as Perl itself
#############################################################################

package XML::Smart::HTMLParser                                 ;

use 5.006                                                      ;

use strict                                                     ;
use warnings                                                   ;

use XML::Smart::Shared qw( _unset_sig_warn _reset_sig_warn )   ;

our ($VERSION , @ISA) ;
$VERSION = '1.12' ;

#######
# NEW #
#######

sub new { 
  my $this = shift ;
  my $class = ref($this) || $this ;
  return $this if ref $this ;

  $this = bless {} => $class ;
  
  _unset_sig_warn() ;
  my %args = @_ ;
  _reset_sig_warn() ;
  $this->setHandlers(%args) ;
  
  $this->{NOENTITY} = 1 ;
  
  return $this ;
}

###############
# SETHANDLERS #
###############

sub setHandlers {
  my $this = shift ;
  _unset_sig_warn() ;
  my %args = @_;
  _reset_sig_warn() ;
    
  $this->{Init}  = $args{Init} || sub{} ;
  $this->{Start} = $args{Start} || sub{} ;
  $this->{Char}  = $args{Char} || sub{} ;
  $this->{End}   = $args{End} || sub{} ;
  $this->{Final} = $args{Final} || sub{} ;
  
  return( 1 ) ;

}

#########
# PARSE #
#########

sub parse {

    my $this = shift ;
    my $data = shift ;
    
    $data =~ s/\r\n?/\n/gs ;
    
    $data =~ s/^\s*<\?xml.*?>//gsi ;
    
    my @parsed ;
    
    while( $data =~ /(.*?)<(.*?)>/gsi ) {

	my $cont = $1 ;
	my $markup = $2 ;
	
	my ( $more_q , @args ) = &parse_tag( $markup ) ;
	
	while( $more_q ) {
	    my $more ;
	    ( $more ) = ( $data =~ /\G(.*?)>/s ) ;
	    pos( $data ) += length( $more ) + 1 ;
	    $markup = $markup.'>'.$more ;
	    ( $more_q , @args ) = &parse_tag( $markup ) ;
	}