view release on metacpan or  search on metacpan

lib/HTML/DOM.pm  view on Meta::CPAN

		                     # sidering what        # dler's view
		                   # _tweak_~text does       # of the tree

		# Web browsers preserve whitespace, at least from the point
		# of view of the DOM; but the main reason we are using this
		# option is that  a  parser  for  innerHTML  doesn’t  know
		# whether the nodes will be inserted in a <pre>.
		no_space_compacting $tb 1;

		$tb->handler(text => "text",         # so we can get line
		    "self, text, is_cdata, offset"); # numbers for scripts
		$tb->handler(start => "start",
		  "self, tagname, attr, attrseq, offset, tokenpos");
		$tb->handler((declaration=>)x2,'self,tagname,tokens,text');

		$tb->{_HTML_DOM_tweakall} = $tb->{'_tweak_*'};

		my %opts = @_;
		$tb->{_HTML_DOM_no_mg} = delete $opts{no_magic_forms};
		  # used by an element’s innerHTML

lib/HTML/DOM.pm  view on Meta::CPAN

		  start_h => [ 
		    sub { $doc_elem->start(@_) },
		   'tagname, attr, attrseq'
		  ],
		  end_h => [ 
		    sub { $doc_elem->end(@_) },
		   'tagname, text'
		  ],
		  text_h => [ 
		    sub { $doc_elem->text(@_) },
		   'text, is_cdata'
		  ],
		 );

		$p->unbroken_text(1); # push_content, which is called by
		                     # H:TB:text, won't concatenate two
		                   # text portions if the  first  one
		                  # is a node.

		$p->parse(shift);

lib/HTML/DOM/_TreeBuilder.pm  view on Meta::CPAN

}

#==========================================================================
{
    my ( $indent, $nugget );

    sub text {
        return if $_[0]{'_stunted'};

        # Accept a "here's a text token" signal from HTML::Parser.
        my ( $self, $text, $is_cdata ) = @_;

        # the >3.0 versions of Parser may pass a cdata node.
        # Thanks to Gisle Aas for pointing this out.

        return unless length $text;    # I guess that's always right

        my $ignore_text         = $self->{'_ignore_text'};
        my $no_space_compacting = $self->{'_no_space_compacting'};
        my $no_expand_entities  = $self->{'_no_expand_entities'};
        my $pos                 = $self->{'_pos'} || $self;

        HTML::Entities::decode($text)
            unless $ignore_text
                || $is_cdata
                || $HTML::Tagset::isCDATA_Parent{ $pos->{'_tag'} }
                || $no_expand_entities;

        #my($indent, $nugget);
        if (DEBUG) {

           # optimization -- don't figure out depth unless we're in debug mode
            my @lineage_tags = $pos->lineage_tag_names;
            $indent = '  ' x ( 1 + @lineage_tags );