package XHTML::Enpara;
use strict;
use warnings;
use HTML::Tagset 3.02 ();
use Carp;
use Scalar::Util "blessed";

my $isKnown = \%HTML::Tagset::isKnown;
my $canTighten = \%HTML::Tagset::canTighten;
my $isHeadElement = \%HTML::Tagset::isHeadElement;
my $isBodyElement = \%HTML::Tagset::isBodyElement;
my $isPhraseMarkup = \%HTML::Tagset::isPhraseMarkup;
my $isHeadOrBodyElement = \%HTML::Tagset::isHeadOrBodyElement;
my $isList = \%HTML::Tagset::isList;
my $isTableElement = \%HTML::Tagset::isTableElement;
my $isFormElement = \%HTML::Tagset::isFormElement;
my $p_closure_barriers = \@HTML::Tagset::p_closure_barriers;

my $isBlockLevel = { map {; $_ => 1 }
                     grep { ! ( $isPhraseMarkup->{$_} || $isFormElement->{$_} ) }
                     keys %{$isBodyElement}
                 };


sub enpara {
    my ( $c, $root ) = @_;

#    blessed($root) 
# check for documentElement or ownerDoc and proceed from there.

    $root->normalize;
    my $doc = $root->getOwnerDocument;

#    my @enpara_nodes = $root->childNodes;
#    push @enpara_nodes, $root->findnodes('//[@class]');
    for my $named_enpara ( $root->findnodes('*[@class]') )
    {
        next unless $isBlockLevel->{$named_enpara->nodeName};
        next unless $named_enpara->getAttribute("class") =~ /\benpara\b/;
        _enpara_this_nodes_content($named_enpara, $doc);
    }
    _enpara_this_nodes_content($root, $doc);
}

sub _enpara_this_nodes_content {
    my ( $parent, $doc ) = @_;

    my $lastChild = $parent->lastChild;
    my @naked_block;
    for my $node ( $parent->childNodes )
    {
        if ( $isBlockLevel->{$node->nodeName}
             or
             $node->nodeName eq "a" # special case block level, so IGNORE
             and
             grep { $_->nodeName eq "img" } $node->childNodes
             )
        {
            next unless @naked_block; # nothing to enblock
            my $p = $doc->createElement("p");
            $p->setAttribute("enpara","enpara");
            $p->appendChild($_) for @naked_block;
            $parent->insertBefore( $p, $node )
                if $p->textContent =~ /\S/;
            @naked_block = ();
        }
        elsif ( $node->nodeName eq "#text"
                and
                $node->nodeValue =~ /(?:[^\S\n]*\n){2,}/
                )
        {
            my $text = $node->nodeValue;

            my @text_part = map { $doc->createTextNode($_) }
                split /([^\S\n]*\n){2,}/, $text;

            my @new_node;
            for ( my $x = 0; $x < @text_part; $x++ )
            {
                if ( $text_part[$x]->nodeValue =~ /\S/ )
                {
                    push @naked_block, $text_part[$x];
                }
                else # it's a blank newline node so _STOP_
                {
                    next unless @naked_block;
                    my $p = $doc->createElement("p");
                    $p->setAttribute("enpara","enpara");
                    $p->appendChild($_) for @naked_block;
                    @naked_block = ();
                    push @new_node, $p;
                }
            }
            if ( @new_node )
            {
                $parent->insertAfter($new_node[0], $node);
                for ( my $x = 1; $x < @new_node; $x++ )
                {
                    $parent->insertAfter($new_node[$x], $new_node[$x-1]);
                }
            }
            $node->unbindNode;
        }
        else
        {
            push @naked_block, $node; # if $node->nodeValue =~ /\S/;
        }

        if ( $node->isSameNode( $lastChild )
             and @naked_block )
        {
            my $p = $doc->createElement("p");
            $p->setAttribute("enpara","enpara");
            $p->appendChild($_) for ( @naked_block );
            $parent->appendChild($p) if $p->textContent =~ /\S/;
        }
    }

    my $newline = $doc->createTextNode("\n");
    my $br = $doc->createElement("br");

    for my $p ( $parent->findnodes('//p[@enpara="enpara"]') )
    {
        $p->removeAttribute("enpara");
        $parent->insertBefore( $newline->cloneNode, $p );
        $parent->insertAfter( $newline->cloneNode, $p );

        my $frag = $doc->createDocumentFragment();

        my @kids = $p->childNodes();
        for ( my $i = 0; $i < @kids; $i++ )
        {
            my $kid = $kids[$i];
            next unless $kid->nodeName eq "#text";
            my $text = $kid->nodeValue;
            $text =~ s/\A\n// if $i == 0;
            $text =~ s/\n\z// if $i == $#kids;

            my @lines = map { $doc->createTextNode($_) }
                split /(\n)/, $text;

            for ( my $i = 0; $i < @lines; $i++ )
            {
                $frag->appendChild($lines[$i]);
                unless ( $i == $#lines
                         or
                         $lines[$i]->nodeValue eq "\n" )
                {
                    $frag->appendChild($br->cloneNode);
                }
            }
            $kid->replaceNode($frag);
        }
    }
}


1;


__END__
typedef enum {
    XML_ELEMENT_NODE=           1,
    XML_ATTRIBUTE_NODE=         2,
    XML_TEXT_NODE=              3,
    XML_CDATA_SECTION_NODE=     4,
    XML_ENTITY_REF_NODE=        5,
    XML_ENTITY_NODE=            6,
    XML_PI_NODE=                7,
    XML_COMMENT_NODE=           8,
    XML_DOCUMENT_NODE=          9,
    XML_DOCUMENT_TYPE_NODE=     10,
    XML_DOCUMENT_FRAG_NODE=     11,
    XML_NOTATION_NODE=          12,
    XML_HTML_DOCUMENT_NODE=     13,
    XML_DTD_NODE=               14,
    XML_ELEMENT_DECL=           15,
    XML_ATTRIBUTE_DECL=         16,
    XML_ENTITY_DECL=            17,
    XML_NAMESPACE_DECL=         18,
    XML_XINCLUDE_START=         19,
    XML_XINCLUDE_END=           20
#ifdef LIBXML_DOCB_ENABLED
   ,XML_DOCB_DOCUMENT_NODE=     21
#endif
} xmlElementType;

my @block_level = qw( blockquote fieldset address noscript iframe
                      object param script table tbody thead tfoot form
                      div map pre dl h1 h2 h3 h4 h5 h6 hr ol ul dd dt
                      li td th tr p colgroup applet multicol xmp
                      listing area center dir col menu del ins nolayer
                      isindex plaintext caption bgsound ilayer legend
                      );

my %block_level = map { $_ => 1 } @block_level;


is_deeply($isBlockLevel, \%block_level);

use Test::More tests => 1;

XML::LibXML based only at first because it's easier.

actual markup
remove_markup("leaving content")

entire Nodes
remove_tags("

enpara

translate_tags

traverse("/*") -> callback

strip_styles(* or [list])
strip_attributes()

inline_stylesheets(names/paths)

fragment_to_xhtml


We WILL NOT be covering other well known and well done implementations like HTML::Entities or URI::Escape

   use Rose::HTML::Util qw(:all);

   $esc = escape_html($str);
   $str = unescape_html($esc);

   $esc = escape_uri($str);
   $str = unescape_uri($esc);

   $comp = escape_uri_component($str);

   $esc = encode_entities($str);