test/html-webhacc/mkdescription.pl

#!/usr/bin/perl
use strict;
use encoding 'us-ascii', STDOUT => 'utf-8';

use lib qw[/home/httpd/html/www/markup/html/whatpm
           /home/wakaba/work/manakai/lib
           /home/wakaba/public_html/-temp/wiki/lib];

my $HTML_NS = q<http://www.w3.org/1999/xhtml>;
my $SRC_NS = q<http://suika.fam.cx/~wakaba/archive/2007/wdcc-desc/>;
my $XML_NS = q<http://www.w3.org/XML/1998/namespace>;

require Message::DOM::DOMImplementation;
my $dom = Message::DOM::DOMImplementation->new;

my $doc;
{
  my $source_file_name = shift or die "$0: No source file specified\n";
  open my $source_file, '<', $source_file_name
      or die "$0: $source_file_name: $!";
  require Message::DOM::XMLParserTemp;
  $doc = Message::DOM::XMLParserTemp->parse_byte_stream
      ($source_file => $dom, undef, charset => 'utf-8');
  $doc->manakai_is_html (1);
}

my $target_lang = 'en';
my @node = (@{$doc->child_nodes});
while (@node) {
  my $node = shift @node;
  if ($node->node_type == $node->ELEMENT_NODE) {
    if ($node->namespace_uri eq $HTML_NS) {
      if ($node->manakai_local_name eq 'title') {
        unless ($node->get_attribute_ns ($XML_NS, 'lang') eq $target_lang) {
          $node->parent_node->remove_child ($node);
        }
      } else {
        unshift @node, @{$node->child_nodes};
      }
    } elsif ($node->namespace_uri eq $SRC_NS) {
      if ($node->manakai_local_name eq 'item') {
        my $message;
        my $desc;
        for (@{$node->child_nodes}) {
          if ($_->node_type == $_->ELEMENT_NODE and
              $_->namespace_uri eq $SRC_NS) {
            if ($_->manakai_local_name eq 'desc') {
              if ($_->get_attribute_ns ($XML_NS, 'lang') eq $target_lang) {
                $desc = $_;
                next;
              } else {
                $desc ||= $_;
              }
            } elsif ($_->manakai_local_name eq 'message') {
              if ($_->get_attribute_ns ($XML_NS, 'lang') eq $target_lang) {
                $message = $_;
                next;
              } else {
                $message ||= $_;
              }
            }
          }
        }

        my $name = $node->get_attribute_ns (undef, 'name');
        $name =~ tr/ /-/;
        my $level = $node->get_attribute_ns (undef, 'level');
        $name = $level . ':' . $name if defined $level;
        my $section = $doc->create_element_ns ($HTML_NS, 'div');
        $section->set_attribute_ns
            (undef, class => 'section ' .
                 $node->get_attribute_ns (undef, 'class'));
        $section->set_attribute_ns (undef, id => $name);

        my @message_child = @{$message->child_nodes};
        my $msg = $section->append_child
            ($doc->create_element_ns ($HTML_NS, 'h3'));
        $msg->append_child ($_) for @message_child;

        if ($desc) {
          my @desc_child = @{$desc->child_nodes};
          $section->append_child ($_) for @desc_child;
        }

        $node->parent_node->insert_before ($section, $node);
        $node->parent_node->remove_child ($node); ## TODO: replace_child is not yet implemented
      } elsif ($node->manakai_local_name eq 'catalog') {
        $node->parent_node->remove_child ($node); 
      } else {
        warn "$0: ", $node->manakai_local_name, " is not supported\n";
      }
    }
  }
}
$doc->document_element->set_attribute_ns (undef, lang => $target_lang);

print $doc->inner_html;
1	wakaba	1.1	#!/usr/bin/perl
2			use strict;
3			use encoding 'us-ascii', STDOUT => 'utf-8';
4
5			use lib qw[/home/httpd/html/www/markup/html/whatpm
6			/home/wakaba/work/manakai/lib
7			/home/wakaba/public_html/-temp/wiki/lib];
8
9			my $HTML_NS = q<http://www.w3.org/1999/xhtml>;
10			my $SRC_NS = q<http://suika.fam.cx/~wakaba/archive/2007/wdcc-desc/>;
11			my $XML_NS = q<http://www.w3.org/XML/1998/namespace>;
12
13			require Message::DOM::DOMImplementation;
14	wakaba	1.2	my $dom = Message::DOM::DOMImplementation->new;
15	wakaba	1.1
16			my $doc;
17			{
18			my $source_file_name = shift or die "$0: No source file specified\n";
19			open my $source_file, '<', $source_file_name
20			or die "$0: $source_file_name: $!";
21			require Message::DOM::XMLParserTemp;
22			$doc = Message::DOM::XMLParserTemp->parse_byte_stream
23			($source_file => $dom, undef, charset => 'utf-8');
24	wakaba	1.2	$doc->manakai_is_html (1);
25	wakaba	1.1	}
26
27			my $target_lang = 'en';
28			my @node = (@{$doc->child_nodes});
29			while (@node) {
30			my $node = shift @node;
31			if ($node->node_type == $node->ELEMENT_NODE) {
32			if ($node->namespace_uri eq $HTML_NS) {
33			if ($node->manakai_local_name eq 'title') {
34			unless ($node->get_attribute_ns ($XML_NS, 'lang') eq $target_lang) {
35			$node->parent_node->remove_child ($node);
36			}
37			} else {
38	wakaba	1.2	unshift @node, @{$node->child_nodes};
39	wakaba	1.1	}
40			} elsif ($node->namespace_uri eq $SRC_NS) {
41			if ($node->manakai_local_name eq 'item') {
42			my $message;
43			my $desc;
44			for (@{$node->child_nodes}) {
45			if ($_->node_type == $_->ELEMENT_NODE and
46			$_->namespace_uri eq $SRC_NS) {
47			if ($_->manakai_local_name eq 'desc') {
48			if ($_->get_attribute_ns ($XML_NS, 'lang') eq $target_lang) {
49			$desc = $_;
50			next;
51			} else {
52			$desc \|\|= $_;
53			}
54			} elsif ($_->manakai_local_name eq 'message') {
55			if ($_->get_attribute_ns ($XML_NS, 'lang') eq $target_lang) {
56			$message = $_;
57			next;
58			} else {
59			$message \|\|= $_;
60			}
61			}
62			}
63			}
64
65			my $name = $node->get_attribute_ns (undef, 'name');
66			$name =~ tr/ /-/;
67	wakaba	1.2	my $level = $node->get_attribute_ns (undef, 'level');
68			$name = $level . ':' . $name if defined $level;
69	wakaba	1.1	my $section = $doc->create_element_ns ($HTML_NS, 'div');
70			$section->set_attribute_ns
71			(undef, class => 'section ' .
72			$node->get_attribute_ns (undef, 'class'));
73			$section->set_attribute_ns (undef, id => $name);
74
75			my @message_child = @{$message->child_nodes};
76			my $msg = $section->append_child
77			($doc->create_element_ns ($HTML_NS, 'h3'));
78			$msg->append_child ($_) for @message_child;
79
80	wakaba	1.2	if ($desc) {
81			my @desc_child = @{$desc->child_nodes};
82			$section->append_child ($_) for @desc_child;
83			}
84	wakaba	1.1
85			$node->parent_node->insert_before ($section, $node);
86			$node->parent_node->remove_child ($node); ## TODO: replace_child is not yet implemented
87	wakaba	1.2	} elsif ($node->manakai_local_name eq 'catalog') {
88	wakaba	1.3	$node->parent_node->remove_child ($node);
89	wakaba	1.1	} else {
90			warn "$0: ", $node->manakai_local_name, " is not supported\n";
91			}
92			}
93			}
94			}
95			$doc->document_element->set_attribute_ns (undef, lang => $target_lang);
96
97	wakaba	1.2	print $doc->inner_html;