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/manakai2/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 'cat') {
        $node->parent_node->remove_child ($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	wakaba	1.4	/home/wakaba/work/manakai2/lib];
7	wakaba	1.1
8			my $HTML_NS = q<http://www.w3.org/1999/xhtml>;
9			my $SRC_NS = q<http://suika.fam.cx/~wakaba/archive/2007/wdcc-desc/>;
10			my $XML_NS = q<http://www.w3.org/XML/1998/namespace>;
11
12			require Message::DOM::DOMImplementation;
13	wakaba	1.2	my $dom = Message::DOM::DOMImplementation->new;
14	wakaba	1.1
15			my $doc;
16			{
17			my $source_file_name = shift or die "$0: No source file specified\n";
18			open my $source_file, '<', $source_file_name
19			or die "$0: $source_file_name: $!";
20			require Message::DOM::XMLParserTemp;
21			$doc = Message::DOM::XMLParserTemp->parse_byte_stream
22			($source_file => $dom, undef, charset => 'utf-8');
23	wakaba	1.2	$doc->manakai_is_html (1);
24	wakaba	1.1	}
25
26			my $target_lang = 'en';
27			my @node = (@{$doc->child_nodes});
28			while (@node) {
29			my $node = shift @node;
30			if ($node->node_type == $node->ELEMENT_NODE) {
31			if ($node->namespace_uri eq $HTML_NS) {
32			if ($node->manakai_local_name eq 'title') {
33			unless ($node->get_attribute_ns ($XML_NS, 'lang') eq $target_lang) {
34			$node->parent_node->remove_child ($node);
35			}
36			} else {
37	wakaba	1.2	unshift @node, @{$node->child_nodes};
38	wakaba	1.1	}
39			} elsif ($node->namespace_uri eq $SRC_NS) {
40			if ($node->manakai_local_name eq 'item') {
41			my $message;
42			my $desc;
43			for (@{$node->child_nodes}) {
44			if ($_->node_type == $_->ELEMENT_NODE and
45			$_->namespace_uri eq $SRC_NS) {
46			if ($_->manakai_local_name eq 'desc') {
47			if ($_->get_attribute_ns ($XML_NS, 'lang') eq $target_lang) {
48			$desc = $_;
49			next;
50			} else {
51			$desc \|\|= $_;
52			}
53			} elsif ($_->manakai_local_name eq 'message') {
54			if ($_->get_attribute_ns ($XML_NS, 'lang') eq $target_lang) {
55			$message = $_;
56			next;
57			} else {
58			$message \|\|= $_;
59			}
60			}
61			}
62			}
63
64			my $name = $node->get_attribute_ns (undef, 'name');
65			$name =~ tr/ /-/;
66	wakaba	1.2	my $level = $node->get_attribute_ns (undef, 'level');
67			$name = $level . ':' . $name if defined $level;
68	wakaba	1.1	my $section = $doc->create_element_ns ($HTML_NS, 'div');
69			$section->set_attribute_ns
70			(undef, class => 'section ' .
71			$node->get_attribute_ns (undef, 'class'));
72			$section->set_attribute_ns (undef, id => $name);
73
74			my @message_child = @{$message->child_nodes};
75			my $msg = $section->append_child
76			($doc->create_element_ns ($HTML_NS, 'h3'));
77			$msg->append_child ($_) for @message_child;
78
79	wakaba	1.2	if ($desc) {
80			my @desc_child = @{$desc->child_nodes};
81			$section->append_child ($_) for @desc_child;
82			}
83	wakaba	1.1
84			$node->parent_node->insert_before ($section, $node);
85			$node->parent_node->remove_child ($node); ## TODO: replace_child is not yet implemented
86	wakaba	1.5	} elsif ($node->manakai_local_name eq 'cat') {
87			$node->parent_node->remove_child ($node);
88	wakaba	1.1	} else {
89			warn "$0: ", $node->manakai_local_name, " is not supported\n";
90			}
91			}
92			}
93			}
94			$doc->document_element->set_attribute_ns (undef, lang => $target_lang);
95
96	wakaba	1.2	print $doc->inner_html;