html5/spec-ja/make2.pl

#!/usr/bin/perl
use strict;

BEGIN { require 'common.pl' }

my $source_file_name = shift;
my $result_file_name = shift;
my $status_file_name = shift;

my %data;
my %pattern;

my $all = get_all_entries ();
for my $key (keys %$all) {
  for (keys %{$all->{$key}->{exact} or {}}) {
    my $entry = $all->{$key}->{exact}->{$_};
    $data{normalize ($entry->{en})} = [$entry->{ja}, $_];
  }

  for (keys %{$all->{$key}->{pattern} or {}}) {
    my $entry = $all->{$key}->{pattern}->{$_};
    $pattern{create_pattern1 (normalize ($entry->{en}))}
        = [$entry->{ja}, $_];
  }
}

my @pattern = sort {length $b <=> length $a} keys %pattern;

my $source_text;
{
  warn "$source_file_name...\n";
  open my $source_file, '<:utf8', $source_file_name
      or die "$0: $source_file_name: $!";
  local $/ = undef;
  $source_text = <$source_file>;
}

my $part = 'cover';

my $status = {};
my $all_status = {};
my $tbt_added = {};

warn "Generating...\n";
$source_text =~ s{(<(?>p(?>re)?|li|d[td]|t[dh]|h[1-6]|caption)(?>\s[^>]*)?>)((?>(?!</?(?>p(?>re)?|li|d(?>[tdl]|iv)|t(?>[dr]|h(?>ead)?|able|body|foot)|h[1-6r]|ul|ol|caption)(?>\s[^>]*)?>).)+)}
{
  my ($tag, $text) = ($1, $2);
  my $prefix = '';
  if ($text =~ s#^(<span class=secno>[^<>]+</span>)##) {
    $prefix = $1;
  }
  my $n_text = normalize ($text);

  if ($tag =~ /^<h2 id=(\w+)/) {
    $part = $1;
  }
  
  if (length $n_text) {
    my $ja_text = $data{$n_text};
    
    $status->{all}++;
    $all_status->{$part}->{all}++;
    
    if (defined $ja_text) {
      $status->{ja}++;
      $all_status->{$part}->{ja}++;
      add_class ($tag, 'has-ja-translation', $ja_text->[1]) .
      q[<span class=en-original lang=en>] .
      $prefix . 
      escape_id ($text, 'en-') . q[</span>] .
      q[<span class=ja-translation lang=ja>] .
      $prefix . 
      $ja_text->[0] . q[</span>];
    } else {
      my $v;
      for my $pattern (@pattern) {
        if ($n_text =~ /^$pattern$/) {
          $status->{ja}++;
          $all_status->{$part}->{ja}++;
          my $real_hash = get_hash ($n_text);
          $v = add_class ($tag, 'has-ja-translation',
                          $pattern{$pattern}->[1], $real_hash) .
              q[<span class=en-original lang=en>] .
              $prefix . 
              escape_id ($text, 'en-') . q[</span>] .
              q[<span class=ja-translation lang=ja>] .
              $prefix . 
              replace_pattern2 ($pattern{$pattern}->[0], $1, $2, $3, $4, $5) .
              q[</span>];

          unless ($tbt_added->{$n_text}) {
            set_fallback_entry ($real_hash => {en => $text});
            $tbt_added->{$n_text} = 1;
          }
          last;
        }
      }

      unless (defined $v) {
        my $hash = get_hash ($n_text);
        $v = add_class ($tag, 'no-ja-translation', $hash) .
            '<span class=en-original lang=en>' .
            $prefix . $text .
            '</span>';
        
        unless ($tbt_added->{$n_text}) {
          set_fallback_entry ($hash => {en => $text});
          $tbt_added->{$n_text} = 1;
        }
      }

      $v;
    }
  } else {
    $1 . $2;
  }
}ges;
$source_text =~ s{(<((?>link|img|script))\s[^>]+>)}{
  my $tag = $1;
  my $tag_name = $2;
  my $n_text = normalize ($tag);
  my $ja_text = $data{$n_text};
  if (defined $ja_text) {
    $ja_text->[0];
  } else {
    unless ($tbt_added->{$n_text}) {
      set_fallback_entry (scalar (get_hash ($n_text)),
                          {en => $tag, tags => [$tag_name . '-tag']});
      $tbt_added->{$n_text} = 1;
    }
    $tag;
  }
}ges;

$source_text =~ s{\[\[([A-Z ]+):([^]]+)\]\]}
{<em class=rfc2119 title="$1">$2</em>}gs;

#$source_text =~ s[<title>][<base href="http://www.whatwg.org/specs/web-apps/current-work/"><title>];

{
  warn "$result_file_name...\n";
  open my $result_file, '>:utf8', $result_file_name
      or die "$0: $result_file_name: $!";
  print $result_file $source_text;
}

{
  my $time = time;
  my @item = ($time, $status->{ja}, $status->{all});
  for my $part (qw(cover introduction infrastructure dom semantics browsers
                   editing comms syntax rendering no)) {
    push @item, $all_status->{$part}->{ja};
    push @item, $all_status->{$part}->{all};
  }

  open my $status_file, '>>', $status_file_name
      or die "$0: $status_file_name: $!";
  print $status_file join "\t", @item;
  print $status_file "\n";
  close $status_file;
}

save_fallback_entries ();

sub add_class ($$$;$) {
  my $tag = shift;
  my $new_class = shift; # should not contain bare & and bare "
  my $hash = shift;
  my $real_hash = shift;
  $real_hash = qq[ data-ja-real-hash="$real_hash"] if defined $real_hash;

  if ($tag =~ /^<li\b/) {
    ## NOTE: This |p| wrapper is necessary, otherwise, if |li| element
    ## is set to |display: table|, then no list marker is shown.
    $tag .= qq[<p class="$new_class ja-translation-inserted" data-ja-hash="$hash"$real_hash>];
  } elsif ($tag =~ /\bclass="/) {
    $tag =~ s/\bclass="([^"]*)"/class="$1 $new_class" data-ja-hash="$hash"$real_hash/;
  } elsif ($tag =~ /\bclass=/) {
    $tag =~ s/\bclass=([^\s>]+)/class="$1 $new_class" data-ja-hash="$hash"$real_hash/g;
  } else {
    $tag =~ s/>/ class="$new_class" data-ja-hash="$hash"$real_hash>/;
  }
  
  return $tag;
} # add_class

sub escape_id ($$) {
  my $content = shift;
  my $id_prefix = shift; # should not contain bare & and bare "
  
  $content =~ s{<([a-zA-Z0-9-][^<>]+)>}{
    my $tag_content = $1;

    if ($tag_content =~ /\bid="/) {
      $tag_content =~ s/\bid="([^"]*)"/id="$id_prefix$1"/;
    } elsif ($tag_content =~ /\bid=/) {
      $tag_content =~ s/\bid=(\S+)/id="$id_prefix$1"/;
    }

#    if ($tag_content =~ /\bhref=#/) {
#      $tag_content =~ s/\bhref=#(\S+)/href=#$id_prefix$1/;
#    }

    '<' . $tag_content . '>';
  }ge;

  return $content;
} # escape_id

## Author: Wakaba <w@suika.fam.cx>.
## License: Copyright 2008 Wakaba.  You are granted a license to use,
##     reproduce and create derivative works of this script.
## $Date: 2008/10/27 04:52:39 $
1	wakaba	1.1	#!/usr/bin/perl
2			use strict;
3
4			BEGIN { require 'common.pl' }
5
6			my $source_file_name = shift;
7			my $result_file_name = shift;
8			my $status_file_name = shift;
9
10			my %data;
11			my %pattern;
12
13			my $all = get_all_entries ();
14			for my $key (keys %$all) {
15			for (keys %{$all->{$key}->{exact} or {}}) {
16			my $entry = $all->{$key}->{exact}->{$_};
17			$data{normalize ($entry->{en})} = [$entry->{ja}, $_];
18			}
19
20			for (keys %{$all->{$key}->{pattern} or {}}) {
21			my $entry = $all->{$key}->{pattern}->{$_};
22			$pattern{create_pattern1 (normalize ($entry->{en}))}
23			= [$entry->{ja}, $_];
24			}
25			}
26
27			my @pattern = sort {length $b <=> length $a} keys %pattern;
28
29			my $source_text;
30			{
31			warn "$source_file_name...\n";
32			open my $source_file, '<:utf8', $source_file_name
33			or die "$0: $source_file_name: $!";
34			local $/ = undef;
35			$source_text = <$source_file>;
36			}
37
38			my $part = 'cover';
39
40			my $status = {};
41			my $all_status = {};
42			my $tbt_added = {};
43
44			warn "Generating...\n";
45	wakaba	1.4	$source_text =~ s{(<(?>p(?>re)?\|li\|d[td]\|t[dh]\|h[1-6]\|caption)(?>\s[^>])?>)((?>(?!</?(?>p(?>re)?\|li\|d(?>[tdl]\|iv)\|t(?>[dr]\|h(?>ead)?\|able\|body\|foot)\|h[1-6r]\|ul\|ol\|caption)(?>\s[^>])?>).)+)}
46	wakaba	1.1	{
47			my ($tag, $text) = ($1, $2);
48			my $prefix = '';
49			if ($text =~ s#^(<span class=secno>[^<>]+</span>)##) {
50			$prefix = $1;
51			}
52			my $n_text = normalize ($text);
53
54			if ($tag =~ /^<h2 id=(\w+)/) {
55			$part = $1;
56			}
57
58			if (length $n_text) {
59			my $ja_text = $data{$n_text};
60
61			$status->{all}++;
62			$all_status->{$part}->{all}++;
63
64			if (defined $ja_text) {
65			$status->{ja}++;
66			$all_status->{$part}->{ja}++;
67			add_class ($tag, 'has-ja-translation', $ja_text->[1]) .
68			q[<span class=en-original lang=en>] .
69			$prefix .
70			escape_id ($text, 'en-') . q[</span>] .
71			q[<span class=ja-translation lang=ja>] .
72			$prefix .
73			$ja_text->[0] . q[</span>];
74			} else {
75			my $v;
76			for my $pattern (@pattern) {
77			if ($n_text =~ /^$pattern$/) {
78			$status->{ja}++;
79			$all_status->{$part}->{ja}++;
80	wakaba	1.2	my $real_hash = get_hash ($n_text);
81	wakaba	1.1	$v = add_class ($tag, 'has-ja-translation',
82	wakaba	1.2	$pattern{$pattern}->[1], $real_hash) .
83	wakaba	1.1	q[<span class=en-original lang=en>] .
84			$prefix .
85			escape_id ($text, 'en-') . q[</span>] .
86			q[<span class=ja-translation lang=ja>] .
87			$prefix .
88			replace_pattern2 ($pattern{$pattern}->[0], $1, $2, $3, $4, $5) .
89			q[</span>];
90	wakaba	1.2
91			unless ($tbt_added->{$n_text}) {
92			set_fallback_entry ($real_hash => {en => $text});
93			$tbt_added->{$n_text} = 1;
94			}
95	wakaba	1.1	last;
96			}
97			}
98
99			unless (defined $v) {
100			my $hash = get_hash ($n_text);
101	wakaba	1.2	$v = add_class ($tag, 'no-ja-translation', $hash) .
102			'<span class=en-original lang=en>' .
103			$prefix . $text .
104			'</span>';
105	wakaba	1.1
106	wakaba	1.2	unless ($tbt_added->{$n_text}) {
107	wakaba	1.1	set_fallback_entry ($hash => {en => $text});
108	wakaba	1.2	$tbt_added->{$n_text} = 1;
109	wakaba	1.1	}
110			}
111
112			$v;
113			}
114			} else {
115			$1 . $2;
116			}
117			}ges;
118	wakaba	1.5	$source_text =~ s{(<((?>link\|img\|script))\s[^>]+>)}{
119	wakaba	1.1	my $tag = $1;
120	wakaba	1.5	my $tag_name = $2;
121	wakaba	1.1	my $n_text = normalize ($tag);
122			my $ja_text = $data{$n_text};
123			if (defined $ja_text) {
124			$ja_text->[0];
125			} else {
126	wakaba	1.5	unless ($tbt_added->{$n_text}) {
127			set_fallback_entry (scalar (get_hash ($n_text)),
128			{en => $tag, tags => [$tag_name . '-tag']});
129			$tbt_added->{$n_text} = 1;
130			}
131	wakaba	1.1	$tag;
132			}
133			}ges;
134
135			$source_text =~ s{\[\[([A-Z ]+):([^]]+)\]\]}
136			{<em class=rfc2119 title="$1">$2</em>}gs;
137
138			#$source_text =~ s[<title>][<base href="http://www.whatwg.org/specs/web-apps/current-work/"><title>];
139
140			{
141			warn "$result_file_name...\n";
142			open my $result_file, '>:utf8', $result_file_name
143			or die "$0: $result_file_name: $!";
144			print $result_file $source_text;
145			}
146
147			{
148			my $time = time;
149			my @item = ($time, $status->{ja}, $status->{all});
150			for my $part (qw(cover introduction infrastructure dom semantics browsers
151			editing comms syntax rendering no)) {
152			push @item, $all_status->{$part}->{ja};
153			push @item, $all_status->{$part}->{all};
154			}
155
156			open my $status_file, '>>', $status_file_name
157			or die "$0: $status_file_name: $!";
158			print $status_file join "\t", @item;
159			print $status_file "\n";
160			close $status_file;
161			}
162
163			save_fallback_entries ();
164
165	wakaba	1.2	sub add_class ($$$;$) {
166	wakaba	1.1	my $tag = shift;
167			my $new_class = shift; # should not contain bare & and bare "
168			my $hash = shift;
169	wakaba	1.2	my $real_hash = shift;
170			$real_hash = qq[ data-ja-real-hash="$real_hash"] if defined $real_hash;
171	wakaba	1.1
172			if ($tag =~ /^<li\b/) {
173			## NOTE: This \|p\| wrapper is necessary, otherwise, if \|li\| element
174			## is set to \|display: table\|, then no list marker is shown.
175	wakaba	1.2	$tag .= qq[<p class="$new_class ja-translation-inserted" data-ja-hash="$hash"$real_hash>];
176	wakaba	1.1	} elsif ($tag =~ /\bclass="/) {
177	wakaba	1.2	$tag =~ s/\bclass="([^"]*)"/class="$1 $new_class" data-ja-hash="$hash"$real_hash/;
178	wakaba	1.1	} elsif ($tag =~ /\bclass=/) {
179	wakaba	1.2	$tag =~ s/\bclass=([^\s>]+)/class="$1 $new_class" data-ja-hash="$hash"$real_hash/g;
180	wakaba	1.1	} else {
181	wakaba	1.2	$tag =~ s/>/ class="$new_class" data-ja-hash="$hash"$real_hash>/;
182	wakaba	1.1	}
183
184			return $tag;
185			} # add_class
186
187			sub escape_id ($$) {
188			my $content = shift;
189			my $id_prefix = shift; # should not contain bare & and bare "
190
191			$content =~ s{<([a-zA-Z0-9-][^<>]+)>}{
192			my $tag_content = $1;
193
194			if ($tag_content =~ /\bid="/) {
195			$tag_content =~ s/\bid="([^"]*)"/id="$id_prefix$1"/;
196			} elsif ($tag_content =~ /\bid=/) {
197			$tag_content =~ s/\bid=(\S+)/id="$id_prefix$1"/;
198			}
199
200			# if ($tag_content =~ /\bhref=#/) {
201			# $tag_content =~ s/\bhref=#(\S+)/href=#$id_prefix$1/;
202			# }
203
204			'<' . $tag_content . '>';
205			}ge;
206
207			return $content;
208			} # escape_id
209	wakaba	1.6
210			## Author: Wakaba <w@suika.fam.cx>.
211			## License: Copyright 2008 Wakaba. You are granted a license to use,
212			## reproduce and create derivative works of this script.
213			## $Date: 2008/10/27 04:52:39 $