| 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 |
$source_text =~ s{(<(?>p(?>re)?|li|d[td]|t[dh]|h[1-6]|caption)(?>\s[^>]*)?>)((?>(?!</?(?>p(?>re)?|li|d(?>[tdl]|iv)|t(?>[dr]|h(?>ead)?|able|body|foot)|h[1-6r]|ul|ol|caption)(?>\s[^>]*)?>).)+)} |
| 46 |
{ |
| 47 |
my ($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 |
my $real_hash = get_hash ($n_text); |
| 81 |
$v = add_class ($tag, 'has-ja-translation', |
| 82 |
$pattern{$pattern}->[1], $real_hash) . |
| 83 |
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, $5) . |
| 89 |
q[</span>]; |
| 90 |
|
| 91 |
unless ($tbt_added->{$n_text}) { |
| 92 |
set_fallback_entry ($real_hash => {en => $text}); |
| 93 |
$tbt_added->{$n_text} = 1; |
| 94 |
} |
| 95 |
last; |
| 96 |
} |
| 97 |
} |
| 98 |
|
| 99 |
unless (defined $v) { |
| 100 |
my $hash = get_hash ($n_text); |
| 101 |
$v = add_class ($tag, 'no-ja-translation', $hash) . |
| 102 |
'<span class=en-original lang=en>' . |
| 103 |
$prefix . $text . |
| 104 |
'</span>'; |
| 105 |
|
| 106 |
unless ($tbt_added->{$n_text}) { |
| 107 |
set_fallback_entry ($hash => {en => $text}); |
| 108 |
$tbt_added->{$n_text} = 1; |
| 109 |
} |
| 110 |
} |
| 111 |
|
| 112 |
$v; |
| 113 |
} |
| 114 |
} else { |
| 115 |
$1 . $2; |
| 116 |
} |
| 117 |
}ges; |
| 118 |
$source_text =~ s{(<(?>link|img|script)\s[^>]+>)}{ |
| 119 |
my $tag = $1; |
| 120 |
my $n_text = normalize ($tag); |
| 121 |
my $ja_text = $data{$n_text}; |
| 122 |
if (defined $ja_text) { |
| 123 |
$ja_text->[0]; |
| 124 |
} else { |
| 125 |
$tag; |
| 126 |
} |
| 127 |
}ges; |
| 128 |
|
| 129 |
$source_text =~ s{\[\[([A-Z ]+):([^]]+)\]\]} |
| 130 |
{<em class=rfc2119 title="$1">$2</em>}gs; |
| 131 |
|
| 132 |
#$source_text =~ s[<title>][<base href="http://www.whatwg.org/specs/web-apps/current-work/"><title>]; |
| 133 |
|
| 134 |
{ |
| 135 |
warn "$result_file_name...\n"; |
| 136 |
open my $result_file, '>:utf8', $result_file_name |
| 137 |
or die "$0: $result_file_name: $!"; |
| 138 |
print $result_file $source_text; |
| 139 |
} |
| 140 |
|
| 141 |
{ |
| 142 |
my $time = time; |
| 143 |
my @item = ($time, $status->{ja}, $status->{all}); |
| 144 |
for my $part (qw(cover introduction infrastructure dom semantics browsers |
| 145 |
editing comms syntax rendering no)) { |
| 146 |
push @item, $all_status->{$part}->{ja}; |
| 147 |
push @item, $all_status->{$part}->{all}; |
| 148 |
} |
| 149 |
|
| 150 |
open my $status_file, '>>', $status_file_name |
| 151 |
or die "$0: $status_file_name: $!"; |
| 152 |
print $status_file join "\t", @item; |
| 153 |
print $status_file "\n"; |
| 154 |
close $status_file; |
| 155 |
} |
| 156 |
|
| 157 |
save_fallback_entries (); |
| 158 |
|
| 159 |
sub add_class ($$$;$) { |
| 160 |
my $tag = shift; |
| 161 |
my $new_class = shift; # should not contain bare & and bare " |
| 162 |
my $hash = shift; |
| 163 |
my $real_hash = shift; |
| 164 |
$real_hash = qq[ data-ja-real-hash="$real_hash"] if defined $real_hash; |
| 165 |
|
| 166 |
if ($tag =~ /^<li\b/) { |
| 167 |
## NOTE: This |p| wrapper is necessary, otherwise, if |li| element |
| 168 |
## is set to |display: table|, then no list marker is shown. |
| 169 |
$tag .= qq[<p class="$new_class ja-translation-inserted" data-ja-hash="$hash"$real_hash>]; |
| 170 |
} elsif ($tag =~ /\bclass="/) { |
| 171 |
$tag =~ s/\bclass="([^"]*)"/class="$1 $new_class" data-ja-hash="$hash"$real_hash/; |
| 172 |
} elsif ($tag =~ /\bclass=/) { |
| 173 |
$tag =~ s/\bclass=([^\s>]+)/class="$1 $new_class" data-ja-hash="$hash"$real_hash/g; |
| 174 |
} else { |
| 175 |
$tag =~ s/>/ class="$new_class" data-ja-hash="$hash"$real_hash>/; |
| 176 |
} |
| 177 |
|
| 178 |
return $tag; |
| 179 |
} # add_class |
| 180 |
|
| 181 |
sub escape_id ($$) { |
| 182 |
my $content = shift; |
| 183 |
my $id_prefix = shift; # should not contain bare & and bare " |
| 184 |
|
| 185 |
$content =~ s{<([a-zA-Z0-9-][^<>]+)>}{ |
| 186 |
my $tag_content = $1; |
| 187 |
|
| 188 |
if ($tag_content =~ /\bid="/) { |
| 189 |
$tag_content =~ s/\bid="([^"]*)"/id="$id_prefix$1"/; |
| 190 |
} elsif ($tag_content =~ /\bid=/) { |
| 191 |
$tag_content =~ s/\bid=(\S+)/id="$id_prefix$1"/; |
| 192 |
} |
| 193 |
|
| 194 |
# if ($tag_content =~ /\bhref=#/) { |
| 195 |
# $tag_content =~ s/\bhref=#(\S+)/href=#$id_prefix$1/; |
| 196 |
# } |
| 197 |
|
| 198 |
'<' . $tag_content . '>'; |
| 199 |
}ge; |
| 200 |
|
| 201 |
return $content; |
| 202 |
} # escape_id |
Parent Directory
| 
Revision Log