| 1 |
wakaba |
1.1 |
#!/usr/bin/perl |
| 2 |
|
|
use strict; |
| 3 |
|
|
|
| 4 |
wakaba |
1.11 |
BEGIN { require 'common.pl' } |
| 5 |
|
|
|
| 6 |
wakaba |
1.6 |
my $source_file_name = shift; |
| 7 |
|
|
my $result_file_name = shift; |
| 8 |
wakaba |
1.9 |
my $status_file_name = shift; |
| 9 |
wakaba |
1.13 |
my $tbt_file_name = shift; |
| 10 |
wakaba |
1.1 |
|
| 11 |
|
|
my %data; |
| 12 |
|
|
my %pattern; |
| 13 |
wakaba |
1.11 |
for_each_data_file (sub ($) { |
| 14 |
|
|
my $data_file_name = shift; |
| 15 |
|
|
warn "$data_file_name...\n"; |
| 16 |
|
|
load_data_file ($data_file_name, \%data, \%pattern); |
| 17 |
|
|
}); |
| 18 |
wakaba |
1.2 |
my @pattern = sort {length $b <=> length $a} keys %pattern; |
| 19 |
wakaba |
1.1 |
|
| 20 |
|
|
my $source_text; |
| 21 |
|
|
{ |
| 22 |
|
|
warn "$source_file_name...\n"; |
| 23 |
|
|
open my $source_file, '<:utf8', $source_file_name |
| 24 |
|
|
or die "$0: $source_file_name: $!"; |
| 25 |
|
|
local $/ = undef; |
| 26 |
|
|
$source_text = <$source_file>; |
| 27 |
|
|
} |
| 28 |
|
|
|
| 29 |
wakaba |
1.13 |
open my $tbt_file, '>:utf8', $tbt_file_name or die "$0: $tbt_file_name: $!"; |
| 30 |
|
|
|
| 31 |
wakaba |
1.9 |
my $status = {}; |
| 32 |
|
|
|
| 33 |
wakaba |
1.1 |
warn "Generating...\n"; |
| 34 |
wakaba |
1.12 |
$source_text =~ s{(<(?>p(?>re)?|li|d[td]|t[dh]|h[1-6])(?>\s[^>]*)?>)((?>(?!</?(?>p(?>re)?|li|d(?>[tdl]|iv)|t(?>[dr]|h(?>ead)?|able|body|foot)|h[1-6r]|ul|ol)(?>\s[^>]*)?>).)+)} |
| 35 |
wakaba |
1.1 |
{ |
| 36 |
|
|
my ($tag, $text) = ($1, $2); |
| 37 |
|
|
my $n_text = normalize ($text); |
| 38 |
wakaba |
1.10 |
|
| 39 |
|
|
if (length $n_text) { |
| 40 |
|
|
my $ja_text = $data{$n_text}; |
| 41 |
|
|
|
| 42 |
|
|
$status->{all}++; |
| 43 |
|
|
|
| 44 |
|
|
if (defined $ja_text) { |
| 45 |
|
|
$status->{ja}++; |
| 46 |
|
|
$tag . q[<span class=ja-translation lang=ja>] . $ja_text . q[</span>]; |
| 47 |
|
|
} else { |
| 48 |
|
|
my $v = $tag . $text; |
| 49 |
wakaba |
1.13 |
my $has_ja; |
| 50 |
wakaba |
1.10 |
for my $pattern (@pattern) { |
| 51 |
|
|
if ($n_text =~ /^$pattern$/) { |
| 52 |
|
|
$status->{ja}++; |
| 53 |
|
|
$v = $tag . q[<span class=ja-translation lang=ja>] . |
| 54 |
|
|
replace_pattern2 ($pattern{$pattern}, $1, $2, $3, $4, $5) . |
| 55 |
|
|
q[</span>]; |
| 56 |
wakaba |
1.13 |
$has_ja = 1; |
| 57 |
wakaba |
1.10 |
last; |
| 58 |
|
|
} |
| 59 |
wakaba |
1.1 |
} |
| 60 |
wakaba |
1.13 |
|
| 61 |
|
|
unless ($has_ja) { |
| 62 |
|
|
$text =~ s/^\s+//; |
| 63 |
|
|
$text =~ s/\s+\z//; |
| 64 |
|
|
$text =~ s/\x0D?\x0A(?:\x0D?\x0A)+/\n/g; |
| 65 |
|
|
print $tbt_file $text; |
| 66 |
|
|
print $tbt_file "\n\n"; |
| 67 |
|
|
} |
| 68 |
|
|
|
| 69 |
wakaba |
1.10 |
$v; |
| 70 |
wakaba |
1.1 |
} |
| 71 |
wakaba |
1.10 |
} else { |
| 72 |
|
|
$1 . $2; |
| 73 |
wakaba |
1.1 |
} |
| 74 |
|
|
}ges; |
| 75 |
|
|
$source_text =~ s{(<(?>link|img|script)\s[^>]+>)}{ |
| 76 |
|
|
my $tag = $1; |
| 77 |
|
|
my $n_text = normalize ($tag); |
| 78 |
|
|
my $ja_text = $data{$n_text}; |
| 79 |
|
|
if (defined $ja_text) { |
| 80 |
|
|
$ja_text; |
| 81 |
|
|
} else { |
| 82 |
|
|
$tag; |
| 83 |
|
|
} |
| 84 |
|
|
}ges; |
| 85 |
|
|
|
| 86 |
wakaba |
1.4 |
$source_text =~ s{\[\[([A-Z ]+):([^]]+)\]\]} |
| 87 |
wakaba |
1.1 |
{<em class=rfc2119 title="$1">$2</em>}gs; |
| 88 |
|
|
|
| 89 |
|
|
#$source_text =~ s[<title>][<base href="http://www.whatwg.org/specs/web-apps/current-work/"><title>]; |
| 90 |
|
|
|
| 91 |
|
|
{ |
| 92 |
|
|
warn "$result_file_name...\n"; |
| 93 |
|
|
open my $result_file, '>:utf8', $result_file_name |
| 94 |
|
|
or die "$0: $result_file_name: $!"; |
| 95 |
|
|
print $result_file $source_text; |
| 96 |
wakaba |
1.9 |
} |
| 97 |
|
|
|
| 98 |
|
|
{ |
| 99 |
|
|
my $time = time; |
| 100 |
|
|
open my $status_file, '>>', $status_file_name |
| 101 |
|
|
or die "$0: $status_file_name: $!"; |
| 102 |
|
|
print $status_file "$time\t$status->{ja}\t$status->{all}\n"; |
| 103 |
wakaba |
1.1 |
} |
Parent Directory
| 
Revision Log