html5/spec-ja/make.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 $tbt_file_name_stem = shift;

my %data;
my %pattern;
for_each_data_file (sub ($) {
  my $data_file_name = shift;
  warn "$data_file_name...\n";
  load_data_file ($data_file_name, \%data, \%pattern);
});
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 = {};

open my $tbt_file, '>:utf8', $tbt_file_name_stem . '.dat' or
    die "$0: $tbt_file_name_stem.dat: $!";
open my $part_tbt_file, '>:utf8', $tbt_file_name_stem . '-' . $part . '.dat' or
    die "$0: $tbt_file_name_stem-$part.dat: $!";

warn "Generating...\n";
$source_text =~ s{(<(?>p(?>re)?|li|d[td]|t[dh]|h[1-6])(?>\s[^>]*)?>)((?>(?!</?(?>p(?>re)?|li|d(?>[tdl]|iv)|t(?>[dr]|h(?>ead)?|able|body|foot)|h[1-6r]|ul|ol)(?>\s[^>]*)?>).)+)}
{
  my ($tag, $text) = ($1, $2);
  my $n_text = normalize ($text);

  if ($tag =~ /^<h2 id=(\w+)/) {
    $part = $1;
    open $part_tbt_file, '>:utf8', $tbt_file_name_stem . '-' . $part . '.dat' or
        die "$0: $tbt_file_name_stem-$part.dat: $!";
  }
  
  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}++;
      $tag . q[<span class=ja-translation lang=ja>] . $ja_text . q[</span>];
    } else {
      my $v = $tag . $text;
      my $has_ja;
      for my $pattern (@pattern) {
        if ($n_text =~ /^$pattern$/) {
          $status->{ja}++;
          $all_status->{$part}->{ja}++;
          $v = $tag . q[<span class=ja-translation lang=ja>] .
              replace_pattern2 ($pattern{$pattern}, $1, $2, $3, $4, $5) .
              q[</span>];
          $has_ja = 1;
          last;
        }
      }

      unless ($has_ja) {
        $text =~ s/^\s+//;
        $text =~ s/\s+\z//;
        $text =~ s/\x0D?\x0A(?:\x0D?\x0A)+/\n/g;
        unless ($tbt_added->{$text}) {
          print $tbt_file ($text);
          print $tbt_file ("\n\n");
          print $part_tbt_file ($text);
          print $part_tbt_file ("\n\n");
          $tbt_added->{$text} = 1;
        }
      }

      $v;
    }
  } else {
    $1 . $2;
  }
}ges;
$source_text =~ s{(<(?>link|img|script)\s[^>]+>)}{
  my $tag = $1;
  my $n_text = normalize ($tag);
  my $ja_text = $data{$n_text};
  if (defined $ja_text) {
    $ja_text;
  } else {
    $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;
}
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	1.15	my $tbt_file_name_stem = shift;
10	wakaba	1.1
11			my %data;
12			my %pattern;
13	wakaba	1.11	for_each_data_file (sub ($) {
14			my $data_file_name = shift;
15			warn "$data_file_name...\n";
16	wakaba	1.17	load_data_file ($data_file_name, \%data, \%pattern);
17	wakaba	1.11	});
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	1.15	my $part = 'cover';
30	wakaba	1.9	my $status = {};
31	wakaba	1.15	my $all_status = {};
32	wakaba	1.14	my $tbt_added = {};
33	wakaba	1.9
34	wakaba	1.15	open my $tbt_file, '>:utf8', $tbt_file_name_stem . '.dat' or
35			die "$0: $tbt_file_name_stem.dat: $!";
36			open my $part_tbt_file, '>:utf8', $tbt_file_name_stem . '-' . $part . '.dat' or
37			die "$0: $tbt_file_name_stem-$part.dat: $!";
38
39	wakaba	1.1	warn "Generating...\n";
40	wakaba	1.12	$source_text =~ s{(<(?>p(?>re)?\|li\|d[td]\|t[dh]\|h[1-6])(?>\s[^>])?>)((?>(?!</?(?>p(?>re)?\|li\|d(?>[tdl]\|iv)\|t(?>[dr]\|h(?>ead)?\|able\|body\|foot)\|h[1-6r]\|ul\|ol)(?>\s[^>])?>).)+)}
41	wakaba	1.1	{
42			my ($tag, $text) = ($1, $2);
43			my $n_text = normalize ($text);
44	wakaba	1.15
45			if ($tag =~ /^<h2 id=(\w+)/) {
46			$part = $1;
47			open $part_tbt_file, '>:utf8', $tbt_file_name_stem . '-' . $part . '.dat' or
48			die "$0: $tbt_file_name_stem-$part.dat: $!";
49			}
50	wakaba	1.10
51			if (length $n_text) {
52			my $ja_text = $data{$n_text};
53
54			$status->{all}++;
55	wakaba	1.15	$all_status->{$part}->{all}++;
56	wakaba	1.10
57			if (defined $ja_text) {
58			$status->{ja}++;
59	wakaba	1.15	$all_status->{$part}->{ja}++;
60	wakaba	1.10	$tag . q[<span class=ja-translation lang=ja>] . $ja_text . q[</span>];
61			} else {
62			my $v = $tag . $text;
63	wakaba	1.13	my $has_ja;
64	wakaba	1.10	for my $pattern (@pattern) {
65			if ($n_text =~ /^$pattern$/) {
66			$status->{ja}++;
67	wakaba	1.18	$all_status->{$part}->{ja}++;
68	wakaba	1.10	$v = $tag . q[<span class=ja-translation lang=ja>] .
69			replace_pattern2 ($pattern{$pattern}, $1, $2, $3, $4, $5) .
70			q[</span>];
71	wakaba	1.13	$has_ja = 1;
72	wakaba	1.10	last;
73			}
74	wakaba	1.1	}
75	wakaba	1.13
76			unless ($has_ja) {
77			$text =~ s/^\s+//;
78			$text =~ s/\s+\z//;
79			$text =~ s/\x0D?\x0A(?:\x0D?\x0A)+/\n/g;
80	wakaba	1.14	unless ($tbt_added->{$text}) {
81	wakaba	1.15	print $tbt_file ($text);
82			print $tbt_file ("\n\n");
83			print $part_tbt_file ($text);
84			print $part_tbt_file ("\n\n");
85	wakaba	1.14	$tbt_added->{$text} = 1;
86			}
87	wakaba	1.13	}
88
89	wakaba	1.10	$v;
90	wakaba	1.1	}
91	wakaba	1.10	} else {
92			$1 . $2;
93	wakaba	1.1	}
94			}ges;
95			$source_text =~ s{(<(?>link\|img\|script)\s[^>]+>)}{
96			my $tag = $1;
97			my $n_text = normalize ($tag);
98			my $ja_text = $data{$n_text};
99			if (defined $ja_text) {
100			$ja_text;
101			} else {
102			$tag;
103			}
104			}ges;
105
106	wakaba	1.4	$source_text =~ s{\[\[([A-Z ]+):([^]]+)\]\]}
107	wakaba	1.1	{<em class=rfc2119 title="$1">$2</em>}gs;
108
109			#$source_text =~ s[<title>][<base href="http://www.whatwg.org/specs/web-apps/current-work/"><title>];
110
111			{
112			warn "$result_file_name...\n";
113			open my $result_file, '>:utf8', $result_file_name
114			or die "$0: $result_file_name: $!";
115			print $result_file $source_text;
116	wakaba	1.9	}
117
118			{
119			my $time = time;
120	wakaba	1.15	my @item = ($time, $status->{ja}, $status->{all});
121			for my $part (qw(cover introduction infrastructure dom semantics browsers
122			editing comms syntax rendering no)) {
123			push @item, $all_status->{$part}->{ja};
124			push @item, $all_status->{$part}->{all};
125			}
126
127	wakaba	1.9	open my $status_file, '>>', $status_file_name
128			or die "$0: $status_file_name: $!";
129	wakaba	1.15	print $status_file join "\t", @item;
130			print $status_file "\n";
131			close $status_file;
132	wakaba	1.1	}