| 1 |
#!/usr/bin/perl |
#!/usr/bin/perl |
| 2 |
use strict; |
use strict; |
| 3 |
|
|
| 4 |
|
BEGIN { require 'common.pl' } |
| 5 |
|
|
| 6 |
my $source_file_name = shift; |
my $source_file_name = shift; |
|
my $data_dir_name = q[data/]; |
|
|
my $data_suffix = q[.dat]; |
|
| 7 |
my $result_file_name = shift; |
my $result_file_name = shift; |
| 8 |
my $status_file_name = shift; |
my $status_file_name = shift; |
| 9 |
|
|
|
sub normalize ($) { |
|
|
my $s = shift; |
|
|
$s =~ s/\s+/ /g; |
|
|
$s =~ s/^ //; |
|
|
$s =~ s/ $//g; |
|
|
return $s; |
|
|
} # normalize |
|
|
|
|
|
sub create_pattern1 ($) { |
|
|
my $s = quotemeta shift; |
|
|
$s =~ s/\\\*/(.+)/g; |
|
|
return $s; |
|
|
} # create_pattern1 |
|
|
|
|
|
sub replace_pattern2 ($@) { |
|
|
my $s = shift; |
|
|
my @arg = @_; |
|
|
$s =~ s/\$(\d+)/$arg[$1 - 1]/g; |
|
|
return $s; |
|
|
} # replace_pattern2 |
|
|
|
|
| 10 |
my %data; |
my %data; |
| 11 |
my %pattern; |
my %pattern; |
| 12 |
{ |
for_each_data_file (sub ($) { |
| 13 |
opendir my $data_dir, $data_dir_name or die "$0: $data_dir_name: $!"; |
my $data_file_name = shift; |
| 14 |
local $/ = undef; |
warn "$data_file_name...\n"; |
| 15 |
for (readdir $data_dir) { |
load_data_file ($data_file_name, \%data, \%pattern); |
| 16 |
next if /^\./; # hidden files |
}); |
|
my $data_file_name = qq[$data_dir_name$_]; |
|
|
next unless $data_file_name =~ /\Q$data_suffix\E$/; |
|
|
warn "$data_file_name...\n"; |
|
|
open my $data_file, '<:utf8', $data_file_name |
|
|
or die "$0: $data_file_name: $!"; |
|
|
my $data = <$data_file>; |
|
|
$data =~ s/\x0D?\x0A/\n/g; |
|
|
for (split /\n\n+(?=#)/, $data) { |
|
|
my ($en, $ja) = split /\n#ja\n/, $_; |
|
|
if ($en =~ s/^#en\n//) { |
|
|
$data{normalize ($en)} = $ja; |
|
|
} elsif ($en =~ s/^#pattern\n//) { |
|
|
$pattern{create_pattern1 (normalize ($en))} = $ja; |
|
|
} |
|
|
} |
|
|
} |
|
|
} |
|
| 17 |
my @pattern = sort {length $b <=> length $a} keys %pattern; |
my @pattern = sort {length $b <=> length $a} keys %pattern; |
| 18 |
|
|
| 19 |
my $source_text; |
my $source_text; |