#!/usr/bin/perl -w
#
# mlt file spliter
#
# ./split.pl file_name.mlt [prefix]
#
#
use strict;
use warnings;
use utf8;
use File::Basename 'fileparse';
use File::Spec;
use NKF;
my $file = $ARGV[0];
my $prefix = $ARGV[1] ? $ARGV[1] : '';
my ($basename, $dirname, $ext) = fileparse($file, qr/\..*$/);
my $dir = $basename;
if (! -d $dir) {
mkdir ($dir);
}
open( my $in, $file) || die "$!";
my $list_file_name = File::Spec->catfile($dir, '_list.html');
open(LIST, ">" . $list_file_name) || die "$!";
my $header_html = <<"EOD";
EOD
print (LIST nkf('-w', $header_html));
my $count = 0 ;
my $skip = 0;
my $buf;
my $pattern = nkf("-w" , "^(【.*】|最終更新日)");
while( <$in> ) {
$_ = nkf("-w -x --ic=CP932", $_);
if (/$pattern/) {
$skip = 1;
} elsif (/^\[SPLIT/) {
if ($skip != 1) {
$count += 1;
my $output_file_name = File::Spec->catfile($dir, $prefix . $count . '.txt');
open(OUT, ">" . $output_file_name) || die "$!";
print(OUT $buf);
close(OUT);
my $brbuf = $buf;
$brbuf =~ s/\n/
/g;
print(LIST "id:$count
\n" .
$brbuf . "
\n")
}
$buf = "";
$skip = 0;
} else {
$buf = $buf . $_;
}
}
print(LIST "");
close(LIST);