#!/usr/bin/perl -w
use strict;

## digioPDF2text
##    with pdftotext 1.00 (included in Xpdf 1.00)
## tested with RedHatLinux7.3 (perl-5.6.1, pdftotext1.00, wget1.8.1)
##
## Time-stamp: "Jan 25 2003"
## author: itouh <wmwm(a)nifty.com>

my @channel = qw(403 446 473);

mainfunc();
exit();

sub mainfunc {
    my $issueday = getMondayOfThisWeek();

    foreach my $ch (@channel){
        my @day = sort glob("$ch-*.txt");
        if($#day >= 0){
            my ($d) = $day[$#day] =~ /^$ch\-([0-9\-]*)\.txt$/;
            if($issueday eq $d){
                ## 既に取得済み
                print "  $ch-$d.txt is already there, will not retrieve.\n";
                next;
            }
        }
        my $pdffile = getPDF($ch);
        pdf2text($pdffile, "$ch-$issueday.txt");
    }
}

sub getMondayOfThisWeek {
    my $t = time();
    ## am0:00〜3:59は昨日のプログラムを放送中なので、4時間引く
    if ((localtime($t))[2] < 4){
        $t -= 4*60*60;
    }
    ## 月曜になるまで 24時間ずつ引く
    while((localtime($t))[6] != 1){
        $t -= 24*60*60;
    }
    ## 月曜の日付を返す
    my ($d, $m, $y) = (localtime($t))[3..5];
    return sprintf("%04d-%02d%02d", $y+1900, $m+1, $d);
}


sub getPDF {
    my ($ch) = @_;
    if(-s "$ch.pdf"){
        unlink("$ch.pdf");
    }
    system("wget http://www.stardigio.com/weekly/proglist/$ch.pdf")
        && die "cannot get $ch.pdf!";
    return "$ch.pdf";
}

sub pdf2text {
    my ($pdffile, $txtfile) = @_;
    my $tmpfile = "tmp.txt";

    if(-s $tmpfile){
        unlink($tmpfile);
    }
    system("pdftotext -enc EUC-JP -raw $pdffile $tmpfile")
        && die "cannot read $pdffile!";
    open(IN,$tmpfile) or die "cannot read $tmpfile!";
    open(OUT,">$txtfile") or die "cannot write $txtfile!";

    while(<IN>){
        chomp();
        if(/^/){
            next;
        }

        ## 空白をある程度削って、等幅フォント環境で演奏者名を揃える
        my $buf = $_;
        $buf =~ s/   .*$//;                #曲タイトルだけ抽出
        my $zen = $buf =~ tr/\x80-\xfe//;  #曲タイトルが含む全角文字の数(の2倍)
        if($zen > 0){
            my $rmsp = " "x($zen / 2); #remove spaces
            $_ =~ s/$rmsp//;
        }

#        if(/^ / or /^　/){
#            ## 行頭に半角／全角空白がある場合、前の行から改行しない
#            print OUT "$_";
#        }else{
#            print OUT "\n$_";
#        }

        print OUT "\n$_";
    }
    print OUT "\n";
    close(OUT);
    close(IN);
}
## end of file