#! /usr/local/bin/perl -w
# -*- Perl -*-
#
# $Id: gcnmz.in,v 1.8.8.5 2006-10-14 14:16:27 opengl2772 Exp $
#
# gcnmz.pl - by furukawa@tcp-ip.or.jp
#

my $PKGDATADIR = $ENV{'pkgdatadir'} || "/usr/local/share/namazu";
push(@INC, $PKGDATADIR . "/pl");
require 'nmzidx.pl';
require 'util.pl';

use IO::File;
STDOUT->autoflush(1);
STDERR->autoflush(1);

my $tim = time();
my $tintvl = 10;
my $verbose = 0;
my $backup = 1;

while (@ARGV && $ARGV[0] =~ s/^\-//){
    my $argv = shift;

    &usage, exit if $argv eq '-help';
    $backup = 0, next if $argv eq '-no-backup';
    $verbose = 1, next if $argv eq '-verbose';
    $verbose = -1, next if $argv eq '-quiet';

    while ($argv =~ s/^(.)//){
        $backup = 0 if $1 eq 'b';
        $verbose = 1 if $1 eq 'v';
        $verbose = -1 if $1 eq 'q';
    }
}

if (@ARGV){
    for my $argv (@ARGV){
        $argv =~ s/NMZ$// unless -d $argv;
        $argv = '.' if $argv eq '';
        &gcnmz($argv, $backup) if -d $argv;
    }
}else{
    &usage;
}

sub eprint{
    print STDERR $_[0] if $verbose >= 0;
    $tim = time();
}

sub veprint{
    print STDERR $_[0] if $verbose > 0;
    $tim = time();
}

sub gcnmz{
    my $dir = shift;
    my $backup = shift;
    my %table = ();
    my $nmzi = new nmzidx($dir);
    defined($nmzi) or die("NMZ.lock2 found. Maybe this index is being updated by another process now.");

    my $nmzfi = $nmzi->open_flist;
    unless (defined $nmzfi->{'t'}) {
        $nmzfi->close;
        $nmzi->close;
        return;
    }

    my $gcflag = 0;
    my $nmzo = new nmzidx($dir, 'w');

    my $file_num = 0;
    my %list;
    my $nmzfo = $nmzo->open_flist;
    my $file_total = $nmzfi->{'size'};
    &eprint("processing NMZ.{t,r,field}\n");
    while (defined $nmzfi->read(\%list)){
        if ($list{'t'} == -1){
            &eprint("$file_num/$file_total ($list{'r'}) deleted\n");
            $gcflag = 1;
        }else{
            $table{$file_num} = $nmzfo->{'offset'};
            $nmzfo->write(\%list);
            &veprint("$file_num/$file_total\n") if time() - $tim >= $tintvl;
        }
        ++$file_num;
    }
    $nmzfi->close;
    $nmzfo->close;

    if ($gcflag){
        my $word;
        my %list_i;

        my $nmzwi = $nmzi->open_word;
        my $nmzwo = $nmzo->open_word;
        my $word_total = $nmzwi->{'size'};
        my $word_num = 0;
        &eprint("processing NMZ.{i,w}\n");
        while (defined $nmzwi->read(\$word, \%list_i)){
            my %list_o;
            for my $key (keys %list_i){
                if (defined $table{$key}){
                    $list_o{$table{$key}} = $list_i{$key};
                }
            }
            if (keys %list_o){
                $nmzwo->write($word, \%list_o);
                &veprint("$word_num/$word_total\n") if time() - $tim >= $tintvl;
            }else{
                &eprint("$word_num/$word_total ($word) deleted\n");
            }
            $word_num++;
        }
        $nmzwi->close;
        $nmzwo->close;

        my $nmzpi = $nmzi->open_phrase;
        my $nmzpo = $nmzo->open_phrase;
        my $phrase_total = $nmzpi->{'size'};
        &eprint("processing NMZ.p\n");
        for (my $x = 0; $x < 0x10000; $x++){
            my @list_i;
            my @list_o;
            $nmzpi->read(\@list_i);
            for my $key (@list_i){
                if (defined $table{$key}){
                    push(@list_o, $table{$key});
                }
            }
            if (@list_i && !@list_o){
                &eprint("$x/$phrase_total deleted\n");
            }elsif (time() - $tim >= $tintvl){
                &veprint("$x/$phrase_total\n");
            }
            $nmzpo->write(\@list_o);
        }
        $nmzpi->close;
        $nmzpo->close;

        $nmzo->write_status($nmzi);
        $nmzo->replace_db($backup);
        if (my $log = $nmzo->log_open("[Garbage Collection]")){
            $log->printf("%-20s %s\n", "Collected Entry:", util::commas($nmzi->{'flist'}->{'offset'} - $nmzo->{'flist'}->{'offset'}));
            $log->printf("%-20s %s\n", "Total Files:", util::commas($nmzo->{'flist'}->{'offset'}));
            $log->printf("%-20s %s\n", "Total Keywords:", util::commas($nmzo->{'word'}->{'offset'}));
            $nmzo->log_close;
        }
    }else{
        $nmzo->remove_tmpdb;
    }
    $nmzi->close;
}

sub usage{
    print
        ("Usage: gcnmz [options] <target>...\n" .
         "  --help              show this help and exit.\n" .
         "  -b, --no-backup     do not backup original file.\n" .
         "  -v, --verbose       verbose mode.\n" .
         "  -q, --quiet         quiet mode.\n"
         );
}