calccomp

Tokyo Cabinet的Tokyo Cabinet 是一个DBM的实现。这里的数据库由一系列key-value对的记录构成。key和value都可以是任意长度的字节序列,既可以是二进制也可以是字符

#! /usr/bin/perl

#================================================================
# calccomp
# Measure elapsed time and database size of compression algos
#================================================================

use strict;
use warnings;
use Time::HiRes qw(gettimeofday);

use constant {
    WORDFILE => '/usr/share/dict/words',
    TSVFILE => 'words.tsv',
    NULLFILE => '/dev/null',
    BNUM => 262144,
    LCNUM => 8192,
    TESTCOUNT => 20,
    REMOVETOP => 2,
    REMOVEBOTTOM => 8,
};

open(IN, '<' . WORDFILE) or die(WORDFILE . ' could not be opened');
open(OUT, '>' . TSVFILE) or die(TSVFILE . ' could not be opened');
while(defined(my $word = <IN>)){
    $word =~ s/[\t\r\n]//g;
    next if(length($word) < 1);
    printf OUT ("%s\t%d\n", $word, int(rand(10000)));
}
close(OUT);
close(IN);

my @commands = (
    './tchmgr create casket-hash ' . BNUM . ' ; ./tchmgr importtsv casket-hash ' . TSVFILE,
    './tchmgr list casket-hash >' . NULLFILE,
    './tcbmgr create casket-btree ' . LCNUM . ' ; ./tcbmgr importtsv casket-btree ' . TSVFILE,
    './tcbmgr list casket-btree >' . NULLFILE,
    './tcbmgr create -td casket-bt-td ' . LCNUM . ' ; ./tcbmgr importtsv casket-bt-td ' . TSVFILE,
    './tcbmgr list casket-bt-td >' . NULLFILE,
    './tcbmgr create -tb casket-bt-tb ' . LCNUM . ' ; ./tcbmgr importtsv casket-bt-tb ' . TSVFILE,
    './tcbmgr list casket-bt-tb >' . NULLFILE,
    './tcbmgr create -tt casket-bt-tt ' . LCNUM . ' ; ./tcbmgr importtsv casket-bt-tt ' . TSVFILE,
    './tcbmgr list casket-bt-tt >' . NULLFILE,
    './tcbmgr create -tx casket-bt-tx ' . LCNUM . ' ; ./tcbmgr importtsv casket-bt-tx ' . TSVFILE,
    './tcbmgr list casket-bt-tx >' . NULLFILE,
    );

my @names = (
    'casket-hash',
    'casket-btree',
    'casket-bt-td',
    'casket-bt-tb',
    'casket-bt-tt',
    'casket-bt-tx',
    );

foreach my $name (@names){
    my @paths = glob("$name*");
    foreach my $path (@paths){
        unlink($path);
    }
}

my @table;
foreach my $command (@commands){
    system('sync ; sync');
    $ENV{'HIDEPRGR'} = 1;
    my @result;
    for(my $i = 0; $i < TESTCOUNT; $i++){
        my $stime = gettimeofday();
        system("$command >/dev/null 2>&1");
        $stime = gettimeofday() - $stime;
        printf("%s\t%d\t%0.5f\n", $command, $i + 1, $stime);
        push(@result, $stime);
    }
    @result = sort { $a <=> $b } @result;
    for(my $i = 0; $i < REMOVETOP; $i++){
        shift(@result);
    }
    for(my $i = 0; $i < REMOVEBOTTOM; $i++){
        pop(@result);
    }
    my $sum = 0;
    foreach my $result (@result){
        $sum += $result;
    }
    my $avg = $sum / scalar(@result);
    push(@table, [$command, $avg]);
}

printf("\n\nRESULT\n\n");
foreach my $row (@table){
    printf("%s\t%0.5f\n", $$row[0], $$row[1]);
}
printf("\n");

my @sizes;
foreach my $name (@names){
    my @sbuf = stat($name);
    my $size = $sbuf[7];
    printf("%s\t%s\n", $name, $size);
    push(@sizes, $size);
}
my @sbuf = stat(TSVFILE);
my $size = $sbuf[7];
printf("\n");

printf("%s,%.5f,%.5f,%d\n", "ORIGINAL", 0, 0, $size);
printf("%s,%.5f,%.5f,%d\n", "HASH", $table[0][1], $table[1][1], $sizes[0]);
printf("%s,%.5f,%.5f,%d\n", "BTREE", $table[2][1], $table[3][1], $sizes[1]);
printf("%s,%.5f,%.5f,%d\n", "BTREE-DEFLATE", $table[4][1], $table[5][1], $sizes[2]);
printf("%s,%.5f,%.5f,%d\n", "BTREE-BZIP2", $table[6][1], $table[7][1], $sizes[3]);
printf("%s,%.5f,%.5f,%d\n", "BTREE-TCBS", $table[8][1], $table[9][1], $sizes[4]);
printf("%s,%.5f,%.5f,%d\n", "BTREE-EXCODEC", $table[10][1], $table[11][1], $sizes[5]);



# END OF FILE