#!/usr/pkg/bin/perl
#
# compare speed of egrep and fgrep for many keywords.
#


# configuration params
$egrep="/usr/bin/egrep";
#$egrep="/usr/local/gnu/bin/egrep";
$fgrep="/usr/bin/fgrep";
#$fgrep="/usr/local/gnu/bin/fgrep";

$words="/usr/share/dict/words";
$doc="/tmp/bigtext";

# don't touch following lines
($nwords)=split(' ', `wc -l $words`);
$keywords_e_pref="/tmp/ke";
$keywords_f_pref="/tmp/kf";
$iteration=5;

do efgrep(4);
do efgrep(8);
do efgrep(16);
do efgrep(32);
do efgrep(64);
do efgrep(128);
do efgrep(256);
do efgrep(512);
do efgrep(1024);
do efgrep(2048);

sub efgrep {
    local($nkeywords) = @_;
    local($keywords_e) = ("$keywords_e_pref$nkeywords");
    local($keywords_f) = ("$keywords_f_pref$nkeywords");

    open(WORDS, $words) || die "cannot open $word.\n";
    open(KEYWORDSE, ">$keywords_e") || die "cannot open $keywords_e.\n";
    open(KEYWORDSF, ">$keywords_f") || die "cannot open $keywords_f.\n";
    printf KEYWORDSE "(";
    srand(time + $$);
    $k = 0;
    while (<WORDS>) {
	if (rand($nwords - $. + 1) < ($nkeywords - $k)) {
	    printf KEYWORDSF;
	    chop;
	    if (++$k == $nkeywords) {
		printf KEYWORDSE "$_)";
		last;
	    } else {
		printf KEYWORDSE "$_|";
	    }
	}
    }
    close(WORDS);
    close(KEYWORDSE);
    close(KEYWORDSF);

    printf STDERR "$nkeywords keywords:\n";
    # -q option is not used intentionally. 
    do time_command("$fgrep -f $keywords_f $doc > /dev/null");
    do time_command("$egrep -f $keywords_e $doc > /dev/null");
    printf STDERR "\n";

    #unlink($keywords_f);
    #unlink($keywords_e);

}

sub time_command {
	local($command) = @_;

	printf STDERR "execute: $command\n";
	for ($i = 0; $i < $iteration; $i++) {
		system("time $command");
	}
}