index maker

[Jun-ichiro Itoh <itojun @ mt . cs . keio . ac . jp>]

	Hello, this is Jun-ichiro ITOH, from Japan.
	It's my first post to this list.

	I'm using majordomo 1.62 for about a month at my site.
	It works really fine. It's really great!

	I've added a small script that does a subject-search for 'index'
	command. Feel free to use/modify it.

itojun@mt.cs.keio.ac.jp(Jun-ichiro ITOH)


---sample output
>>>> index dist-test
3      	94/01/09 Jun-ichiro ITOH      (dist-test 3) test                      
4      	94/01/09 Jun-ichiro ITOH      (dist-test 4) test                      
5      	94/01/09 Jun-ichiro ITOH      (dist-test 5) test                      
6      	94/01/09 Jun-ichiro ITOH      (dist-test 6) test                      
7      	94/01/09 Jun-ichiro ITOH      (distrubute test 7) test                
8      	94/01/09 Jun-ichiro Itoh      (dist-test 8) distribute test           
9      	94/01/09 Jun-ichiro Itoh      (dist-test 9) Re: distribute test       
10     	94/01/09 Koichi Moriyama      (dist-test 10) test                     

---majordomo.cf
# $index_command = "/bin/ls -lRL";
$index_command = "$homedir/indexer";

---indexer
#! /usr/local/bin/perl

#	output index, with caching
#	by itojun@mt.cs.keio.ac.jp (Jun-ichiro ITOH)
#	$Header$

$version = '1.0';
$versionString = '***VERSION***';

$cache = '.cache';
# for News, Sun
$cachefile = "$cache.pag";
# for BSD/386
# $cachefile = "$cache.db";

if (-e $cachefile) {
	$cacheDate = (-M $cachefile);
} else {
	$cacheDate = -1;
}

dbmopen(fileCache, $cache, 0644);

# version check
if ($fileCache{$versionString} < $version) {
	$cacheDate = -1;
}

# search flles below here, updating fileCache
&printDir('.', 'DIR000');

# delete removed files from fileCache
foreach $file (keys %fileCache) {
	delete $fileCache{$file} if (!$mark{$file});
}

# output names, with some ordering
foreach $file (sort sortstring grep(!/[0-9]+/, keys %fileCache)) {
	print $fileCache{$file};
}
foreach $file (sort sortnumber grep(/[0-9]+/, keys %fileCache)) {
	print $fileCache{$file};
}

$fileCache{$versionString} = $version;
dbmclose(fileCache);

exit 0;

#------------------------------------------------------------
sub printDir {
	local($directory) = $_[0];
	local($handle) = $_[1];
	local(@allfiles);
	local($file, $shortfile);

	$handle++;
	opendir($handle, $directory);
	@allfiles = grep(!/^\./, readdir($handle));
	foreach $file (@allfiles) {
		if (-B $file) {
			next;
		} elsif (-d $file) {
			&printDir("$directory/$file", $handle);
		} else {
			$shortfile = "$directory/$file";
			$shortfile =~ s/^\.\///;
			$mark{$shortfile} = 1;
			# update not needed if cache is newer than target
			next if (0 < $cacheDate && -M $file > $cacheDate);
			&printFile($shortfile);
		}
	}
	closedir($handle);
}

sub printFile {
	local($file) = $_[0];
	local($from, $subject) = ('', '');
	local($time, $year, $month, $day);

# binary files: will never fall into here
#	if (-B $file) {
#		$fileCache{$file} = "$file\n";
#		return;
#	}

# text files: assume mail or news
	open(FILE, $file);
	while (<FILE>) {
		chop;
		$from = $1 if /^From:\s+(\S.*)$/;
		$subject = $1 if /^Subject:\s+(\S.*)$/;
		last if ($from && $subject);
	}
	close(FILE);

	$time = (stat($file))[9];
	($year, $month, $day) = (gmtime($time))[5,4,3];;
	$month++;

	$subject = '(no subject)' if ($subject eq '');
	$from = $1 if ($from =~ /^([^<]+)\s+<[^>]+>$/);
	$from = '(no name)' if ($from eq '');
	if (length($file) <= 7) {
		$fileCache{$file} = sprintf("%-7s\t%8s %-20s %-40s\n",
			$file, &formatdate($year, $month, $day),
			substr($from, 0, 20), substr($subject, 0, 40));
	} else {
		$fileCache{$file} = sprintf("%s\n\t%8s %-20s %-40s\n",
			$file, &formatdate($year, $month, $day),
			substr($from, 0, 20), substr($subject, 0, 40));
	}
}

sub formatdate {
	local($year, $month, $day) = @_[0,1,2];

	return sprintf("%02d/%02d/%02d", $year, $month, $day);
}

#------------------------------------------------------------
sub sortstring {
	$a cmp $b;
}

sub sortnumber {
	$a <=> $b;
}