#dbfix, Copyright 2000 Boutell.Com, Inc.
#
#Recreates database.dat files based on data found in Wusage 6 
#and Wusage 7 reports for individual time periods.
#This script assumes you have not altered your
#report formatting too drastically. If you have, you will
#need to make adjustments. TBB 02/02/00
#
#DON'T RUN THIS if your historical charts are fine!
#Only customers who used Wusage 7.0 P11 are likely to need this.
#7.0 P11 was only available for download for about 24 hours.
#
#Your old database.dat files will be backed up to 
#database.bak, just in case you ran this program by mistake.
#Please note that database.bak can be overwritten if you run it twice.
#
#TO USE:
#
#1. VERIFY THAT YOU EVER HAD 7.0 P11, WHICH WAS ONLY AVAILABLE
#ON 1/31/00 FOR 24 HOURS, OR THAT YOUR HISTORICAL TOTALS GRAPHS
#ARE COMPLETELY ZEROED OUT FOR SOME OTHER REASON.
#
#2. Type "perl dbfix /path/to/my/conf/file"
#
#3. If you don't get good results, please send your configuration
#file to us for analysis, and explain that you were trying to use
#dbfix and did not get the expected results.
#
#
#END OF INSTRUCTIONS

@frequencies = (
	"daily", "weekly", "monthly", "quarterly", "annual" 
);

@hashes = (
        "documents",
        "notfound",
        "sites",
        "proxysites",
        "useragents",
        "referers",
        "referrers",
        "referringsites",
        "searchkeywords",
        "authusers",
        "domains",
        "proxydomains",
        "os",
        "types",
        "visits",
        "trails"
);

$found = 0;

$conf = $ARGV[0];
if ($conf eq "") {
	die "Usage: perlfix /path/to/my/conf/file\n";
}

open(IN, $conf) || die "Can't open configuration file $conf\n";
while ($line = <IN>) {
	if ($line =~ /^reportorder/) {
		while ($line = <IN>) {
			if ($line =~ /^end/) {
				last;
			}
		}
	}
	if ($line =~ /^reportdir\s+(\S+)/) {
		$reportDir = $1;
	}
	if (($line =~ /^totals\s*$/) || ($line =~ /^totalsgroup\s*$/)) {
		while ($line = <IN>) {
			if ($line =~ /^end/) {
				last;
			}
			$hashes = join('|', @hashes);
			if ($line =~ /^($hashes) /i) {
				$hash = $1;	
			} else {
				$hash = "documents";
			}
			$line =~ s/^((function|nograph|nototal|yeartoyear) )+//;
			if ($line =~ /^((?:$hashes )?(?:(?:\{.*?\})|[^\s\{\}]+)+)\s+(.*)\s*$/) {
				($pattern, $title) = ($1, $2);	
				$pattern = "$hash $pattern";
				$patterns{$title} = $pattern;
				$titles{$pattern} = $title;
				push @patterns, $pattern;
			}
		}
	}	
}
close(IN);

for $f (@frequencies)
{
	opendir(DIR, "$reportDir/$f") || next;
	if (open(IN, "$reportDir/$f/database.dat")) {
		open(OUT, ">$reportDir/$f/database.bak");
		while ($line = <IN>) {
			print OUT $line;
		}
		close(IN);
		close(OUT);
	}
	open(OUT, ">$reportDir/$f/database.dat");
	print OUT "Date\tDays";	
	for $p (@patterns) {
		print OUT "\t$p accesses\t$p bytes\t$p visits";
	}
	print OUT "\n";	
	$found++;
	%accesses = ( );
	%bytes = ( );
	%visits = ( );
	for ($year = 1990; ($year < 2010); $year++) {
		opendir(YDIR, "$reportDir/$f/$year") || next;
		closedir(YDIR);			
		for ($month = 1; ($month <= 12); $month++) {
			$month = sprintf("%02d", $month);
			opendir(MDIR, "$reportDir/$f/$year/$month") || next;
			closedir(MDIR);			
			for ($day = 1; ($day <= 31); $day++) {
				$day = sprintf("%02d", $day);
				opendir(DDIR, "$reportDir/$f/$year/$month/$day") || next;
				closedir(DDIR);			
				if (!(open(IN, "$reportDir/$f/$year/$month/$day/exec.html") ||
					open(IN, "$reportDir/$f/$year/$month/$day/exec.htm") ||
					open(IN, "$reportDir/$f/$year/$month/$day/index.html") ||
					open(IN, "$reportDir/$f/$year/$month/$day/index.htm"))) {
					next;
				}
				$data = "";
				while ($line = <IN>) {
					$data .= $line;	
				}
				close(IN);
				$data =~ s/\,//g;
				# Wusage 6 style
				if ($data =~ /\<h3\>Totals\<\/h3\>\s+\<table.*?\>(.*?)\<\/table\>/is) {
					$data = $1;
					while ($data =~ /\<td\>(.*?)\<\/td\>\s+\<td.*?\>\s*([\d\,\.]+)\s*\<\/td\>\s+\<td.*?\>\s*([\d\,\.]+)\s*\<\/td\>(.*)$/is) {
						($title, $accesses, $bytes, $data) = ($1, $2, $3, $4);
						$title =~ s/^\s+//;
						$title =~ s/\s+$//;
						$accesses{$title} = $accesses;
						$bytes{$title} = $bytes;
					}			
					next;	
				}
				# Wusage 7 style
				while ($data =~ /\<tr\>\s+\<td\>(.*?)\<\/td\>\s+\<td.*?\>\s*([\d\,\.]+)\s*\<\/td\>\s+\<td.*?\>\s*([\d\,\.]+)\s*\<\/td\>\s+\<td.*?\>\s*([\d\,\.]+)\s*\<\/td\>(.*)$/is) {
					($title, $accesses, $bytes, $visits, $data) = ($1, $2, $3, $4, $5);
					$title =~ s/^\s+//;
					$title =~ s/\s+$//;
					$accesses{$title} = $accesses;
					$bytes{$title} = $bytes;
					$visits{$title} = $visits;
				}
				print OUT "$month/$day/$year\t", &length($f, $month, $day, $year);
				for $p (@patterns) {
					$a = $accesses{$titles{$p}};
					$b = $bytes{$titles{$p}};
					$v = $visits{$titles{$p}};
					$a = sprintf("%20.02f", $a);
					$b = sprintf("%20.02f", $b);
					$v = sprintf("%20.02f", $v);
					print OUT "\t$a\t$b\t$v";
				}
				print OUT "\n";
			}	
		}
	}
	close(OUT);
	closedir(DIR);	
}

if (!$found) {
	die "Didn't find any of the report subdirectories. You are probably\n" .
		"not in your report directory. cd there first.\n";
}

sub length
{
	my($f, $month, $day, $year) = @_;
	if ($f eq "daily") {
		return 1;
	} elsif ($f eq "weekly") {
		return 7;
	} elsif ($f eq "monthly") {
		return 31;
	} elsif ($f eq "quarterly") {
		return 93;
	} elsif ($f eq "annual") {
		return 365;
	}
}