#!/usr/bin/perl -T
#
# vlogger - smarter logging for apache
# steve j. kondik  <shade@chemlab.org>
#
# this script will take piped logs in STDIN, break off the first component
# and log the line into the proper directory under $LOGDIR.  it will roll the
# logs over at midnight on-the-fly and maintain a symlink to the most recent log.
#
#
# This library is free software; you can redistribute it and/or
# modify it under the terms of the GNU Library General Public
# License as published by the Free Software Foundation; either
# version 2 of the License, or (at your option) any later version.
#
# This library is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
# Library General Public License for more details.
#
# 
# CHANGELOG:
# 0.1	initial release
# 0.2	cleanups, added -e option for errorlogs, added strict stuff
# 0.3	cleanups, bugfixes, docs, added -r size rotation option
# 0.4   added dbi usage tracking option, code cleanups from cz@digitalfreaks.org
# 1.0	small bugfixes, first production release
# 1.1   bugfix release
# 1.2   support for mod_logio
# 1.3	various contributed bugfixes
# 1.4   automatically creates two levels of subdirs.
# 1.4p1 implemented anonymization of ipv4 and ipv6
# 1.4p2 implemented anonymization in error log part
# 1.4p3 implemented stronger anonymization
# 1.4p4 implemented re-opening on externally rotated error.log
# 1.4p5 backported old ISPConfig changes (improved DBI error handling, naming)
#
# 
# TODO:
# configurable file compression using Compress::Zlib, maybe.
# 

package vlogger;

$ENV{PATH} = "/bin:/usr/bin";

my $VERSION = "1.4p5"; # modified for ISPConfig

=head1 NAME

vlogger - flexible log rotation and usage tracking in perl

=head1 SYNOPSIS

vlogger [OPTIONS]... [LOGDIR]

=head1 DESCRIPTION

Vlogger is designed to make webserver log rotation simple and easy to manage.
It deals with VirtualHost logs automatically, so only one directive is required
to manage all hosts on a webserver.  Vlogger takes piped output from Apache or
another webserver, splits off the first field, and writes the logs to logfiles
in subdirectories.  It uses a filehandle cache to avoid resource limitations.
It will start a new logfile at the beginning of a new day, and optionally start
new files when a certain filesize is reached.  It can maintain a symlink to
the most recent log for easy access.  Optionally, host parsing can be disabled
for use in ErrorLog directives.  

To use vlogger, you need to add a "%v" to the first part of your LogFormat:

LogFormat "%v %h %l %u %t \"%r\" %>s %b \"%{Referer}i\" \"%{User-Agent}i\"" combined

Then call it from a customlog:

CustomLog "| /usr/local/sbin/vlogger -s access.log -u www-logs -g www-logs /var/log/apache" combined

=head1 OPTIONS

Options are given in short format on the command line.

-a
	Do not autoflush files.  This may improve performance but may break logfile
analyzers that depend on full entries in the logs.

-e
	ErrorLog mode.  In this mode, the host parsing is disabled, and the file is
written out using the template under the specified LOGDIR.
	
-n
	Disables rotation.  This option disables rotation altogether.

-f MAXFILES
	Maximum number of filehandles to keep open.  Defaults to 100.  Setting this
value too high may result in the system running out of file descriptors.
Setting it too low may affect performance.

-u UID
	Change user to UID when running as root.

-g GID
	Change group to GID when running as root.

-t TEMPLATE
	Filename template using Date::Format codes.  Default is "%m%d%Y-access.log",
or "%m%d%Y-error.log".  When using the -r option, the default becomes
"%m%d%Y-%T-access.log" or "%m%d%Y-%T-error.log".

-s SYMLINK
	Specifies the name of a symlink to the current file.

-r SIZE
	Rotate files when they reach SIZE.  SIZE is given in bytes.

-d CONFIG
	Use the DBI usage tracker.

-p
	Privacy mode (delete last octet of ipv4 or last 4 sections of ipv6)

-P
	Extended privacy mode (delete 2 octets ipv4 or 5 sections ipv6)

-h
	Displays help.

-v
	Prints version information.

=head1 DBI USAGE TRACKER

	Vlogger can automatically keep track of per-virtualhost usage statistics in a
database.  DBI and the relevant drivers (eg. DBD::mysql) needs to be installed for
this to work.  Create a table in your database to hold the data.  A "mysql_create.sql"
script is provided for using this feature with MySQL.  Configure the dsn, user, pass
and dump values in the vlogger-dbi.conf file.  The "dump" parameter controls how often
vlogger will dump its stats into the database (the default is 30 seconds).  Copy this
file to somewhere convienient on your filesystem (like /etc/apache/vlogger-dbi.conf) and
start vlogger with "-d /etc/apache/vlogger-dbi.conf".  You might want to use this feature
to easily bill customers on a daily/weekly/monthly basis for bandwidth usage.

=head1 SEE ALSO
cronolog(1), httplog(1)

=head1 BUGS
None, yet.

=head1 AUTHORS
Steve J. Kondik <shade@chemlab.org>
	
WWW: http://n0rp.chemlab.org/vlogger
	
=cut

# a couple modules we need
use strict;
no strict "refs";
use warnings;
use sigtrap qw(handler exitall HUP USR1 TERM INT PIPE);
use Date::Format;
use Getopt::Std;
use IO::Handle;
use File::Path qw(make_path);
use File::Basename;

# get command line options
our %OPTS;
getopts( 'f:t:s:hu:g:aeinpPvr:d:', \%OPTS );

# print out version
if ( $OPTS{'v'} ) {
    print "VLogger $VERSION (apache logfile parser)\n";
    print "Written by Steve J. Kondik <shade\@chemlab.org>\n\n";
    print "This is free software; see the source for copying conditions.  There is NO\n";
    print "warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n";
    exit;
}

# print help
if ( $OPTS{'h'} || !$ARGV[0] ) {
    usage();
    exit;
}

# log directory
my $LOGDIR;
if ( $ARGV[0] ) {
    if ( !-d $ARGV[0] || -l $ARGV[0] ) {
        print STDERR
          "[vlogger] target directory $ARGV[0] does not exist or is a symlink - exiting.\n\n";
        exit;
    }
    $LOGDIR = $ARGV[0];
}
$LOGDIR =~ /(.*)/;
$LOGDIR = $1;

# change uid/gid if requested (and running as root)
if ( $> == "0" ) {
    if ( $OPTS{'g'} ) {
        my $guid = getgrnam( $OPTS{'g'} );
        if ( !defined $guid || $guid == 0 ) {
            print STDERR
              "[vlogger] cannot run as root or nonexistant group.\n\n";
            exit;
        }

        $) = $guid;
        $( = $guid;
        if ( $) != $guid and $) != ( $guid - 2**32 ) ) {
            die "fatal: setgid to gid $guid failed\n";
        }
    }

    if ( $OPTS{'u'} ) {
        my $uuid = getpwnam( $OPTS{'u'} );
        if ( !defined $uuid || $uuid == 0 ) {
            print STDERR
              "[vlogger] cannot run as root or nonexistant user.\n\n";
            exit;
        }

        $> = $uuid;
        $< = $uuid;
        if ( $> != $uuid and $> != ( $uuid - 2**32 ) ) {
            die "fatal: setuid to uid $uuid failed\n";
        }
    }
}

# set up dbi stuffs

my $DBI_DSN;
my $DBI_USER;
my $DBI_PASS;
my $DBI_DUMP;
if ( $OPTS{'d'} ) {
    if ( $OPTS{'e'} ) {
        print "-d not valid with -e.  exiting.\n";
        exit;
    }

    eval "use DBI";

    open CONF, $OPTS{'d'};
    while (<CONF>) {
        chomp;
        my @conf = split (/\s/);
        if ( $conf[0] eq "dsn" ) {
            $DBI_DSN = $conf[1];
        }
        elsif ( $conf[0] eq "user" ) {
            $DBI_USER = $conf[1];
        }
        elsif ( $conf[0] eq "pass" ) {
            $DBI_PASS = $conf[1];
        }
        elsif ( $conf[0] eq "dump" ) {
            $DBI_DUMP = $conf[1];
        }
    }
    close CONF;

    unless ( $DBI_DSN && $DBI_USER && $DBI_PASS && $DBI_DUMP ) {
        print "All values for DBI configuration are not properly defined.\n\n";
        exit;
    }

    # test the connection
    eval {
        my $dbh = DBI->connect( $DBI_DSN, $DBI_USER, $DBI_PASS )
          or die "DBI Error: $!";
        $dbh->disconnect;
    };
    if($@) {
        print "MySQL Connection problem\n";
    }

    # SIGALRM dumps the tracker hash
    $SIG{ALRM} = \&dump_tracker;

    alarm $DBI_DUMP;

}

# max files to keep open
my $MAXFILES;
if ( $OPTS{'f'} ) {
    $MAXFILES = $OPTS{'f'};
}
else {
    $MAXFILES = "100";
}

# filesize rotation
my $MAXSIZE;
if ( $OPTS{'r'} ) {
    $MAXSIZE = $OPTS{'r'};
}

# filename template
my $TEMPLATE;
if ( $OPTS{'t'} ) {
    $TEMPLATE = $OPTS{'t'};
    $TEMPLATE =~ /(.*)/;
    $TEMPLATE = $1;

}
elsif ( $OPTS{'e'} ) {
    if ( $OPTS{'r'} ) {
        $TEMPLATE = "%m%d%Y-%T-error.log";
    }
    else {
        $TEMPLATE = "%m%d%Y-error.log";
    }
}
else {
    if ( $OPTS{'r'} ) {
        $TEMPLATE = "%m%d%Y-%T-access.log";
    }
    else {
        $TEMPLATE = "%m%d%Y-access.log";
    }
}

# symlink
if ( $OPTS{'s'} ) {
    $OPTS{'s'} =~ /(.*)/;
    $OPTS{'s'} = $1;
}

# chroot to the logdir
chdir($LOGDIR);
#chroot("."); #we better do not chroot as DBI requires to load a module on the fly -> error!

my %logs     = ();
my %tracker  = ();
my $LASTDUMP = time();

# pick a mode
if ( $OPTS{'e'} ) {

    $0 = "vlogger (error log)";
    # errorlog mode
    open ELOG, ">>" . time2str( $TEMPLATE, time() )
      or die ( "can't open $LOGDIR/" . time2str( $TEMPLATE, time() ) );

    unless ( $OPTS{'a'} ) {
        ELOG->autoflush(1);
    }
    if ( $OPTS{'s'} ) {
        if ( -l $OPTS{'s'} ) {
            unlink( $OPTS{'s'} );
        }
        symlink( time2str( $TEMPLATE, time() ), $OPTS{'s'} );
    }

    my $LASTWRITE = time();

    while ( my $log_line = <STDIN> ) {
        unless ( $OPTS{'n'} ) {
            if ( time2str( "%Y%m%d", time() ) >
                time2str( "%Y%m%d", $LASTWRITE ) )
            {

                # open a new file
                close ELOG;
                open_errorlog();
            }
            elsif ( $OPTS{'r'} ) {

                # check the size
                my @filesize = ELOG->stat;
                print $filesize[7] . "\n";
                if ( $filesize[7] > $MAXSIZE ) {
                    close ELOG;
                    open_errorlog();
                }
            }

            $LASTWRITE = time();
        }

#        my ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,
#            $atime,$mtime,$ctime,$blksize,$blocks)
#               = stat(time2str( $TEMPLATE, time() ));
        my @fstat = ELOG->stat();
        my $nlinks = $fstat[3];
        if($nlinks < 1) {
			close ELOG;
			open ELOG, ">>" . time2str( $TEMPLATE, time() )
				or die ( "can't open $LOGDIR/" . time2str( $TEMPLATE, time() ) );
		}

        if ( $OPTS{'P'} ) {
			$log_line =~ s/^(.*?\[client\s+\d+\.\d+)\.\d+\.\d+((?::\d+)?\]\s+.*)/$1.0.0$2/g;
			$log_line =~ s/^(.*?\[client\s+[a-f0-9]*:[a-f0-9]*:[a-f0-9]*)(?::[a-f0-9]*){2,5}((?::\d+)?\]\s+.*)/$1::0$2/g;
		} elsif ( $OPTS{'p'} ) {
			$log_line =~ s/^(.*?\[client\s+\d+\.\d+\.\d+)\.\d+((?::\d+)?\]\s+.*)/$1.0$2/g;
			$log_line =~ s/^(.*?\[client\s+[a-f0-9]*:[a-f0-9]*:[a-f0-9]*:[a-f0-9]*)(?::[a-f0-9]*){1,4}((?::\d+)?\]\s+.*)/$1::0$2/g;
        }

        # we dont need to do any other parsing at all, so write the line.
        print ELOG $log_line;
    }

}
else {

    # accesslog mode
    $0 = "vlogger (access log)";
    while ( my $log_line = <STDIN> ) {

        
		if ( $OPTS{'P'} ) {
			$log_line =~ s/^(\S*\s+\d+\.\d+)\.\d+\.\d+(\s+.*)/$1.0.0$2/g;
			$log_line =~ s/^(\S*\s+[a-f0-9]*:[a-f0-9]*:[a-f0-9]*)(?::[a-f0-9]*){2,5}(\s+.*)/$1::0$2/g;
		} elsif($OPTS{'p'}) {
			$log_line =~ s/^(\S*\s+\d+\.\d+\.\d+)\.\d+(\s+.*)/$1.0$2/g;
			$log_line =~ s/^(\S*\s+[a-f0-9]*:[a-f0-9]*:[a-f0-9]*:[a-f0-9]*)(?::[a-f0-9]*){1,4}(\s+.*)/$1::0$2/g;
		}
        
        # parse out the first word (the vhost)
        my @this_line = split ( /\s/, $log_line );
        my ($vhost) = $this_line[0];
        my $reqsize = $this_line[10];
        $vhost = lc($vhost) || "default";
        if ( $vhost =~ m#[/\\]# ) { $vhost = "default" }
        $vhost =~ /(.*)/o;
        $vhost = $1;
        $vhost = 'default' unless $vhost;

        if ( $OPTS{'i'} ) {
            $reqsize = $this_line[1] + $this_line[2];
        }
        
        # if we're writing to a log, and it rolls to a new day, close all files.
        unless ( $OPTS{'n'} ) {
            if ( $logs{$vhost}
                && ( time2str( "%Y%m%d", time() ) >
                    time2str( "%Y%m%d", $logs{$vhost} ) ) )
            {
                foreach my $key ( keys %logs ) {
                    close $key;
                }
                %logs = ();
            }
            elsif ( $OPTS{'r'} && $logs{$vhost} ) {

                # check the size
                my @filesize = $vhost->stat;
                if ( $filesize[7] > $MAXSIZE ) {
                    close $vhost;
                    delete( $logs{$vhost} );
                }
            }
        }

        # open a new log
        if ( !$logs{$vhost} ) {

            # check how many files we have open, close the oldest one
            if ( keys(%logs) > $MAXFILES ) {
                my ( $key, $value ) =
                  sort { $logs{$a} <=> $logs{$b} } ( keys(%logs) );
                close $key;
                delete( $logs{$key} );
            }

            my $filename = "${vhost}/" . time2str( $TEMPLATE, time() );
            my $filepath = dirname($filename);

            # check if directory is there
            unless ( -d $filepath ) {
                print "mkdir: " . $filepath . "\n";
                make_path($filepath)
                  or die ( "can't mkdir $filepath" );
            }
            
            # Dont log to symlinks
            if( -l "${vhost}/".time2str( $TEMPLATE, time() ) ) {
                die("Log target is a symlink: $LOGDIR/${vhost}/".time2str( $TEMPLATE, time() ));
            }

            # open the file using the template
            open $vhost, ">>".$filename
              or die ( "can't open $filename" );

            # autoflush the handle unless -a
            if ( !$OPTS{'a'} ) {
                $vhost->autoflush(1);
            }

            # make a symlink if -s
            if ( $OPTS{'s'} ) {
                chdir("${vhost}");
                if ( -l $OPTS{'s'} ) {
                    unlink( $OPTS{'s'} );
                }
                symlink( time2str( $TEMPLATE, time() ), $OPTS{'s'} );
                chdir("..");
            }
        }

        # update the timestamp and write the line
        $logs{$vhost} = time();
        if ($OPTS{'i'}) {
            $log_line =~ s/^\S*\s+\S*\s+\S*\s+//o;
        }
        else {
            $log_line =~ s/^\S*\s+//o;
        }

        if ($reqsize ne "-") {
			  if ( $reqsize =~ m/\d|/ && $reqsize > 0 ) {
   	         $tracker{$vhost} += $reqsize;
      	  }
		  }

        print $vhost $log_line;

    }
}

# sub to close all files
sub closeall {
    if ( $OPTS{'e'} ) {
        close ELOG;
    }
    else {
        foreach my $key ( keys %logs ) {
            close $key;
        }
        %logs = ();
        if ( $OPTS{'d'} ) {
            vlogger::dump_tracker();
        }
    }
}

sub exitall {
    vlogger::closeall;
    exit;
}

# sub to open new errorlog
sub open_errorlog {
    open ELOG, ">>" . time2str( $TEMPLATE, time() )
      or die ( "can't open $LOGDIR/" . time2str( $TEMPLATE, time() ) );
    if ( $OPTS{'s'} ) {
        if ( -l $OPTS{'s'} ) {
            unlink( $OPTS{'s'} );
        }
        symlink( time2str( $TEMPLATE, time() ), $OPTS{'s'} );
    }

    # autoflush it unless -a
    unless ( $OPTS{'a'} ) {
        ELOG->autoflush(1);
    }
}

# sub to update the database with the tracker data
sub dump_tracker {
    eval {
    if ( keys(%tracker) > 0 ) {
        my $dbh = DBI->connect( $DBI_DSN, $DBI_USER, $DBI_PASS )
          or warn "DBI Error: $!";
        foreach my $key ( keys(%tracker) ) {
            my $ts = time2str( "%Y-%m-%d", time() );
            my $sth =
              $dbh->prepare( "select * from web_traffic where hostname='" . $key
                . "' and traffic_date='" . $ts . "'" );
            $sth->execute;
            if ( $sth->rows ) {
                my $query =
                  "update web_traffic set traffic_bytes=traffic_bytes+"
                  . $tracker{$key}
                  . " where hostname='" . $key
                  . "' and traffic_date='" . $ts . "'";
                $dbh->do($query);
            }
            else {
                my $query = "insert into web_traffic (hostname, traffic_date, traffic_bytes) values ('$key', '$ts', '$tracker{$key}')";
                $dbh->do($query);
            }
        }
        $dbh->disconnect;
        %tracker = ();
    }
    alarm $DBI_DUMP;
    };
    
    if ($@) {
        print "Unable to store vlogger data in database\n";
    }
}

# print usage info
sub usage {
    print "Usage: vlogger [OPTIONS]... [LOGDIR]\n";
    print "Handles a piped logfile from a webserver, splitting it into it's\n";
    print "host components, and rotates the files daily.\n\n";
    print "  -a                          do not autoflush files\n";
    print "  -e                          errorlog mode\n";
    print "  -n                          don't rotate files\n";
    print "  -f MAXFILES                 max number of files to keep open\n";
    print "  -u UID                      uid to switch to when running as root\n";
    print "  -g GID                      gid to switch to when running as root\n";
    print "  -t TEMPLATE                 filename template (see perldoc Date::Format)\n";
    print "  -s SYMLINK                  maintain a symlink to most recent file\n";
    print "  -r SIZE                     rotate when file reaches SIZE\n";
    print "  -d CONFIG                   use DBI usage tracker (see perldoc vlogger)\n";
    print "  -p                          Privacy mode (delete last octet of ipv4 or\n";
    print "                              last 4 sections of ipv6)\n";
    print "  -P                         Extended privacy mode (delete 2 octets ipv4\n";
    print "                              or last 5 sections of ipv6\n";
    print "  -i                          extract mod_logio instead of filesize\n";
    print "  -h                          display this help\n";
    print "  -v                          output version information\n\n";
    print "TEMPLATE may be a filename with Date::Format codes.  The default template\n";
    print "is %m%d%Y-access.log.  SYMLINK is the name of a file that will be linked to\n";
    print "the most recent file inside the log directory.  The default is access.log.\n";
    print "MAXFILES is the maximum number of filehandles to cache.  This defaults to 100.\n";
    print "When running with -a, performance may improve, but this might confuse some\n";
    print "log analysis software that expects complete log entries at all times.\n";
    print "Errorlog mode is used when running with an Apache errorlog.  In this mode,\n";
    print "virtualhost parsing is disabled, and a single file is written in LOGDIR\n";
    print "using the TEMPLATE (%m%d%Y-error.log is default for -e).  When running with\n";
    print "-r, the template becomes %m%d%Y-%T-xxx.log.  SIZE is given in bytes.\n\n";
    print "Report bugs to <shade\@chemlab.org>.\n";
}