#!/usr/bin/perl
# bad-links.pl
# read the error list and create the error reports
# if the -s argument is specified, error reports are sent to individual users
#  via email

$DOCROOT='/www/data';
$TEMP='/tmp/bad-links.tmp';
$ADMINUSER='beecher';			# send root & unknown reports to this user
$ERRCOLUMN=58;

# users who said "no, thanks" (they will not get any error reports)
$no_thanks='-thor-mo107-jmg72-bcw6-mgz1-chh1-ly38-';

chop($SORT=`which sort`);
die("can't find sort\n") if $SORT =~ /^no /;

$MAILPROG='/usr/lib/sendmail';
die("$MAILPROG not found\n") unless -x $MAILPROG;

$send='';
while ($ARGV[0]) {
    if ($ARGV[0] eq '-s') {		# -s means send mail
	shift;
	$send=1;
    } else {
	print "unexpected option: @ARGV\n";
	exit;
    }
}

&write_temp_file;

&get_header;
$prev_group  = '';
$prev_user   = '';
$prev_req    = '';
$prev_url    = '';
$prev_ecode  = '';
$prev_ecount = '';
$mopen       = '';
$mailsent    = 0;

if (!open(SOR, "$SORT $TEMP |")) {
    die "can't sort $TEMP: $!\n";
}

while (<SOR>) {
    chop;
    ($group, $user, $person, $req, $url, $ecode, $ecount, $time1, $time2) = split /\t/;
    if    ($group  ne $prev_group)	{ &new_group; }
    elsif ($user   ne $prev_user)	{ &new_user;  }
    elsif ($req    ne $prev_req)	{ &new_req;   }
    elsif ($url    ne $prev_url)	{ &new_url;   }
    else				{ &new_ecode; }

    $prev_group  = $group;
    $prev_user   = $user;
    $prev_req    = $req;
    $prev_url    = $url;
}
close(SOR);
#unlink($TEMP);


if ($mopen) {
    close(MAIL);
    $mopen='';
}
print STDERR "\n$mailsent messages sent\n";


########################################

sub write_temp_file {
# read the error file created by getstats.  here is a sample record:
# /acis/rad/index.html	/~beecher/99503.html	404	2	20/Dec/1997:11:36       20/Dec/1997:14:30

# lookup the user and group, creating the temp file
    local($fromcgi, $tocgi, $mailto, $badfile, $goodfile) = (0,0,0,0,0);
    if (!open(TMP, ">$TEMP")) {
	die "can't create $TEMP: $!\n";
    }
    while (<STDIN>) {
	chop;
	($req, $url, $rest) = split(/\t/, $_, 3);
	if (($req =~ m|^/cgi-bin|) ||		# ignore bad links from CGI
	    ($req =~ m|^/sec-cgi-bin|)) {
	  $fromcgi++;
	  next;
	}
	if (($url =~ m|^/cgi-bin|) ||		# ignore bad links to CGI
	    ($url =~ m|^/sec-cgi-bin|)) {
	  $tocgi++;
	  next;
	}
	if ($url =~ /mailto:/) {		# ignore links to mailto:
	  $mailto++;
	  next;
	}
	@fileinfo = stat("$DOCROOT$req");	# stat the file
	$group    = 'Unknown';
	$user     = $ADMINUSER;
	$person   = 'Unknown';
	if ($fileinfo[1] &&			# file exists
	  ($fileinfo[4]!=-1)) {			# uid is not -1
	  @userinfo=getpwuid($fileinfo[4]);	# username and group
	  $user=$userinfo[0] if $userinfo[0];
	  $group=getgrgid($userinfo[3]);
	  $person = (split(',',$userinfo[6],2))[0];
	  $person =~ s/  / /g;
	  $person = 'Unknown' unless $person;

	  print TMP join("\t", $group, $user, $person, $req, $url, $rest)."\n";
	  $goodfile++;
	} else {
	  $badfile++;
	}
    }
    close(TMP);
    print STDERR "$fromcgi links from CGI files ignored\n";
    print STDERR "$tocgi links to CGI files ignored\n";
    print STDERR "$mailto mailto links ignored\n";
    print STDERR "$badfile files not found\n";
    print STDERR "$goodfile files included in report\n";
}


sub new_group {
    print "\n\n-------- GROUP $group --------";
    &new_user;
}

sub new_user {
    local($muser);
    if ($mopen) {
	close(MAIL);
	$mopen='';
    }
    if (!$send ||			# -s command line option
	($no_thanks =~ /-$user-/)) {	# no mail for these users
	print "\n\nReport for $person <$user>\n";
	&new_req;
	return;
    }
    $muser=$user;
    $muser=$ADMINUSER if $user eq 'root';

    print "\n\nSending mail to $person <$muser>\n";
    die("can't run $MAILPROG: $!\n")
		unless open(MAIL, "|$MAILPROG -t");
    $mopen=1;
    $mailsent++;
    print MAIL "To: $muser
Subject: web error report
$header
      *** Web Error Report for $person <$user> ***
";    
    &new_req;
}

sub new_req {
    local($request) = $req;
    $request =~ s/#.*//;
    print MAIL "\n$request\n" if $mopen;
    print "\n$request\n";
    &new_url;
}

sub new_url {
    local($msg)= "     bad link to $url\n";
    print MAIL $msg if $mopen;
    print $msg;
    &new_ecode;
}

sub new_ecode {
    local($errmsg);
    if    ($ecode eq '400') { $errmsg='Bad syntax'; }
    elsif ($ecode eq '401') { $errmsg='Access denied'; }
    elsif ($ecode eq '403') { $errmsg='Protected'; }
    elsif ($ecode eq '404') { $errmsg='Not found'; }
    elsif ($ecode eq '405') { $errmsg='Method not allowed'; }
    elsif ($ecode eq '408') { $errmsg='Timeout'; }
    elsif ($ecode eq '500') { $errmsg='Server error'; }
    elsif ($ecode eq '501') { $errmsg='Not implemented'; }
    else                    { $errmsg=$errcode; }
    if ($ecount==1) {
	print MAIL "                  $errmsg at $time1\n" if $mopen;
	print      "                  $errmsg at $time1\n";
    } else {
	print MAIL "                  $errmsg [$ecount times] $time1 to $time2\n" if $mopen;
	print      "                  $errmsg [$ecount times] $time1 to $time2\n";
    }
}

sub get_header {			# you can put informative messages here
    $header="
      *** About the Weekly Error Report for Web Pages ***

Here is your web error report for the past week.

";
}