# Blosxom Plugin: clicktrack
# Author: Nelson Minar <nelson@monkey.org>
# Version: 20030524
# http://www.nelson.monkey.org/~nelson/weblog/
# License: Public Domain
# Thanks to Todd Larason http://molelog.molehill.org/blox/

# The clicktrack plugin helps you track clicks to offsite links.
# It does two things:
#
# The start method looks for urls of the form blosxom.cgi?redirectURL=...
# If this URL is present it overrides Blosxom's normal blog behaviour
# and instead tells the browser to redirect to the specified URL.
#
# The story method scans your stories for offsite URLs and replaces
# them with links to the blosxom redirect URL. The links have
# a bit of Javascript to keep the status window tidy. Example:
# <a href="http://www.nelson.monkey.org/cgi-bin/test.cgi?redirectURL=http%3A%2F%2Fmolelog.molehill.org%2Fblox%2F" 
#   onMouseOver="window.status='to http://molelog.molehill.org/blox/'; return true;" 
#   onMouseOut="window.status=''; return true;">Todd</a>
#
# To install simply drop it in your plugins dir. This script wants to
# run early, so in most cases you should name it "00clicktrack". I encourage
# you to disclose you're running this on your blog - let users know!
# By default we don't escape URLs - breaks the CGI standard, but makes
# prettier URLs. You can turn on escaping below.
#
# Possible features:
#   Log the clicks. I find it simpler to just look at my access.log
#   Option to set a cookie to opt out of this tracking

package clicktrack;

# Configurable option - escape URLs. Escaping is more correct, but ugly.
$escapeURLs = 0;

use CGI qw/:standard/;

$query;

# If the query contains redirectURL, override Blosxom and send the 302
sub start {
    $query = new CGI;

    # Hunt for &redirectURL= in the query string
    if ($escapeURLs) {
        $url = $query->param('redirectURL');
    } else {
        if ($ENV{QUERY_STRING} =~ m/redirectURL=(.*)$/) {
            $url = $1;
        }
    }

    # Issue redirect
    if ($url) {
        print "Status: 302 Moved Temporarily\r\n",
              "Location: $url\r\n",
              "Content-Type: text/html\r\n\r\n",
              "<a href=\"$url\">$url</a>\r\n";

        exit 0;         # Not using skip because we're subverting Blosxom
    }
    1;
}

# Rewrite the href of the anchor to route through our click tracking
sub rewriteHref {
    my ($tag) = @_;

    # Extract the href (skip if we don't find one)
    if (!($tag =~ m/^(.*)href="([^"]+)"(.*)$/is)) {
        return $tag;
    }
    $prehref = $1;
    $href = $2;
    $posthref = $3;

    # Check that the href goes to an offsite http URL
    return $tag if !($href =~ m/^http/i);

    $msg = $href;
    $msg =~ s%^http://%%;
    $msg = "to $msg";
    if ($escapeURLs) {
        $redirURL = CGI::escape($href);
    } else {
        $redirURL = $href;
    }
    # Rewrite the link to include our own clicktrack URL
    return $prehref . 
           "href=\"" . $query->url() .
           "?redirectURL=" . $redirURL . "\" " . 
           "onMouseOver=\"window.status='" . $msg . "'; return true;\" " .
           "onMouseOut=\"window.status=''; return true;\"" .
           "$posthref";
}

# Modify the body of stories by rewriting offsite hrefs to our own redir
sub story {
    my($pkg, $path, $filename, $story_ref, $title_ref, $body_ref) = @_;
    # Rewrite any <a> tag in the post body
    $$body_ref =~ s/(<a\s[^>]+>)/rewriteHref($1)/geis;
    1;
}

1;