#!/usr/bin/perl -w
# read dcache4.url from Opera logs, write links and time.

use Time::ParseDate;
use POSIX qw(strftime);
use strict;

my $fh;
my $t;
my $url = '';
my $date = '';

#my $file = 'dcache4.url';     # TESTING
my $file = shift || die "usage: $0 file\n";

# Get today's date starting at midnight.
my $today = time();
my ($s, $m, $h) = (localtime($today))[0, 1, 2];
my $offset = ($h * 3600) + ($m * 60) + $s;
$today -= $offset;

# Read the strings from the URL cache file.

open($fh, "/bin/strings $file |");

while (<$fh>) {
    chomp;

    # order: link, then *first* date string, then "opr".

    if (m!^.(.*://.*)!) {
        $url = $1;
        $date = '';
    }

    elsif (m!^(.* GMT)!) {
        if ($t = parsedate($1)) {
            if ($t >= $today) {
                $date = strftime("%T", localtime($t));
            }
        }
    }

    elsif (m!^opr! || m!\(!) {
        if ($url =~ m/\.(png|css|jpg|gif)$/) {
            $url = '';
        }

        if ($url =~ m!http://example.com/!) {
            $url = '';
        }

        if (length($url)) {
            $date = '??:??:??' unless length($date);
            print "$date $url\n";
            $url = $date = '';
        }
    }
}

close($fh);
exit(0);