home | O'Reilly's CD bookshelfs | FreeBSD | Linux | Cisco | Cisco Exam  


14.11. Program: ggh - Grep Netscape Global History

This program divulges the contents of Netscape's history.db file. It can be called with full URLs or with a (single) pattern. If called without arguments, it displays every entry in the history file. The ~/.netscape/history.db file is used unless the -database option is given.

Each output line shows the URL and its access time. The time is converted into localtime representation with -localtime (the default), gmtime representation with -gmtime  - or left in raw form with -epochtime , which is useful for sorting by date.

To specify a pattern to match against, give one single argument without a :// .

To look up one or more URLs, supply them as arguments:

% ggh http://www.perl.com/index.html

To find out a link you don't quite recall, use a regular expression (a single argument without a :// is a pattern):

% ggh perl

To find out all the people you've mailed:

% ggh mailto:

To find out the FAQ sites you've visited, use a snazzy Perl pattern with an embedded /i modifier:

% ggh -regexp '(?i)\bfaq\b'

If you don't want the internal date converted to localtime, use -epoch :

% ggh -epoch http://www.perl.com/perl/

If you prefer gmtime to localtime, use -gmtime :

% ggh -gmtime http://www.perl.com/perl/

To look at the whole file, give no arguments (but perhaps redirect to a pager):

% ggh | less

If you want the output sorted by date, use the -epoch flag:

% ggh -epoch | sort -rn | less

If you want it sorted by date into your local time zone format, use a more sophisticated pipeline:

% ggh -epoch | sort -rn | perl -pe 's/\d+/localtime $&/e' | less

The Netscape release notes claim that they're using NDBM format. This is misleading: they're actually using Berkeley DB format, which is why we require DB_File (not supplied standard with all systems Perl runs on) instead of NDBM_File (which is). The program is shown in Example 14.8 .

Example 14.8: ggh

#!/usr/bin/perl -w
# ggh -- grovel global history in netscape logs
$USAGE = <<EO_COMPLAINT;
usage: $0 [-database dbfilename] [-help]
           [-epochtime | -localtime | -gmtime]
           [ [-regexp] pattern] | href ... ]
EO_COMPLAINT

use Getopt::Long;

($opt_database, $opt_epochtime, $opt_localtime,
 $opt_gmtime,   $opt_regexp,    $opt_help,
 $pattern,                                  )      = (0) x 7;

usage() unless GetOptions qw{ database=s
                              regexp=s
                              epochtime localtime gmtime
                              help
                            };

if ($opt_help) { print $USAGE; exit; }

usage("only one of localtime, gmtime, and epochtime allowed")
    if $opt_localtime + $opt_gmtime + $opt_epochtime > 1;

if ( $opt_regexp ) {
    $pattern = $opt_regexp;
} elsif (@ARGV && $ARGV[0] !~ m(://)) {
    $pattern = shift;
}

usage("can't mix URLs and explicit patterns")
    if $pattern && @ARGV;

if ($pattern && !eval { '' =~ /$pattern/; 1 } ) {
    $@ =~ s/ at \w+ line \d+\.//;
    die "$0: bad pattern $@";
}

require DB_File; DB_File->import();  # delay loading until runtime
$| = 1;                              # feed the hungry PAGERs

$dotdir  = $ENV{HOME}    || $ENV{LOGNAME};
$HISTORY = $opt_database || "$dotdir/.netscape/history.db";

die "no netscape history dbase in $HISTORY: $!" unless -e $HISTORY;
die "can't dbmopen $HISTORY: $!" unless dbmopen %hist_db, $HISTORY, 0666;

# the next line is a hack because the C programmers who did this
# didn't understand strlen vs strlen+1.  jwz told me so. :-)
$add_nulls   = (ord(substr(each %hist_db, -1)) == 0);

# XXX: should now do scalar keys to reset but don't 
#      want cost of full traverse, required on tied hashes.
#   better to close and reopen?

$nulled_href = "";  
$byte_order  = "V";         # PC people don't grok "N" (network order)
    
if (@ARGV) {
    foreach $href (@ARGV) {
        $nulled_href = $href . ($add_nulls && "\0");
        unless ($binary_time = $hist_db{$nulled_href}) {
            warn "$0: No history entry for HREF $href\n";
            next;
        }
        $epoch_secs = unpack($byte_order, $binary_time);
        $stardate   = $opt_epochtime ? $epoch_secs
                                     : $opt_gmtime ? gmtime    $epoch_secs
                                                   : localtime $epoch_secs;
        print "$stardate $href\n";
    }
} else {
    while ( ($href, $binary_time) = each %hist_db ) {
        chop $href if $add_nulls;
        next unless defined $href && defined $binary_time;
        # gnat reports some binary times are missing
        $binary_time = pack($byte_order, 0) unless $binary_time;
        $epoch_secs = unpack($byte_order, $binary_time);
        $stardate   = $opt_epochtime ? $epoch_secs
                                     : $opt_gmtime ? gmtime    $epoch_secs
                                                   : localtime $epoch_secs;
        print "$stardate $href\n" unless $pattern && $href !~ /$pattern/o;
    }
}

sub usage {
    print STDERR "@_\n" if @_;
    die $USAGE;
}