#!/usr/bin/perl -w # ggh - grovel global history in netscape logs $USAGE = < 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; # 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; }