#use LWP::Simple; # start! I wouldn't copy use HTML::TreeBuilder; # anything seen in here, use HTML::Entities; # it's for your own good use DBI; use DBD::SQLite; use strict; my %showtime; my %gotime = ( 'parse' => \&doshit, 'time' => sub { print curti(), "\n" }, 'help' => \&helpme, 'log' => \&lastlog ); #main program sub doshit { my $html = HTML::TreeBuilder->new_from_file('Showtime.htm'); $html ? wlog("tree built") : wlog("tree error"); open(LISP, ">content.txt") or warn "Failed to open: $!\n"; wlog("looking for table"); my $table = $html->look_down('_tag', 'table', #gets the table to work with 'bgcolor', '#bfa89b', sub { table_header_has($_[0], "ETA") }, sub { not table_header_has($_[0], "Starting") }); wlog("table found"); my @fields = qw(gar1 gar2 season station comp airtime eta gar3 gar4); wlog("grabbing values from tree"); foreach my $anime ($table->look_down('_tag', 'tr', #starts looking through the rows 'align', undef, 'bgcolor', '#e9dfdb')){ my $name = $anime->look_down('_tag', 'a', 'name', qr/\d+/)->parent(); #gets the name (for link stays as hash) my $link = $name->parent()->look_down('_tag', 'a', 'href', qr/anidb/); #get the RIGHT link $showtime{$name->as_text}->{'link'} = $link ? anidb_short($link->{'href'}) : undef; #sets up the hash #print LISP "$name -> $link\n"; my $count = 0; # hackish, fix later foreach my $node ($anime->look_down('_tag', 'td')) { $showtime{$name->as_text}->{$fields[$count++]} = $node->as_text; #gets all those other stupid values } } @fields = ("link",@fields[3,5,6]); #dump the "/gar\d/s" ;-) wlog("storing values"); foreach my $key (sort keys %showtime) { print LISP "$key -> @{[ map { $showtime{$key}->{$_} } @fields ]}\n"; } wlog("work done, deleting tree"); $html->delete; close(LISP) or warn "Failed to close: $!\n"; print "All done...."; } my ($subgex,@lastlog); if @ARGV { $ARGV_[0]->() and exit; } #better get it right. ^^; print curti(), " - " and helpme(); while($subgex !~ /exit/){ foreach my $regex (sort keys %gotime) { if($subgex =~ /$regex/) { $gotime{$regex}->(); print "Now what?\n"; } } chomp($subgex = ); } #core subroutines sub table_header_has { my ($table, $str) = @_; wlog("looking for $str"); return defined $table->look_down('_tag', 'th', sub { $_[0]->as_text() eq $str }); } sub anidb_short { #turns the ugly "anidb.info/perl-bin/animedb.pl?show=anime&aid=xxxx" into "anidb.info/axxxx" my $url = shift; $url =~ s/^http.*aid.(\d+)$/$1/i; $url = 'http://anidb.info/a' . $url; return $url; } sub write_database { my $dbh = DBI->connect("dbi:SQLite:dbname=mahou.db","","") or warn "Can't connect: $!\n"; } #stuff for the idler sub wlog { my $write = shift; $write = "@{[ curti() ]} \u$write."; print "$write\n" and push(@lastlog,$write); } # simple logger thing sub curti { my @time = @{[ gmtime ]}[2,1,0]; $time[0] = $time[0]-3 < 0 ? $time[0]+9 : $time[0]-3; return sprintf("%02d:%02d:%02d", @time); } # gets the time... DEAN'S WAY! sub helpme { print "Commands: help, parse, time, log, exit\n"; } sub lastlog { print "No logs.\n" and return unless @lastlog; print foreach map { $_, "\n" } @lastlog; }