All pastes #579795 Raw Edit

v0.0.5

public text v1 · immutable
·published 2007-06-20 20:48 UTC
rendered paste body
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;



 ###                     ###
 #####D#O#N#E###############
  #                       #
   # main hash and array #
  #                       #
 ###########################
 ###                     ###


my %things = (
  'help'   => sub { print curti(), " - Commands: parse, find, time, exit, report\n" },
  'find'   => \&sql_finder,
  'parse'  => \&sql_writer,              # I like the next one! \o/
  'report' => sub { hash_report() and return if shift =~ /hash/; meta_report(); },
  'time'   => sub { print curti(), " There you go.\n" }
);
my @fields = (
  'number',
  'series',    # why is @fields out here?
  'season',    # well billy, I just don't
  'station',   # know~! :) :P ;) xD <3 :3
  'company',
  'airtime',
  'eta',       # it's only used once, may
  'eps',       # as well ditch it soon :)
  'links'
);



 ###                      ###
 #####D#O#N#E################
  #                        #
   # heart of the progrum #
  #                        #
 ############################
 ###                      ###


my $subjex = intro();
while($subjex !~ /exit/io){
  foreach my $regex (sort keys %things) {
    if($subjex =~ /^\s*($regex)(?:\s*(.*))?/i) {
      $things{$1}->($2);
    }
  }
  print curti(), " > ";
  chomp($subjex = <STDIN>);
}



 ###                    ###
 ##########################
  #                      #
   # spiffy subroutines #
  #                      #
 ##########################
 ###                    ###


   #D#O#N#E# # # # # # # # #
   # stuff for anidb links #
   # # # # # # # # # # # # #

sub anidb_link {  # makes pretty URLs
  my $url = shift;
  return '-' if !$url;
  $url = 'http://anidb.info/a' . $url;
  return $url;
}

sub anidb_id {  # return AIDs
  my $url = shift;
  $url =~ s/^http.*aid.(\d+)$/$1/i;
  return $url;
}


   #D#O#N#E# # # # # # #
   # sql stuff... ugh! #
   # # # # # # # # # # #

sub build_hash {
  my %showtime = {};
  my $dbh = DBI->connect(
    'dbi:SQLite:dbname=mahou.db',undef,undef,
    { RaiseError => 1, AutoCommit => 0 }
  );

  eval {
    # build hash from mahou_data...
    # someone should code this. ;-)
  };
  if($@) {
    print curti(), " Build hash: Database error.\n";
  }
  $dbh->disconnect;
  return 0;
}

sub check_meta {
  my @delta = parti(pop);
  print curti(), " Check meta: Checking meta...\n";

  # first: hour is same, minutes > 30
  # second: hour is + 1, minutes < 30
  if ($delta[0] == 0 and $delta[1] > 30) {
    print curti(), " Check meta: Hit!\n";
    return @{[ 1,++$_[0],$_[1],curti() ]};
  } elsif ($delta[0] == 1 and $delta[1] < 30) {
    print curti(), " Check meta: Hit!\n";
    return @{[ 1,++$_[0],$_[1],curti() ]};
  }
  print curti(), " Check meta: Miss!\n";
  return @{[ 0,$_[0],++$_[1],curti() ]};
}

sub get_meta {

  # hash of params for the check
  my %type = (
    'query' => 
      { 'return' => 0, 'update' => 1 },
    'report' =>
      { 'return' => 1, 'update' => 0 }
  ); # this wasn't completely needed;
     # it's here for scalability? ^^;


  my $dbh = DBI->connect(
    'dbi:SQLite:dbname=mahou.db',undef,undef,
    { RaiseError => 1, AutoCommit => 0 }
  );
  
  # check who sent for us
  my ($from,$return) = (shift,[undef,[]]);
  print curti(), " Get meta: Coming from $from, using method:";
  foreach my $lol (keys %{$type{$from}}) {
    print " $lol" if $type{$from}->{$lol};
  } print "\n";

  eval {
    # get meta from db
    @{$return->[1]} = @{[ $dbh->selectrow_array('SELECT * from meta_data') ]};
    print curti(), " Get meta: Pre update: @{$return->[1]}\n";

    if ($type{$from}->{'update'}) {
      # check and get new meta
      ($return->[0],@{$return->[1]}) = check_meta(@{$return->[1]});

      # delete old meta and get insert ready
      $dbh->do('DELETE FROM meta_data');
      print curti(), " Get meta: Deleted old data.\n";
      my $sth = $dbh->prepare('INSERT INTO meta_data values(?,?,?)');

      # insert and commit
      $sth->execute(@{$return->[1]});
      $dbh->commit;
    }
  };

  if($@) {
    print curti(), " Get meta: Database error.\n";
    $dbh->rollback;
  }

  print curti(), " Get meta: Meta check complete, new meta: @{$return->[1]}\n";
  $dbh->disconnect;

  # only send back what we need
  return \$return if $type{$from}->{'return'} and $type{$from}->{'update'};
  return \@{$return->[1]} if $type{$from}->{'return'};
  return $return->[0];
}


   #D#O#N#E# # # # # # # # # # # # # #
   # core sql search/writer thingies #
   # # # # # # # # # # # # # # # # # #

sub sql_finder {
  # set need stuff, check metadata
  my $search = shift; my @data;
  sql_writer() if get_meta('query');

  # connect to db, finally
  my $dbh = DBI->connect(
    'dbi:SQLite:dbname=mahou.db',undef,undef,
    { RaiseError => 1, AutoCommit => 0 }
  );

  eval {
    print curti(), " Sql finder: Starting search...\n";

    # search, first checks for aid, then defaults to name
    if($search =~ /^(?:ID:)?(\d+)$/) {
      @data = $dbh->selectrow_array("SELECT * from mahou_data WHERE aid = $1");
    } else {
      $search = '%' . $search . '%';
      my $sth = $dbh->prepare('SELECT * from mahou_data WHERE name like ?');
      $sth->execute($search);
      @data = $sth->fetchrow_array;
    }

    if(@data) {
      print curti(), " Sql finder: Adjusting time from: $data[4]\n";
      $data[4] = fixti($data[4]);
      print curti(), " Query Data:\n";
      print "          $_\n" foreach @data;
    } else {
      print curti(), " Query Data: No results.\n";
    }

  };
  if($@) {
    print curti(), " Sql finder: Database error.\n";
  }
  $dbh->disconnect;
}

sub sql_writer {
  print curti(), " Sql writer: Getting hash...\n";

  # grab hash from parser and connect to db
  my $hash = core_parse();
  print curti(), " Sql writer: Connecting and preparing insert.\n";

  my $dbh = DBI->connect(
    'dbi:SQLite:dbname=mahou.db',undef,undef,
    { RaiseError => 1, AutoCommit => 0 }
  );

  eval {
    # clear table, prepare insert
    $dbh->do('DELETE FROM mahou_data');
    my $sth = $dbh->prepare('INSERT INTO mahou_data values(?,?,?,?,?)');

    # insert
    print curti(), " Sql writer: Inserting rows.\n";
    foreach my $anime (sort keys %{$hash}) {
      $sth->execute($anime,@{[ map { $hash->{$anime}{$_} } ('link','station','airtime','eta') ]});
    }
    $dbh->commit;
  };
  if($@) {
    $dbh->rollback;
    print curti(), " Sql writer: Database error.\n";
  }

  $dbh->disconnect;
  print curti(), " Sql writer: Database updated.\n";
}


   #D#O#N#E# # # # # # # # # # # #
   # download, parse and related #
   # # # # # # # # # # # # # # # #

sub table_header_has {
  my ($table, $str) = @_;
  return defined $table->look_down(
    '_tag', 'th',
    sub { $_[0]->as_text() eq $str }
  );
} # by pc486

sub core_parse {
  # build tree, get shit from it ^-^
  print curti(), " Core parse: Starting parser...\n";
  my ($html,%showtime) = (get("http://mahou.org/Showtime"),{});
  $html = HTML::TreeBuilder->new_from_content($html);

  # first step: find time table
  print curti(), " Core parse: Searching for table...\n";
  my $table = $html->look_down(
    '_tag', 'table',
    'bgcolor', '#bfa89b',
    sub { table_header_has($_[0], "ETA") },
    sub { not table_header_has($_[0], "Starting") }
  ); # by pc486

  # second step: find ACTUAL time table
  print curti(), " Core parse: Constructing hash...\n";
  foreach my $anime ($table->look_down(
    '_tag', 'tr',
    'align', undef, # skip first <tr>
    'bgcolor', '#e9dfdb')
  ){

    # first: get the name, for LINK: stays as hash; pc486!
    # second: get the ^<RIGHT>^ link (which is kinda good)
    my $name = $anime->look_down('_tag', 'a', 'name', qr/\d+/)->parent();
    my $link = $name->parent()->look_down('_tag', 'a', 'href', qr/anidb/);

    # starts the hash (with an aid), then populates it with map
    $showtime{$name->as_text}->{'link'} = $link ? anidb_id($link->{'href'}) : undef;
    map { $showtime{$name->as_text}->{$fields[$_]} =
      @{[ $anime->look_down('_tag', 'td') ]}[$_]->as_text } (3,5,6);

  }

  $html->delete;
  print curti(), " Core parse: Returning hash to parent...\n";
  return \%showtime;
}


   #D#O#N#E# # # # # # # #
   # stuff for the idler #
   # # # # # # # # # # # #

sub meta_report {
  print curti(), " Meta report: Initializing...\n";
  my $meta = get_meta('report');
  print curti(), " Mahou Meta Report:\n",
    "              re-parse hits   -> @{[ $meta->[0] ]}\n",
    "              re-parse misses -> @{[ $meta->[1] ]}\n",
    "              time since last -> @{[ cheti(parti($meta->[2])) ]}\n";

}

sub intro {
  print ":::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::\n",
        ":::::::::                                           :::::::::\n",
        "::::::::: Welcome to the Mahou.org/Showtime parser! :::::::::\n",
        ":::::::::                                           :::::::::\n",
        ":::::::::    `Being dumb since 1823!`               :::::::::\n",
        ":::::::::          `Or something.`        :::::::::::::::::::\n",
        ":::::::::                                 ::By: Dean_Kreger::\n",
        ":::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::\n";
  return('help');
}

 # I may do this if I get bored... 
sub hash_report {
  my %showtime = {};
  my $lol = build_hash(); # returns zero
  if ($lol) {
    foreach my $anime (sort keys %showtime) {
      print "$anime <-<+>-> ", anidb_link($showtime{$anime}->{'link'}),
            "\n  Station -> ", $showtime{$anime}->{'station'},
            "\n  Airtime -> ", $showtime{$anime}->{'airtime'},
            "\n  ETA     -> ", $showtime{$anime}->{'eta'}, "\n\n";
    }
  }
  print curti(), " The hash report function has yet to be implemented.\n";
}


   # # # # # # # # # # # # # # # # # # #
   # kawaii~ time related subroutines~ #
   # # # # # # # # # # # # # # # # # # #

      ### ##  ##  ##  ##  ##  ##  ##  ## ###
      #  These were my play things. :) <3  #
      ### ##  ##  ##  ##  ##  ##  ##  ## ###

sub fixti {
  my %fixti; my @curti = split(/:/,curti());
  map { $fixti{chop($_)} = $_ } @{[ split(/\s+/,shift) ]};
  return "working on it...";
} # fixes the time... DEAN'S WAY!


sub cheti {
  my $conti = shift;
  return "update already! it's been like a day or more" if $conti < 0;
  return "more than five hours... I think..." if $conti > 5;
  return "like... more than an hour, at least" if $conti > 1;
  return "pretty much less than an hour" if $conti <= 1;
  return "erroneous check; please replace coder";
} # checks the time... DEAN'S WAY! ... Sorta...


sub parti {
  my @parti; push(@parti,shift);
  my @curti; push(@curti,curti());
  print curti(), " Parti: Doing: @curti - @parti\n";
  map { @$_ = split(/:/,$_->[0]) } (\@parti,\@curti);
  return @{[ map { $curti[$_] - $parti[$_] } (0,1,2) ]};
} # parses the time... DEAN'S WAY!


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!