use strict; use warnings; use WWW::Mechanize (); use String::Approx 'adist'; main( @ARGV ); exit; sub main { initialize( @_ ); for my $part_number ( @_ ) { print part_spec( $part_number ) . "\n"; } } sub initialize { } sub part_spec { my $part_number = shift; my ( $bare_number ) = $part_number =~ /^(ST\d+)/; my $www = WWW::Mechanize->new; $www->get( 'http://www.seagate.com/support/disc/' ); $www->submit_form( form_number => 2, fields => { modelcode => $bare_number } ); my $spec_page; for my $test ( qr/.support.disc.specs.+$part_number/i, qr/.support.disc.specs.+$bare_number/i ) { my @candidates = $www->find_all_links( url_regex => $test ); if ( 1 == @candidates ) { $spec_page = $candidates[0]; last; } elsif ( @candidates ) { my @dists = map adist( lc( $part_number ), lc( $_->url ) ), @candidates; $spec_page = $candidates[ ( sort { $dists[$a] <=> $dists[$b] } 0 .. $#dists )[ 0 ] ]; last; } } $www->get( $spec_page->URI ); my ( $capacity, $rpm, $access_time, $desc ) = $www->content =~ /\b ([\d.]+) \s* GB .+? \b (\d+) \s* rpm .+? \b ([\d.]+) \s* ms .+? ((?-s:ST-\d+.+))/xs; return "Desc=$desc\n" . "Capacity=$capacity\n" . "RPM=$rpm\n" . "Access time=$access_time\n"; } sub render_html { require HTML::FormatText; my $html = shift; return HTML::FormatText->format_string( $html ); }