SHOW:
|
|
- or go back to the newest paste.
| 1 | #!/usr/bin/perl | |
| 2 | #---------------------------------------- | |
| 3 | # | |
| 4 | # http://alterchan.net/b/res/31574.html | |
| 5 | # | |
| 6 | #---------------------------------------- | |
| 7 | ||
| 8 | use v5.10; | |
| 9 | use strict; | |
| 10 | use warnings; | |
| 11 | use autodie; | |
| 12 | - | use threads; |
| 12 | + | use Coro; |
| 13 | - | use Thread::Semaphore; |
| 13 | + | use Coro::LWP; |
| 14 | use Coro::Semaphore; | |
| 15 | use LWP; | |
| 16 | ||
| 17 | #---------------------------------------- | |
| 18 | # Настройки | |
| 19 | ||
| 20 | my $password = ""; # Пароль от любого аккаунта, его блеклист будет очищен | |
| 21 | ||
| 22 | my $board = "b"; # Доска | |
| 23 | my $pages = 3; # Количество сканируемых страниц начиная с 0 | |
| 24 | my $thread = 0; # Если указано, то сканируется только один тред | |
| 25 | ||
| 26 | my $target = 0; # Целевой пост | |
| 27 | my $outfile = "posts.txt"; # Куда сохранить найденные посты | |
| 28 | ||
| 29 | #---------------------------------------- | |
| 30 | ||
| 31 | my $lwp = new LWP::UserAgent; | |
| 32 | $lwp->agent( "Opera/9.80 (X11; Linux i686; U; en) Presto/2.10.289 Version/12.02" ); | |
| 33 | $lwp->cookie_jar( {} );
| |
| 34 | - | my $sem = new Thread::Semaphore( 15 ); |
| 34 | + | |
| 35 | my $sem = new Coro::Semaphore( 15 ); | |
| 36 | ||
| 37 | sub login($) {
| |
| 38 | my( $pass ) = @_; | |
| 39 | say "Login"; | |
| 40 | my $res = $lwp->post( "http://alterchan.net/uid.php", | |
| 41 | Content_Type => "application/x-www-form-urlencoded", | |
| 42 | Content => "action=login&pass1=$pass", | |
| 43 | ); | |
| 44 | unless( $res->content =~ /Ist Gut/ ) {
| |
| 45 | die "Can't login:\n" . $res->content; | |
| 46 | } | |
| 47 | } | |
| 48 | ||
| 49 | sub add_to_blacklist($$) {
| |
| 50 | my( $board, $post ) = @_; | |
| 51 | my $res = $lwp->post( "http://alterchan.net/uid.php", | |
| 52 | Content_Type => "application/x-www-form-urlencoded", | |
| 53 | Content => "action=blacklist&number=$post&board=$board", | |
| 54 | ); | |
| 55 | unless( $res->is_success ) {
| |
| 56 | die "Can't hide post '$board/$post':\n" . $res->as_string; | |
| 57 | } | |
| 58 | } | |
| 59 | ||
| 60 | sub clear_blacklist() {
| |
| 61 | my $res = $lwp->post( "http://alterchan.net/uid.php", | |
| 62 | Content_Type => "application/x-www-form-urlencoded", | |
| 63 | Content => "action=erase", | |
| 64 | ); | |
| 65 | unless( $res->is_success ) {
| |
| 66 | die "Can't erase blacklist:\n" . $res->as_string; | |
| 67 | } | |
| 68 | } | |
| 69 | ||
| 70 | sub get_page($$) {
| |
| 71 | my( $board, $page ) = @_; | |
| 72 | say "Get page '$board/$page'"; | |
| 73 | my $url = "http://alterchan.net/$board/" . ( $page ? "$page.html" : "" ); | |
| 74 | my $res = $lwp->get( $url ); | |
| 75 | if( $res->is_success ) {
| |
| 76 | return $res->content; | |
| 77 | } else {
| |
| 78 | die "Can't download page '$board/$page':\n" . $res->as_string; | |
| 79 | } | |
| 80 | } | |
| 81 | ||
| 82 | sub parse_threads($) {
| |
| 83 | my( $html ) = @_; | |
| 84 | my @threads = $html =~ /^<div id="thread(\d+).+">/gm; | |
| 85 | if( @threads ) {
| |
| 86 | return @threads; | |
| 87 | } else {
| |
| 88 | die "No threads found"; | |
| 89 | } | |
| 90 | } | |
| 91 | ||
| 92 | sub get_thread($$) {
| |
| 93 | my( $board, $thread ) = @_; | |
| 94 | say "Get thread '$board/$thread'"; | |
| 95 | my $res = $lwp->get( "http://alterchan.net/$board/res/$thread.html" ); | |
| 96 | if( $res->is_success ) {
| |
| 97 | return $res->content; | |
| 98 | } else {
| |
| 99 | die "Can't download thread '$board/$thread':\n" . $res->as_string; | |
| 100 | } | |
| 101 | } | |
| 102 | ||
| 103 | sub parse_posts($) {
| |
| 104 | my( $html ) = @_; | |
| 105 | my %posts = $html =~ m/^<td class="reply" id="reply(\d+)">.*?^<blockquote>(.*?)^<\/blockquote>/gms; | |
| 106 | for( values %posts ) {
| |
| 107 | s/^\s+|\s+$//g; | |
| 108 | s/<.*?>//g; | |
| 109 | s/>/>/g; | |
| 110 | s/"/"/g; | |
| 111 | } | |
| 112 | return \%posts; | |
| 113 | } | |
| 114 | ||
| 115 | sub parse_all_posts(@) {
| |
| 116 | my( @threads ) = @_; | |
| 117 | - | my @workers = map {
|
| 117 | + | |
| 118 | - | new threads( sub {
|
| 118 | + | my @workers; |
| 119 | for my $thread ( @threads ) {
| |
| 120 | - | my @ret = ( $_, parse_posts( get_thread( $board, $_ ) ) ); |
| 120 | + | push @workers, async {
|
| 121 | $sem->down; | |
| 122 | my @ret = ( $thread, parse_posts( get_thread( $board, $thread ) ) ); | |
| 123 | - | } ); |
| 123 | + | |
| 124 | - | } @threads; |
| 124 | + | |
| 125 | }; | |
| 126 | } | |
| 127 | for( @workers ) {
| |
| 128 | my( $thread, $posts ) = $_->join; | |
| 129 | $threads{$thread} = $posts;
| |
| 130 | } | |
| 131 | return \%threads; | |
| 132 | } | |
| 133 | ||
| 134 | sub write_file($$) {
| |
| 135 | my( $fname, $data ) = @_; | |
| 136 | open my $fh, ">", $fname; | |
| 137 | if( syswrite( $fh, $data ) != length $data ) {
| |
| 138 | warn "File '$fname' written with errors"; | |
| 139 | } | |
| 140 | } | |
| 141 | ||
| 142 | #---------------------------------------- | |
| 143 | ||
| 144 | login( $password ); | |
| 145 | clear_blacklist(); | |
| 146 | ||
| 147 | my @all; | |
| 148 | if($thread) {
| |
| 149 | - | my @workers = map {
|
| 149 | + | |
| 150 | - | new threads ( sub {
|
| 150 | + | |
| 151 | my @workers; | |
| 152 | - | my @ret = parse_threads( get_page( $board, $_ ) ); |
| 152 | + | for my $page ( 0 .. $pages - 1 ) {
|
| 153 | push @workers, async {
| |
| 154 | $sem->down; | |
| 155 | - | } ); |
| 155 | + | my @ret = parse_threads( get_page( $board, $page ) ); |
| 156 | - | } ( 0 .. $pages - 1 ); |
| 156 | + | |
| 157 | return @ret; | |
| 158 | }; | |
| 159 | } | |
| 160 | push @all, $_->join for @workers; | |
| 161 | } | |
| 162 | ||
| 163 | my $threads = parse_all_posts( @all ); | |
| 164 | ||
| 165 | add_to_blacklist( $board, $target ); | |
| 166 | my @after = map { keys %$_ } values %{ parse_all_posts( @all ) };
| |
| 167 | ||
| 168 | clear_blacklist(); | |
| 169 | ||
| 170 | my $text; | |
| 171 | for my $thread ( sort { $a <=> $b } keys %$threads ) {
| |
| 172 | for my $post ( sort { $a <=> $b } keys %{ $threads->{$thread} } ) {
| |
| 173 | next if $post ~~ @after; | |
| 174 | $text .= "http://alterchan.net/$board/res/$thread.html\n"; | |
| 175 | $text .= "Тред $thread, пост $post:\n-----\n$threads->{$thread}->{$post}\n-----\n";
| |
| 176 | } | |
| 177 | } | |
| 178 | ||
| 179 | write_file( $outfile, $text ); |