logbot-web

#!/usr/bin/perl
# mojolicious web handler
# must set env var LOGBOT_CONFIG to config filename
#
# in production runs as a daemon behind a reverse proxy, which handles requests
# for static assets (web/public/static)
#
# use `dev-make` to build static assets
use local::lib;
use v5.10;
use strict;
use warnings;

use FindBin qw( $RealBin );
use lib "$RealBin/lib";

BEGIN { $ENV{TZ} = 'UTC' }

use DateTime ();
use IO::Compress::Gzip qw( gzip );
use LogBot::Config qw( find_config load_all_configs load_config reload_config );
use LogBot::MemCache ();
use LogBot::Util qw( file_for file_time time_to_ymd );
use LogBot::Web::Util qw( channel_from_param channel_topics linkify render_init rewrite_old_urls );
use Mojo::ByteStream ();
use Mojo::Log        ();
use Mojo::Util qw( dumper );
use Mojolicious::Lite qw( app );

# load networks
my $networks = [];
{
    if (my $config = $ENV{LOGBOT_CONFIG}) {

        # load specific networks (mostly for dev)
        foreach my $network (split(/,/, $config)) {
            push @{$networks}, load_config(find_config($network), web => 1);
        }

    } else {

        # load all networks
        foreach my $network (values %{ load_all_configs(web => 1) }) {
            push @{$networks}, $network;
        }
    }

    $networks = [sort { $a->{name} cmp $b->{name} } @{$networks}];
}

# configure mojo
my $is_production = app->mode() eq 'production';
app->secrets('!logbot!');
app->renderer->paths([$RealBin . '/web/templates']);
app->static->paths([$RealBin . '/web/public']);
app->config(
    hypnotoad => {
        listen => ['http://127.0.0.1:' . ($ENV{LOGBOT_PORT} // 3001)],
        pid_file => ($ENV{LOGBOT_PID_FILE} // ($RealBin . '/logbot-web.pid')),
    },
);
app->log(Mojo::Log->new(path => $is_production ? '/var/log/logbot/mojo.log' : 'log/mojo.log'));
plugin AccessLog => {
    log => ($is_production ? '/var/log/logbot/access.log' : 'log/access.log'),
    format => '%h %{X-Network}o - %t "%r" %>s %b "%{Referer}i" "%{User-Agent}i"',
};
plugin 'Status' => {
    route => app->routes->under(
        '_status' => sub {
            my ($c) = @_;
            my $password = $ENV{'LOGBOT_STATUS_PASSWORD'} // 'status';
            return 1 if $password eq '';
            return 1 if ($c->req->url->to_abs->userinfo // '') eq 'logbot:' . $password;
            $c->res->headers->www_authenticate('Basic');
            $c->render(text => 'Authentication required', status => 401);
            return undef;
        }
    )
};

my $memcache = LogBot::MemCache->new(binary => $is_production);
render_init($RealBin);

# per-request initialisation
under sub {
    my ($c) = @_;

    # determine current config
    my $config_index = 0;
    if (scalar(@{$networks}) > 1) {
        my $host = lc($c->req->url->to_abs->host);
        my $dot_posn = index($host, '.');
        if ($dot_posn != -1) {
            $host = substr($host, 0, $dot_posn);
            for (my $i = 0; $i < scalar(@{$networks}); $i++) {
                next unless $networks->[$i]->{name} eq $host;
                $config_index = $i;
                last;
            }
        }
    }

    # reload config
    $networks->[$config_index] = reload_config($networks->[$config_index]);
    my $config = $networks->[$config_index];

    # store network name in response header for logging
    $c->res->headers->add('X-Network' => $config->{name});

    # for client-side list cache
    my $topics_lastmod = file_time(file_for($config, 'topics_lastmod')) // 0;

    $c->stash(
        config          => $config,
        networks        => $networks,
        network         => $config->{name},
        channels        => $config->{_derived}->{visible_channels},
        topics          => channel_topics($config),
        channel         => '',
        date            => '',
        error           => '',
        event_count     => 0,
        bot_event_count => 0,
        page            => '',
        today           => DateTime->now()->truncate(to => 'day'),
        is_today        => 0,
        cache_prefix    => $config->{_derived}->{time} . '.' . $config->{name} . '.',
        channel_list_id => $config->{_derived}->{time} . '.' . $topics_lastmod,
        topics_lastmod  => $topics_lastmod,
    );
    return 1;
};

#

# default => about logbot || search
get '/' => sub {
    my ($c) = @_;

    # redirect old urls
    if (my $url = rewrite_old_urls($c)) {

        # url was formed server-side, no need to bounce through js redirect
        if ($url !~ /#/) {
            return $c->redirect_to($url);
        }

        $c->stash(redirect_to => $url);
        return $c->render('redirect');
    }

    # search
    my $q = $c->req->query_params->param('q');
    if (defined($q)) {
        LogBot::Web::Search::render($c, $q);

    } else {

        # index
        LogBot::Web::Index::render($c);
    }
};

# config
get '/_config' => sub {
    my ($c) = @_;
    LogBot::Web::Config::render($c);
};

# channel list
get '/_channels' => sub {
    my ($c) = @_;
    LogBot::Web::List::render($c);
};

get '/_channels_body' => sub {
    my ($c) = @_;
    LogBot::Web::List::render($c, { body_only => 1 });
};

# network stats
get '/_stats' => sub {
    my ($c) = @_;
    LogBot::Web::Stats::render($c);
};

get '/_stats/meta' => sub {
    my ($c) = @_;
    LogBot::Web::Stats::render_meta($c);
};

get '/_stats/hours' => sub {
    my ($c) = @_;
    LogBot::Web::Stats::render_hours($c);
};

# debugging
if (!$is_production) {
    get '/_stash' => sub {
        my ($c) = @_;
        $c->stash(today => $c->stash('today')->ymd());
        $c->render(text => dumper($c->stash), format => 'txt');
    };
}

# robots.txt
my $robots_txt = <<'EOF';
# http://law.di.unimi.it/BUbiNG.html
# 20% of my traffic was from this bot
User-agent: BUbiNG
Disallow: /

# Marketing/SEO bot
User-agent: SemrushBot
Disallow: /
User-agent: SemrushBot-SA
Disallow: /
EOF
get '/robots.txt' => sub {
    my ($c) = @_;
    $c->render(text => $robots_txt, format => 'txt');
};

# /channel => redirect to current date
get '/#channel' => sub {
    my ($c) = @_;
    my $channel = channel_from_param($c) // return;

    # redirect to current date
    my $path = $c->req->url->path;
    $path .= '/' unless substr($path, -1) eq '/';
    $c->redirect_to($path . time_to_ymd(time()));
};

# /channel/date => show logs
get '/#channel/:date' => [date => qr/\d{8}/] => sub {
    my ($c) = @_;
    LogBot::Web::Channel::render_logs($c);
};

get '/#channel/:date/raw' => [date => qr/\d{8}/] => sub {
    my ($c) = @_;
    LogBot::Web::Channel::render_raw($c);
};

get '/#channel/link/:time/:nick' => [time => qr /\d+/] => sub {
    my ($c) = @_;
    LogBot::Web::Channel::redirect_to($c);
};

get '/#channel/stats' => sub {
    my ($c) = @_;
    LogBot::Web::Stats::render($c, require_channel => 1);
};

get '/#channel/stats/meta' => sub {
    my ($c) = @_;
    LogBot::Web::Stats::render_meta($c, require_channel => 1);
};

get '/#channel/stats/hours' => sub {
    my ($c) = @_;
    LogBot::Web::Stats::render_hours($c);
};

get '/#channel/stats/nicks' => sub {
    my ($c) = @_;
    LogBot::Web::Stats::render_nicks($c);
};

# 404 handler
any '*' => sub {
    my ($c) = @_;
    $c->res->code(404);
    $c->res->message('Not Found');
    LogBot::Web::Index::render($c, { error => 'Not Found.' });
};

my %cache;

# static file with timestamp
helper static => sub {
    my ($c, $file) = @_;
    return $cache{static}->{$file} //= '/static/' . $file . '?' . file_time($RealBin . '/web/public/static/' . $file);
};

# inline svg
helper svg => sub {
    my ($c, $file) = @_;
    return $cache{svg}->{$file} //= Mojo::ByteStream->new(slurp($RealBin . '/web/svg/' . $file . '.svg'));
};

# linkify text
helper linkify => sub {
    my ($c, $text) = @_;
    return linkify($text);
};

# cache
helper cached => sub {
    my ($c, $key, $callback) = @_;
    return $key eq ''
        ? $callback->()
        : Mojo::ByteStream->new($memcache->cached($c->stash('cache_prefix') . $key, $callback));
};

hook after_render => sub {
    my ($c, $output, $format) = @_;
    my $headers = $c->res->headers;

    # CSP
    state $csp = join(
        '; ',
        q{default-src 'self'},
        q{object-src 'none'},
        q{frame-ancestors 'none'},
        q{base-uri 'none'},
        q{style-src 'self' 'unsafe-inline'},  # unsafe-inline for chosen, top-nick graph
        q{img-src 'self' data:},              # data: for pikaday
    );
    $headers->header('Content-Security-Policy' => $csp);

    # preload fonts
    state $link = join(', ',
        map { '<' . $c->url_for($_)->to_abs . '>; rel=preload; as=font' }
            qw( /static/hind-regular.ttf /static/hind-medium.ttf /static/hind-bold.ttf ));
    $headers->header(Link => $link);

    # no need to expose this info
    $headers->remove('Server');

    # gzip compression
    if (($c->req->headers->accept_encoding // '') =~ /gzip/i) {
        $headers->append(Vary => 'Accept-Encoding');
        $headers->content_encoding('gzip');
        gzip($output, \my $compressed);
        ${$output} = $compressed;
    }
};

app->start;