--- trunk/bin/irc-logger.pl 2007/12/16 19:03:35 72
+++ trunk/bin/irc-logger.pl 2008/03/07 10:13:45 91
@@ -20,8 +20,6 @@
=item --log=irc-logger.log
-Name of log file
-
=back
=head1 DESCRIPTION
@@ -32,7 +30,8 @@
## CONFIG
-my $HOSTNAME = `hostname`;
+my $HOSTNAME = `hostname -f`;
+chomp($HOSTNAME);
my $NICK = 'irc-logger';
$NICK .= '-dev' if ($HOSTNAME =~ m/llin/);
@@ -45,31 +44,28 @@
$CHANNEL = '#irc-logger' if ($HOSTNAME =~ m/llin/);
my $IRC_ALIAS = "log";
-my %FOLLOWS =
- (
- ACCESS => "/var/log/apache/access.log",
- ERROR => "/var/log/apache/error.log",
- );
-
my $DSN = 'DBI:Pg:dbname=' . $NICK;
-my $ENCODING = 'ISO-8859-2';
my $TIMESTAMP = '%Y-%m-%d %H:%M:%S';
my $sleep_on_error = 5;
-my $http_port = $NICK =~ m/-dev/ ? 8001 : 8000;
-my $http_hostname = `hostname`;
-chomp( $http_hostname );
+# number of last tags to keep in circular buffer
+my $last_x_tags = 50;
-## END CONFIG
+# don't pull rss feeds more often than this
+my $rss_min_delay = 60;
+$rss_min_delay = 15;
+my $http_port = $NICK =~ m/-dev/ ? 8001 : 8000;
+
+my $url = "http://$HOSTNAME:$http_port";
+## END CONFIG
-use POE qw(Component::IRC Wheel::FollowTail Component::Server::HTTP);
+use POE qw(Component::IRC Component::Server::HTTP);
use HTTP::Status;
use DBI;
-use Encode qw/from_to is_utf8/;
use Regexp::Common qw /URI/;
use CGI::Simple;
use HTML::TagCloud;
@@ -95,14 +91,14 @@
'log:s' => \$log_path,
);
-$SIG{__DIE__} = sub {
- confess "fatal error";
-};
+#$SIG{__DIE__} = sub {
+# confess "fatal error";
+#};
open(STDOUT, '>', $log_path) || warn "can't redirect log to $log_path: $!";
sub _log {
- print strftime($TIMESTAMP,localtime()), ' ', join(" ",@_), $/;
+ print strftime($TIMESTAMP,localtime()) . ' ' . join(" ",@_) . $/;
}
# HTML formatters
@@ -128,7 +124,7 @@
$m =~ s/($escape_re)/$escape{$1}/gs;
$m =~ s#($RE{URI}{HTTP})#e(qq{$1})#egs ||
$m =~ s#\/(\w+)\/#$1#gs;
- $m =~ s#$tag_regex#e(qq{$1})#egs;
+ $m =~ s#$tag_regex#e(qq{$1})#egs;
$m =~ s#\*(\w+)\*#$1#gs;
$m =~ s#_(\w+)_#$1#gs;
@@ -148,9 +144,10 @@
};
my $dbh = DBI->connect($DSN,"","", { RaiseError => 1, AutoCommit => 1 }) || die $DBI::errstr;
+$dbh->do( qq{ set client_encoding = 'UTF-8' } );
my $sql_schema = {
- log => '
+ log => qq{
create table log (
id serial,
time timestamp default now(),
@@ -164,17 +161,31 @@
create index log_time on log(time);
create index log_channel on log(channel);
create index log_nick on log(nick);
- ',
- meta => '
+ },
+ meta => q{
create table meta (
nick text not null,
channel text not null,
name text not null,
value text,
- changed timestamp default now(),
+ changed timestamp default 'now()',
primary key(nick,channel,name)
);
- ',
+ },
+ feeds => qq{
+create table feeds (
+ id serial,
+ url text not null,
+ name text,
+ delay interval not null default '5 min',
+ active boolean default true,
+ last_update timestamp default 'now()',
+ polls int default 0,
+ updates int default 0
+);
+create unique index feeds_url on feeds(url);
+insert into feeds (url,name) values ('http://wiki.razmjenavjestina.org/feed/workspace/razmjenavjestina?category=Recent%20Changes','wiki');
+ },
};
foreach my $table ( keys %$sql_schema ) {
@@ -238,7 +249,7 @@
-my $sth = $dbh->prepare(qq{
+my $sth_insert_log = $dbh->prepare(qq{
insert into log
(channel, me, nick, message, time)
values (?,?,?,?,?)
@@ -482,7 +493,6 @@
=cut
-my $last_x_tags = 5;
my @last_tags;
sub add_tag {
@@ -491,7 +501,6 @@
return unless ($arg->{id} && $arg->{message});
my $m = $arg->{message};
- from_to('UTF-8', 'iso-8859-2', $m) if (is_utf8($m));
my @tags;
@@ -500,14 +509,14 @@
next if (! $tag || $tag =~ m/https?:/i);
push @{ $tags->{$tag} }, $arg->{id};
#warn "+tag $tag: $arg->{id}\n";
- $cloud->add($tag, "?tag=$tag", scalar @{$tags->{$tag}} + 1);
+ $cloud->add($tag, "$url?tag=$tag", scalar @{$tags->{$tag}} + 1);
push @tags, $tag;
}
if ( @tags ) {
- shift @last_tags if $#last_tags == $last_x_tags;
- push @last_tags, { tags => [ @tags ], %$arg };
+ pop @last_tags if $#last_tags == $last_x_tags;
+ unshift @last_tags, { tags => [ @tags ], %$arg };
}
}
@@ -519,14 +528,14 @@
=cut
sub seed_tags {
- my $sth = $dbh->prepare(qq{ select id,message,nick,me,time from log where message like '%//%' });
+ my $sth = $dbh->prepare(qq{ select id,message,nick,me,time from log where message like '%//%' order by time asc });
$sth->execute;
while (my $row = $sth->fetchrow_hashref) {
add_tag( %$row );
}
foreach my $tag (keys %$tags) {
- $cloud->add($tag, "?tag=$tag", scalar @{$tags->{$tag}} + 1);
+ $cloud->add($tag, "$url?tag=$tag", scalar @{$tags->{$tag}} + 1);
}
}
@@ -560,9 +569,7 @@
$a->{me} ? "***" . $a->{nick} : "<" . $a->{nick} . ">",
" " . $a->{message};
- from_to($a->{message}, 'UTF-8', $ENCODING);
-
- $sth->execute($a->{channel}, $a->{me}, $a->{nick}, $a->{message}, $a->{time});
+ $sth_insert_log->execute($a->{channel}, $a->{me}, $a->{nick}, $a->{message}, $a->{time});
add_tag( id => $dbh->last_insert_id(undef,undef,"log",undef), %$a );
}
@@ -601,27 +608,113 @@
exit;
}
+#
+# RSS follow
+#
+
+my $_rss;
+
+
+sub rss_fetch {
+ my ($args) = @_;
+
+ # how many messages to send out when feed is seen for the first time?
+ my $send_rss_msgs = 1;
+
+ _log "RSS fetch", $args->{url};
+
+ my $feed = XML::Feed->parse(URI->new( $args->{url} ));
+ if ( ! $feed ) {
+ _log("can't fetch RSS ", $args->{url});
+ return;
+ }
+ my ( $total, $updates ) = ( 0, 0 );
+ for my $entry ($feed->entries) {
+ $total++;
+
+ # seen allready?
+ return if $_rss->{$feed->link}->{seen}->{$entry->id}++ > 0;
+
+ sub prefix {
+ my ($txt,$var) = @_;
+ $var =~ s/^\s+//g;
+ return $txt . $var if $var;
+ }
+
+ my $msg;
+ $msg .= prefix( 'From: ' , $args->{name} || $feed->title );
+ $msg .= prefix( ' by ' , $entry->author );
+ $msg .= prefix( ' -- ' , $entry->link );
+# $msg .= prefix( ' id ' , $entry->id );
+
+ if ( $args->{kernel} && $send_rss_msgs ) {
+ $send_rss_msgs--;
+ _log('RSS', $msg);
+ $sth_insert_log->execute( $CHANNEL, 1, $NICK, $msg, undef );
+ $args->{kernel}->post( $IRC_ALIAS => notice => $CHANNEL, $msg );
+ $updates++;
+ }
+ }
+
+ my $sql = qq{ update feeds set last_update = now(), polls = polls + 1 };
+ $sql .= qq{, updates = updates + $updates } if $updates;
+ $sql .= qq{where id = } . $args->{id};
+ eval { $dbh->do( $sql ) };
+
+ _log "RSS got $total items of which $updates new";
+
+ return $updates;
+}
+
+sub rss_fetch_all {
+ my $kernel = shift;
+ my $sql = qq{
+ select id, url, name
+ from feeds
+ where active is true
+ };
+ # limit to newer feeds only if we are not sending messages out
+ $sql .= qq{ and last_update + delay < now() } if $kernel;
+ my $sth = $dbh->prepare( $sql );
+ $sth->execute();
+ warn "# ",$sth->rows," active RSS feeds\n";
+ my $count = 0;
+ while (my $row = $sth->fetchrow_hashref) {
+ $row->{kernel} = $kernel if $kernel;
+ $count += rss_fetch( $row );
+ }
+ return "OK, fetched $count posts from " . $sth->rows . " feeds";
+}
+
+
+sub rss_check_updates {
+ my $kernel = shift;
+ my $last_t = $_rss->{last_poll} || time();
+ my $t = time();
+ if ( $t - $last_t > $rss_min_delay ) {
+ $_rss->{last_poll} = $t;
+ _log rss_fetch_all( $kernel );
+ }
+}
+
+# seed rss seen cache so we won't send out all items on startup
+_log rss_fetch_all;
#
# POE handing part
#
-my $SKIPPING = 0; # if skipping, how many we've done
-my $SEND_QUEUE; # cache
my $ping; # ping stats
POE::Component::IRC->new($IRC_ALIAS);
-POE::Session->create( inline_states =>
- {_start => sub {
+POE::Session->create( inline_states => {
+ _start => sub {
$_[KERNEL]->post($IRC_ALIAS => register => 'all');
$_[KERNEL]->post($IRC_ALIAS => connect => $CONNECT);
},
irc_255 => sub { # server is done blabbing
$_[KERNEL]->post($IRC_ALIAS => join => $CHANNEL);
- $_[KERNEL]->post($IRC_ALIAS => join => '#logger');
- $_[KERNEL]->yield("heartbeat"); # start heartbeat
-# $_[KERNEL]->yield("my_add", $_) for keys %FOLLOWS;
$_[KERNEL]->post( $IRC_ALIAS => privmsg => 'nickserv', "IDENTIFY $NICK" );
},
irc_public => sub {
@@ -652,15 +745,16 @@
},
irc_ping => sub {
- warn "pong ", $_[ARG0], $/;
+ _log( "pong ", $_[ARG0] );
$ping->{ $_[ARG0] }++;
+ rss_check_updates( $_[KERNEL] );
},
irc_invite => sub {
my $kernel = $_[KERNEL];
my $nick = (split /!/, $_[ARG0])[0];
my $channel = $_[ARG1];
- warn "invited to $channel by $nick";
+ _log "invited to $channel by $nick";
$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, "how nice of you to invite me to $channel, I'll be right there..." );
$_[KERNEL]->post($IRC_ALIAS => join => $channel);
@@ -671,7 +765,6 @@
my $nick = (split /!/, $_[ARG0])[0];
my $msg = $_[ARG2];
my $channel = $_[ARG1]->[0];
- from_to($msg, 'UTF-8', $ENCODING);
my $res = "unknown command '$msg', try /msg $NICK help!";
my @out;
@@ -715,7 +808,6 @@
foreach my $res (get_from_log( limit => $limit )) {
_log "last: $res";
- from_to($res, $ENCODING, 'UTF-8');
$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res );
}
@@ -730,7 +822,6 @@
search => $what,
)) {
_log "search [$what]: $res";
- from_to($res, $ENCODING, 'UTF-8');
$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res );
}
@@ -800,14 +891,44 @@
$res = "config option $op doesn't exist";
}
}
+ } elsif ($msg =~ m/^rss-update/) {
+ $res = rss_fetch_all( $_[KERNEL] );
+ } elsif ($msg =~ m/^rss-clean/) {
+ $_rss = undef;
+ $dbh->do( qq{ update feeds set last_update = now() - delay } );
+ $res = "OK, cleaned RSS cache";
+ } elsif ($msg =~ m/^rss-list/) {
+ my $sth = $dbh->prepare(qq{ select url,name,last_update,active from feeds });
+ $sth->execute;
+ while (my @row = $sth->fetchrow_array) {
+ $_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, join(' | ',@row) );
+ }
+ $res = '';
+ } elsif ($msg =~ m!^rss-(add|remove|stop|start)\s+(http://\S+)\s*(.*)!) {
+ my $sql = {
+ add => qq{ insert into feeds (url,name) values (?,?) },
+# remove => qq{ delete from feeds where url = ? and name = ? },
+ start => qq{ update feeds set active = true where url = ? },
+ stop => qq{ update feeds set active = false where url = ? },
+
+ };
+ if (my $q = $sql->{$1} ) {
+ my $sth = $dbh->prepare( $q );
+ my @data = ( $2 );
+ push @data, $3 if ( $q =~ s/\?//g == 2 );
+ warn "## $1 SQL $q with ",dump( @data ),"\n";
+ eval { $sth->execute( @data ) };
+ }
+
+ $res = "OK, RSS $1 : $2 - $3";
}
if ($res) {
_log ">> [$nick] $res";
- from_to($res, $ENCODING, 'UTF-8');
$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res );
}
+ rss_check_updates( $_[KERNEL] );
},
irc_477 => sub {
_log "# irc_477: ",$_[ARG1];
@@ -846,66 +967,6 @@
"";
0; # false for signals
},
- my_add => sub {
- my $trailing = $_[ARG0];
- my $session = $_[SESSION];
- POE::Session->create
- (inline_states =>
- {_start => sub {
- $_[HEAP]->{wheel} =
- POE::Wheel::FollowTail->new
- (
- Filename => $FOLLOWS{$trailing},
- InputEvent => 'got_line',
- );
- },
- got_line => sub {
- $_[KERNEL]->post($session => my_tailed =>
- time, $trailing, $_[ARG0]);
- },
- },
- );
-
- },
- my_tailed => sub {
- my ($time, $file, $line) = @_[ARG0..ARG2];
- ## $time will be undef on a probe, or a time value if a real line
-
- ## PoCo::IRC has throttling built in, but no external visibility
- ## so this is reaching "under the hood"
- $SEND_QUEUE ||=
- $_[KERNEL]->alias_resolve($IRC_ALIAS)->get_heap->{send_queue};
-
- ## handle "no need to keep skipping" transition
- if ($SKIPPING and @$SEND_QUEUE < 1) {
- $_[KERNEL]->post($IRC_ALIAS => privmsg => $CHANNEL =>
- "[discarded $SKIPPING messages]");
- $SKIPPING = 0;
- }
-
- ## handle potential message display
- if ($time) {
- if ($SKIPPING or @$SEND_QUEUE > 3) { # 3 msgs per 10 seconds
- $SKIPPING++;
- } else {
- my @time = localtime $time;
- $_[KERNEL]->post($IRC_ALIAS => privmsg => $CHANNEL =>
- sprintf "%02d:%02d:%02d: %s: %s",
- ($time[2] + 11) % 12 + 1, $time[1], $time[0],
- $file, $line);
- }
- }
-
- ## handle re-probe/flush if skipping
- if ($SKIPPING) {
- $_[KERNEL]->delay($_[STATE] => 0.5); # $time will be undef
- }
-
- },
- my_heartbeat => sub {
- $_[KERNEL]->yield(my_tailed => time, "heartbeat", "beep");
- $_[KERNEL]->delay($_[STATE] => 10);
- }
},
);
@@ -913,6 +974,11 @@
my $httpd = POE::Component::Server::HTTP->new(
Port => $http_port,
+ PreHandler => {
+ '/' => sub {
+ $_[0]->header(Connection => 'close')
+ }
+ },
ContentHandler => { '/' => \&root_handler },
Headers => { Server => 'irc-logger' },
);
@@ -958,6 +1024,11 @@
my ($request, $response) = @_;
$response->code(RC_OK);
+ # this doesn't seem to work, so moved to PreHandler
+ #$response->header(Connection => 'close');
+
+ return RC_OK if $request->uri =~ m/favicon.ico$/;
+
my $q;
if ( $request->method eq 'POST' ) {
@@ -970,37 +1041,92 @@
my $search = $q->param('search') || $q->param('grep') || '';
- if ($request->url =~ m#/rss#i) {
+ if ($request->url =~ m#/rss(?:/(tags|last-tag|follow.*)\w*(?:=(\d+))?)?#i) {
+ my $show = lc($1);
+ my $nr = $2;
+
my $type = 'RSS'; # Atom
$response->content_type( 'application/' . lc($type) . '+xml' );
my $html = '';
- warn "create $type feed from ",dump( @last_tags );
+ #warn "create $type feed from ",dump( @last_tags );
my $feed = XML::Feed->new( $type );
+ $feed->link( $url );
- $feed->title( "last $last_x_tags from $CHANNEL" );
-# $feed->link( "http://$http_hostname:$http_port" );
- $feed->description( "collects messages which have tags// in them" );
-
- foreach my $m ( @last_tags ) {
- warn dump( $m );
- #my $tags = join(' ', @{$m->{tags}} );
+ if ( $show eq 'tags' ) {
+ $nr ||= 50;
+ $feed->title( "tags from $CHANNEL" );
+ $feed->link( "$url/tags" );
+ $feed->description( "tag cloud created from messages on channel $CHANNEL which have tags// in them" );
my $feed_entry = XML::Feed::Entry->new($type);
- $feed_entry->title( $m->{nick} . '@' . $m->{time} );
- $feed_entry->author( $m->{nick} );
-# $feed_entry->link( );
- $feed_entry->issued( DateTime::Format::Flexible->build( $m->{time} ) );
- $feed_entry->summary(
- '{nick}->( $m->{nick} ) .
-# '' . $m->{nick} . ' ' .
- $filter->{message}->( $m->{message} ) .
- ']]>'
+ $feed_entry->title( "$nr tags from $CHANNEL" );
+ $feed_entry->author( $NICK );
+ $feed_entry->link( '/#tags' );
+
+ $feed_entry->content(
+ qq{}
+ . $cloud->css
+ . qq{}
+ . $cloud->html( $nr )
+ . qq{]]>}
);
- $feed_entry->category( join(', ', @{$m->{tags}}) );
$feed->add_entry( $feed_entry );
+
+ } elsif ( $show eq 'last-tag' ) {
+
+ $nr ||= $last_x_tags;
+ $nr = $last_x_tags if $nr > $last_x_tags;
+
+ $feed->title( "last $nr tagged messages from $CHANNEL" );
+ $feed->description( "collects messages which have tags// in them" );
+
+ foreach my $m ( @last_tags ) {
+# warn dump( $m );
+ #my $tags = join(' ', @{$m->{tags}} );
+ my $feed_entry = XML::Feed::Entry->new($type);
+ $feed_entry->title( $m->{nick} . '@' . $m->{time} );
+ $feed_entry->author( $m->{nick} );
+ $feed_entry->link( '/#' . $m->{id} );
+ $feed_entry->issued( DateTime::Format::Flexible->build( $m->{time} ) );
+
+ my $message = $filter->{message}->( $m->{message} );
+ $message .= "
\n" unless $message =~ m!<(/p|br/?)>!;
+# warn "## message = $message\n";
+
+ #$feed_entry->summary(
+ $feed_entry->content(
+ ""
+ );
+ $feed_entry->category( join(', ', @{$m->{tags}}) );
+ $feed->add_entry( $feed_entry );
+
+ $nr--;
+ last if $nr <= 0;
+
+ }
+
+ } elsif ( $show =~ m/^follow/ ) {
+
+ $feed->title( "Feeds which this bot follows" );
+
+ my $sth = $dbh->prepare( qq{ select * from feeds order by last_update desc } );
+ $sth->execute;
+ while (my $row = $sth->fetchrow_hashref) {
+ my $feed_entry = XML::Feed::Entry->new($type);
+ $feed_entry->title( $row->{name} );
+ $feed_entry->link( $row->{url} );
+ $feed_entry->issued( DateTime::Format::Flexible->build( $row->{last_update} ) );
+ $feed_entry->content(
+ '' . dump( $row ) . ']]>'
+ );
+ $feed->add_entry( $feed_entry );
+ }
+
+ } else {
+ _log "unknown rss request ",$request->url;
+ return RC_DENY;
}
$response->content( $feed->as_xml );
@@ -1011,21 +1137,24 @@
warn "$@";
}
- $response->content_type("text/html; charset=$ENCODING");
+ $response->content_type("text/html; charset=UTF-8");
my $html =
- qq{
}; - if ($request->url =~ m#/history#) { + } + . $cloud->html(500) + . qq{
};
+
+ if ($request->url =~ m#/tags?#) {
+ # nop
+ } elsif ($request->url =~ m#/history#) {
my $sth = $dbh->prepare(qq{
select date(time) as date,count(*) as nr,sum(length(message)) as len
from log
@@ -1057,9 +1186,9 @@
$cal->weekdays('MON','TUE','WED','THU','FRI');
($l_yyyy,$l_mm) = ($yyyy,$mm);
}
- $cal->setcontent($dd, qq{
- $row->{nr}
$row->{len}
- });
+ $cal->setcontent($dd, qq[
+ $row->{nr}
$row->{len}
+ ]) if $cal;
}
$html .= qq{