--- trunk/bin/irc-logger.pl 2008/02/29 22:11:07 83
+++ trunk/bin/irc-logger.pl 2008/03/07 00:31:58 88
@@ -22,6 +22,10 @@
Name of log file
+=item --follow=file.log
+
+Follows new messages in file
+
=back
=head1 DESCRIPTION
@@ -46,15 +50,11 @@
$CHANNEL = '#irc-logger' if ($HOSTNAME =~ m/llin/);
my $IRC_ALIAS = "log";
-my %FOLLOWS =
- (
- ACCESS => "/var/log/apache/access.log",
- ERROR => "/var/log/apache/error.log",
- );
+# default log to follow and announce messages
+my $follows_path = 'follows.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;
@@ -62,18 +62,19 @@
# number of last tags to keep in circular buffer
my $last_x_tags = 50;
+# 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 HTTP::Status;
use DBI;
-use Encode qw/from_to is_utf8/;
use Regexp::Common qw /URI/;
use CGI::Simple;
use HTML::TagCloud;
@@ -96,19 +97,38 @@
my $log_path;
GetOptions(
'import-dircproxy:s' => \$import_dircproxy,
+ 'follows:s' => \$follows_path,
'log:s' => \$log_path,
);
-$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(" ",@_) . $/;
}
+# LOG following
+
+my %FOLLOWS =
+ (
+# ACCESS => "/var/log/apache/access.log",
+# ERROR => "/var/log/apache/error.log",
+ );
+
+sub add_follow_path {
+ my $path = shift;
+ my $name = $path;
+ $name =~ s/\..*$//;
+ warn "# using $path to announce messages from $name\n";
+ $FOLLOWS{$name} = $path;
+}
+
+add_follow_path( $follows_path ) if ( -e $follows_path );
+
# HTML formatters
my %escape = ('<'=>'<', '>'=>'>', '&'=>'&', '"'=>'"');
@@ -152,9 +172,10 @@
};
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(),
@@ -168,17 +189,31 @@
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 ) {
@@ -242,7 +277,7 @@
-my $sth = $dbh->prepare(qq{
+my $sth_insert_log = $dbh->prepare(qq{
insert into log
(channel, me, nick, message, time)
values (?,?,?,?,?)
@@ -494,7 +529,6 @@
return unless ($arg->{id} && $arg->{message});
my $m = $arg->{message};
- from_to('UTF-8', 'iso-8859-2', $m) if (is_utf8($m));
my @tags;
@@ -563,9 +597,7 @@
$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 );
}
@@ -604,6 +636,97 @@
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: ' , $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 ( $last_t - $t > $rss_min_delay ) {
+ $_rss->{last_poll} = $t;
+ _log rss_fetch_all( $kernel );
+ }
+}
+
+# seed rss seen cache so we won't send out all items on startup
+_log rss_fetch_all;
#
# POE handing part
@@ -615,16 +738,13 @@
POE::Component::IRC->new($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 {
@@ -655,15 +775,16 @@
},
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);
@@ -674,7 +795,6 @@
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;
@@ -718,7 +838,6 @@
foreach my $res (get_from_log( limit => $limit )) {
_log "last: $res";
- from_to($res, $ENCODING, 'UTF-8');
$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res );
}
@@ -733,7 +852,6 @@
search => $what,
)) {
_log "search [$what]: $res";
- from_to($res, $ENCODING, 'UTF-8');
$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res );
}
@@ -803,14 +921,34 @@
$res = "config option $op doesn't exist";
}
}
+ } elsif ($msg =~ m/^rss-update/) {
+ $res = rss_fetch_all( $_[KERNEL] );
+ } elsif ($msg =~ m/^rss-clean/) {
+ $_rss = undef;
+ $res = "OK, cleaned RSS cache";
+ } elsif ($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 );
+ warn "## SQL $q ( $2 | $3 )\n";
+ eval { $sth->execute( $2, $3 ) };
+ }
+
+ $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];
@@ -849,66 +987,6 @@
"";
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);
- }
},
);
@@ -983,7 +1061,7 @@
my $search = $q->param('search') || $q->param('grep') || '';
- if ($request->url =~ m#/rss(?:/(tags|last-tag)\w*(?:=(\d+))?)?#i) {
+ if ($request->url =~ m#/rss(?:/(tags|last-tag|follow.*)\w*(?:=(\d+))?)?#i) {
my $show = lc($1);
my $nr = $2;
@@ -995,6 +1073,7 @@
#warn "create $type feed from ",dump( @last_tags );
my $feed = XML::Feed->new( $type );
+ $feed->link( $url );
if ( $show eq 'tags' ) {
$nr ||= 50;
@@ -1021,7 +1100,6 @@
$nr = $last_x_tags if $nr > $last_x_tags;
$feed->title( "last $nr tagged messages from $CHANNEL" );
- $feed->link( $url );
$feed->description( "collects messages which have tags// in them" );
foreach my $m ( @last_tags ) {
@@ -1036,7 +1114,6 @@
my $message = $filter->{message}->( $m->{message} );
$message .= "
\n" unless $message =~ m!<(/p|br/?)>!;
# warn "## message = $message\n";
- from_to( $message, $ENCODING, 'UTF-8' );
#$feed_entry->summary(
$feed_entry->content(
@@ -1050,8 +1127,25 @@
}
+ } 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 {
- warn "!! unknown rss request for $show\n";
+ _log "unknown rss request ",$request->url;
return RC_DENY;
}
@@ -1063,7 +1157,7 @@
warn "$@";
}
- $response->content_type("text/html; charset=$ENCODING");
+ $response->content_type("text/html; charset=UTF-8");
my $html =
qq{