--- trunk/irc-logger.pl	2006/05/20 10:30:45	26
+++ trunk/irc-logger.pl	2006/10/24 12:51:49	41
@@ -10,6 +10,14 @@
 
 ./irc-logger.pl
 
+=head2 Options
+
+=over 4
+
+=item --import-dircproxy=filename
+
+Import log from C<dircproxy> to C<irc-logger> database
+
 =head1 DESCRIPTION
 
 log all conversation on irc channel
@@ -18,13 +26,17 @@
 
 ## CONFIG
 
+my $HOSTNAME = `hostname`;
+
 my $NICK = 'irc-logger';
+$NICK .= '-dev' if ($HOSTNAME =~ m/llin/);
 my $CONNECT =
   {Server => 'irc.freenode.net',
    Nick => $NICK,
    Ircname => "try /msg $NICK help",
   };
 my $CHANNEL = '#razmjenavjestina';
+$CHANNEL = '#irc-logger' if ($HOSTNAME =~ m/llin/);
 my $IRC_ALIAS = "log";
 
 my %FOLLOWS =
@@ -36,6 +48,9 @@
 my $DSN = 'DBI:Pg:dbname=' . $NICK;
 
 my $ENCODING = 'ISO-8859-2';
+my $TIMESTAMP = '%Y-%m-%d %H:%M:%S';
+
+my $sleep_on_error = 5;
 
 ## END CONFIG
 
@@ -44,10 +59,19 @@
 use POE qw(Component::IRC Wheel::FollowTail Component::Server::HTTP);
 use HTTP::Status;
 use DBI;
-use Encode qw/from_to/;
+use Encode qw/from_to is_utf8/;
 use Regexp::Common qw /URI/;
 use CGI::Simple;
-
+use HTML::TagCloud;
+use POSIX qw/strftime/;
+use HTML::CalendarMonthSimple;
+use Getopt::Long;
+use DateTime;
+
+my $import_dircproxy;
+GetOptions(
+	'import-dircproxy:s' => \$import_dircproxy,
+);
 
 my $dbh = DBI->connect($DSN,"","", { RaiseError => 1, AutoCommit => 1 }) || die $DBI::errstr;
 
@@ -78,10 +102,13 @@
 
 my $sth = $dbh->prepare(qq{
 insert into log
-	(channel, me, nick, message)
-values (?,?,?,?)
+	(channel, me, nick, message, time)
+values (?,?,?,?,?)
 });
 
+my $tags;
+my $tag_regex = '\b([\w-_]+)//';
+
 =head2 get_from_log
 
  my @messages = get_from_log(
@@ -113,8 +140,6 @@
 sub get_from_log {
 	my $args = {@_};
 
-	$args->{limit} ||= 10;
-
 	$args->{fmt} ||= {
 		date => '[%s] ',
 		time => '{%s} ',
@@ -146,8 +171,10 @@
 	my $sql = $context ? $sql_context : $sql_message;
 
 	$sql .= " where message ilike ? or nick ilike ? " if ($args->{search});
+	$sql .= " where id in (" . join(",", @{ $tags->{ $args->{tag} } }) . ") " if ($args->{tag} && $tags->{ $args->{tag} });
+	$sql .= " where date(time) = ? " if ($args->{date});
 	$sql .= " order by log.time desc";
-	$sql .= " limit " . $args->{limit};
+	$sql .= " limit " . $args->{limit} if ($args->{limit});
 
 	my $sth = $dbh->prepare( $sql );
 	if (my $search = $args->{search}) {
@@ -155,6 +182,12 @@
 		$search =~ s/\s+$//;
 		$sth->execute( ( '%' . $search . '%' ) x 2 );
 		warn "search for '$search' returned ", $sth->rows, " results ", $context || '', "\n";
+	} elsif (my $tag = $args->{tag}) {
+		$sth->execute();
+		warn "tag '$tag' returned ", $sth->rows, " results ", $context || '', "\n";
+	} elsif (my $date = $args->{date}) {
+		$sth->execute($date);
+		warn "found ", $sth->rows, " messages for date $date ", $context || '', "\n";
 	} else {
 		$sth->execute();
 	}
@@ -199,19 +232,29 @@
 		}
 	}
 
+	# sprintf which can take coderef as first parametar
+	sub cr_sprintf {
+		my $fmt = shift || return;
+		if (ref($fmt) eq 'CODE') {
+			$fmt->(@_);
+		} else {
+			sprintf($fmt, @_);
+		}
+	}
+
 	foreach my $row (@rows) {
 
 		$row->{time} =~ s#\.\d+##;
 
 		my $msg = '';
 
-		$msg = sprintf($args->{fmt}->{date}, $row->{date}) . ' ' if ($last_row->{date} ne $row->{date});
+		$msg = cr_sprintf($args->{fmt}->{date}, $row->{date}) . ' ' if ($last_row->{date} ne $row->{date});
 		my $t = $row->{time};
 
 		if ($last_row->{channel} ne $row->{channel}) {
-			$msg .= sprintf($args->{fmt}->{time_channel}, $t, $row->{channel});
+			$msg .= cr_sprintf($args->{fmt}->{time_channel}, $t, $row->{channel});
 		} else {
-			$msg .= sprintf($args->{fmt}->{time}, $t);
+			$msg .= cr_sprintf($args->{fmt}->{time}, $t);
 		}
 
 		my $append = 1;
@@ -228,19 +271,19 @@
 
 			$nick = $args->{filter}->{nick}->($nick) if (ref($args->{filter}->{nick}) eq 'CODE');
 
-			$msg .= sprintf( $fmt, $nick );
+			$msg .= cr_sprintf( $fmt, $nick );
 			$append = 0;
 		}
 
 		$args->{fmt}->{message} ||= '%s';
 		if (ref($args->{filter}->{message}) eq 'CODE') {
-			$msg .= sprintf($args->{fmt}->{message},
+			$msg .= cr_sprintf($args->{fmt}->{message},
 				$args->{filter}->{message}->(
 					$row->{message}
 				)
 			);
 		} else {
-			$msg .= sprintf($args->{fmt}->{message}, $row->{message});
+			$msg .= cr_sprintf($args->{fmt}->{message}, $row->{message});
 		}
 
 		if ($append && @msgs) {
@@ -255,14 +298,133 @@
 	return @msgs;
 }
 
+# tags support
+
+my $cloud = HTML::TagCloud->new;
+
+=head2 add_tag
+
+ add_tag( id => 42, message => 'irc message' );
+
+=cut
+
+sub add_tag {
+	my $arg = {@_};
+
+	return unless ($arg->{id} && $arg->{message});
+
+	my $m = $arg->{message};
+	from_to('UTF-8', 'iso-8859-2', $m) if (is_utf8($m));
+
+	while ($m =~ s#$tag_regex##s) {
+		my $tag = $1;
+		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);
+	}
+}
+
+=head2 seed_tags
+
+Read all tags from database and create in-memory cache for tags
+
+=cut
+
+sub seed_tags {
+	my $sth = $dbh->prepare(qq{ select id,message from log where message like '%//%' });
+	$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);
+	}
+}
+
+seed_tags;
+
+
+=head2 save_message
+
+  save_message(
+  	channel => '#foobar',
+	me => 0,
+	nick => 'dpavlin',
+	msg => 'test message',
+	time => '2006-06-25 18:57:18',
+  );
+
+C<time> is optional, it will use C<< now() >> if it's not available.
+
+C<me> if not specified will be C<0> (not C</me> message)
+
+=cut
+
+sub save_message {
+	my $a = {@_};
+	$a->{me} ||= 0;
+	$a->{time} ||= strftime($TIMESTAMP,localtime());
+
+	print
+		$a->{time}, " ",
+		$a->{channel}, " ",
+		$a->{me} ? "***" . $a->{nick} : "<" . $a->{nick} . ">",
+		" " . $a->{msg} . "\n";
+
+	from_to($a->{msg}, 'UTF-8', $ENCODING);
+
+	$sth->execute($a->{channel}, $a->{me}, $a->{nick}, $a->{msg}, $a->{time});
+	add_tag( id => $dbh->last_insert_id(undef,undef,"log",undef),
+		message => $a->{msg});
+}
+
+if ($import_dircproxy) {
+	open(my $l, $import_dircproxy) || die "can't open $import_dircproxy: $!";
+	warn "importing $import_dircproxy...\n";
+	my $tz_offset = 2 * 60 * 60;	# TZ GMT+2
+	while(<$l>) {
+		chomp;
+		if (/^@(\d+)\s(\S+)\s(.+)$/) {
+			my ($time, $nick, $msg) = ($1,$2,$3);
+
+			my $dt = DateTime->from_epoch( epoch => $time + $tz_offset );
+
+			my $me = 0;
+			$me = 1 if ($nick =~ m/^\[\S+]/);
+			$nick =~ s/^[\[<]([^!]+).*$/$1/;
+
+			$msg =~ s/^ACTION\s+// if ($me);
+
+			save_message(
+				channel => $CHANNEL,
+				me => $me,
+				nick => $nick,
+				msg => $msg,
+				time => $dt->ymd . " " . $dt->hms,
+			) if ($nick !~ m/^-/);
+
+		} else {
+			warn "can't parse: $_\n";
+		}
+	}
+	close($l);
+	warn "import over\n";
+	exit;
+}
+
+
+#
+# POE handing part
+#
 
 my $SKIPPING = 0;               # if skipping, how many we've done
 my $SEND_QUEUE;                 # cache
 
 POE::Component::IRC->new($IRC_ALIAS);
 
-POE::Session->create
-  (inline_states =>
+POE::Session->create( inline_states =>
    {_start => sub {      
 		$_[KERNEL]->post($IRC_ALIAS => register => 'all');
 		$_[KERNEL]->post($IRC_ALIAS => connect => $CONNECT);
@@ -280,10 +442,7 @@
 		my $channel = $_[ARG1]->[0];
 		my $msg = $_[ARG2];
 
-		from_to($msg, 'UTF-8', $ENCODING);
-
-		print "$channel: <$nick> $msg\n";
-		$sth->execute($channel, 0, $nick, $msg);
+		save_message( channel => $channel, me => 0, nick => $nick, msg => $msg);
     },
     irc_ctcp_action => sub {
 		my $kernel = $_[KERNEL];
@@ -291,10 +450,7 @@
 		my $channel = $_[ARG1]->[0];
 		my $msg = $_[ARG2];
 
-		from_to($msg, 'UTF-8', $ENCODING);
-
-		print "$channel ***$nick $msg\n";
-		$sth->execute($channel, 1, $nick, $msg);
+		save_message( channel => $channel, me => 1, nick => $nick, msg => $msg);
     },
 	irc_msg => sub {
 		my $kernel = $_[KERNEL];
@@ -322,18 +478,25 @@
 			my $nr = $1 || 10;
 
 			my $sth = $dbh->prepare(qq{
-				select nick,count(*) from log group by nick order by count desc limit $nr
+				select
+					nick,
+					count(*) as count,
+					sum(length(message)) as len
+				from log
+				group by nick
+				order by len desc,count desc
+				limit $nr
 			});
 			$sth->execute();
 			$res = "Top $nr users: ";
 			my @users;
 			while (my $row = $sth->fetchrow_hashref) {
-				push @users,$row->{nick} . ': ' . $row->{count};
+				push @users,$row->{nick} . ': ' . $row->{count} . '/' . $row->{len} . '=' . sprintf("%.2f", $row->{len}/$row->{count});
 			}
 			$res .= join(" | ", @users);
 		} elsif ($msg =~ m/^last.*?\s*(\d*)/i) {
 
-			foreach my $res (get_from_log( limit => $1 )) {
+			foreach my $res (get_from_log( limit => ($1 || 100) )) {
 				print "last: $res\n";
 				from_to($res, $ENCODING, 'UTF-8');
 				$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res );
@@ -379,35 +542,27 @@
 		warn "## indetify $NICK\n";
 		$_[KERNEL]->post( $IRC_ALIAS => privmsg => 'nickserv', "IDENTIFY $NICK" );
 	},
+	irc_disconnected => sub {
+		warn "## disconnected, reconnecting again\n";
+		$_[KERNEL]->post($IRC_ALIAS => connect => $CONNECT);
+	},
+	irc_socketerr => sub {
+		warn "## socket error... sleeping for $sleep_on_error seconds and retry";
+		sleep($sleep_on_error);
+		$_[KERNEL]->post($IRC_ALIAS => connect => $CONNECT);
+	},
 #	irc_433 => sub {
 #		print "# irc_433: ",$_[ARG1], "\n";
 #		warn "## indetify $NICK\n";
 #		$_[KERNEL]->post( $IRC_ALIAS => privmsg => 'nickserv', "IDENTIFY $NICK" );
 #	},
-	irc_372 => sub {
-		print "MOTD: ", $_[ARG1], "\n";
-	},
-	irc_snotice => sub {
-		print "(server notice): ", $_[ARG0], "\n";
-	},
-    (map
-     {
-       ;"irc_$_" => sub { }}
-     qw(
-		)),
-#	join
-#     	ctcp_version
-#	connected snotice ctcp_action ping notice mode part quit
-#	001 002 003 004 005
-#	250 251 252 253 254 265 266
-#	332 333 353 366 372 375 376
-#	477
     _child => sub {},
     _default => sub {
-      printf "%s: session %s caught an unhandled %s event.\n",
-        scalar localtime(), $_[SESSION]->ID, $_[ARG0];
-      print "The $_[ARG0] event was given these parameters: ",
-        join(" ", map({"ARRAY" eq ref $_ ? "[@$_]" : "$_"} @{$_[ARG1]})), "\n";
+		printf "%s #%s %s %s\n",
+			strftime($TIMESTAMP,localtime()), $_[SESSION]->ID, $_[ARG0],
+			ref($_[ARG1]) eq "ARRAY"	?	join(",", map { ref($_) eq "ARRAY" ? join(";", @{$_}) : $_ } @{ $_[ARG1] })	:
+			$_[ARG1]					?	$_[ARG1]					:
+			"";
       0;                        # false for signals
     },
     my_add => sub {
@@ -487,7 +642,7 @@
 my $style = <<'_END_OF_STYLE_';
 p { margin: 0; padding: 0.1em; }
 .time, .channel { color: #808080; font-size: 60%; }
-.date { float: right; background: #404040; color: #e0e0e0; font-size: 120%; padding: 0.5em; border-top: 1px dashed #e0e0e0; }
+.date { float: right; background: #e0e0e0; color: #404040; font-size: 120%; padding: 0.25em; border: 1px dashed #808080; }
 .nick { color: #000000; font-size: 80%; padding: 2px; font-family: courier, courier new, monospace ; }
 .message { color: #000000; font-size: 100%; }
 .search { float: right; }
@@ -496,6 +651,9 @@
 .col-2 { background: #99ff99 }
 .col-3 { background: #ff9999 }
 .col-4 { background: #ff66ff }
+a:link.tag, a:visited.tag { border: 1px dashed #ccc; backgound: #ccc; text-decoration: none }
+a:hover.tag { border: 1px solid #eee }
+hr { border: 1px dashed #ccc; height: 1px; clear: both; }
 _END_OF_STYLE_
 
 my $max_color = 4;
@@ -519,20 +677,55 @@
 
 	my $search = $q->param('search') || $q->param('grep') || '';
 
-	$response->content(
-		qq{<html><head><title>$NICK</title><style type="text/css">$style</style></head><body>
-		<form method="post" class="search">
+	my $html =
+		qq{<html><head><title>$NICK</title><style type="text/css">$style} .
+		$cloud->css .
+		qq{</style></head><body>} .
+		qq{
+		<form method="post" class="search" action="/">
 		<input type="text" name="search" value="$search" size="10">
 		<input type="submit" value="search">
 		</form>
-		<p>
 		} .
-		join("</p><p>",
+		$cloud->html(500) .
+		qq{<p>};
+	if ($request->url =~ m#/history#) {
+		my $sth = $dbh->prepare(qq{
+			select date(time) as date,count(*) as nr
+				from log
+				group by date(time)
+				order by date(time) desc
+		});
+		$sth->execute();
+		my ($l_yyyy,$l_mm) = (0,0);
+		my $cal;
+		while (my $row = $sth->fetchrow_hashref) {
+			# this is probably PostgreSQL specific, expects ISO date
+			my ($yyyy,$mm,$dd) = split(/-/, $row->{date});
+			if ($yyyy != $l_yyyy || $mm != $l_mm) {
+				$html .= $cal->as_HTML() if ($cal);
+				$cal = new HTML::CalendarMonthSimple('month'=>$mm,'year'=>$yyyy);
+				$cal->border(2);
+				($l_yyyy,$l_mm) = ($yyyy,$mm);
+			}
+			$cal->setcontent($dd, qq{
+				<a href="/?date=$row->{date}">$row->{nr}</a>
+			});
+		}
+		$html .= $cal->as_HTML() if ($cal);
+
+	} else {
+		$html .= join("</p><p>",
 			get_from_log(
-				limit => $q->param('last') || 100,
-				search => $q->param('search') || $q->param('grep') || undef,
+				limit => $q->param('last') || $q->param('date') ? undef : 100,
+				search => $search || undef,
+				tag => $q->param('tag') || undef,
+				date => $q->param('date') || undef,
 				fmt => {
-					date => '<hr size="1"/><div class="date">%s</div> ',
+					date => sub {
+						my $date = shift || return;
+						qq{<hr/><div class="date"><a href="/?date=$date">$date</a></div>};
+					},
 					time => '<span class="time">%s</span> ',
 					time_channel => '<span class="channel">%s %s</span> ',
 					nick => '%s:&nbsp;',
@@ -544,6 +737,7 @@
 						my $m = shift || return;
 						$m =~ s/($escape_re)/$escape{$1}/gs;
 						$m =~ s#($RE{URI}{HTTP})#<a href="$1">$1</a>#gs;
+						$m =~ s#$tag_regex#<a href="?tag=$1" class="tag">$1</a>#g;
 						return $m;
 					},
 					nick => sub {
@@ -558,9 +752,15 @@
 					},
 				},
 			)
-		) .
-		qq{</p></body></html>}
-	);
+		);
+	}
+
+	$html .= qq{</p>
+	<hr/>
+	<p>See <a href="/history">history</a> of all messages.</p>
+	</body></html>};
+
+	$response->content( $html );
 	return RC_OK;
 }