--- trunk/irc-logger.pl	2006/03/12 14:19:00	13
+++ trunk/irc-logger.pl	2006/06/16 21:28:46	30
@@ -18,7 +18,7 @@
 
 ## CONFIG
 
-my $NICK = 'irc-logger-dev';
+my $NICK = 'irc-logger';
 my $CONNECT =
   {Server => 'irc.freenode.net',
    Nick => $NICK,
@@ -33,7 +33,9 @@
    ERROR => "/var/log/apache/error.log",
   );
 
-my $DSN = 'DBI:Pg:dbname=irc-logger';
+my $DSN = 'DBI:Pg:dbname=' . $NICK;
+
+my $ENCODING = 'ISO-8859-2';
 
 ## END CONFIG
 
@@ -42,18 +44,26 @@
 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;
 
 my $dbh = DBI->connect($DSN,"","", { RaiseError => 1, AutoCommit => 1 }) || die $DBI::errstr;
 
-=for SQL schema
+eval {
+	$dbh->do(qq{ select count(*) from log });
+};
+
+if ($@) {
+	warn "creating database table in $DSN\n";
+	$dbh->do(<<'_SQL_SCHEMA_');
 
-$dbh->do(qq{
 create table log (
 	id serial,
 	time timestamp default now(),
 	channel text not null,
+	me boolean default false,
 	nick text not null,
 	message text not null,
 	primary key(id)
@@ -63,16 +73,17 @@
 create index log_channel on log(channel);
 create index log_nick on log(nick);
 
-});
-
-=cut
+_SQL_SCHEMA_
+}
 
 my $sth = $dbh->prepare(qq{
 insert into log
-	(channel, nick, message)
-values (?,?,?)
+	(channel, me, nick, message)
+values (?,?,?,?)
 });
 
+my $tags;
+
 =head2 get_from_log
 
  my @messages = get_from_log(
@@ -82,10 +93,23 @@
 		time => '{%s} ',
 		time_channel => '{%s %s} ',
 		nick => '%s: ',
+		me_nick => '***%s ',
 		message => '%s',
 	},
+	filter => {
+		message => sub {
+			# modify message content
+			return shift;
+		}
+	},
+	context => 5,
  );
 
+Order is important. Fields are first passed through C<filter> (if available) and
+then throgh C<< sprintf($fmt->{message}, $message >> if available.
+
+C<context> defines number of messages around each search hit for display.
+
 =cut
 
 sub get_from_log {
@@ -94,28 +118,49 @@
 	$args->{limit} ||= 10;
 
 	$args->{fmt} ||= {
+		date => '[%s] ',
 		time => '{%s} ',
 		time_channel => '{%s %s} ',
 		nick => '%s: ',
+		me_nick => '***%s ',
 		message => '%s',
 	};
 
-	my $sql = qq{
+	my $sql_message = qq{
 		select
 			time::date as date,
 			time::time as time,
 			channel,
+			me,
 			nick,
 			message
 		from log
 	};
-	$sql .= " where message ilike ? " if ($args->{search});
+
+	my $sql_context = qq{
+		select
+			id
+		from log
+	};
+
+	my $context = $1 if ($args->{search} && $args->{search} =~ s/\s*\+(\d+)\s*/ /);
+
+	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 .= " order by log.time desc";
 	$sql .= " limit " . $args->{limit};
 
 	my $sth = $dbh->prepare( $sql );
-	if ($args->{search}) {
-		$sth->execute( $args->{search} );
+	if (my $search = $args->{search}) {
+		$search =~ s/^\s+//;
+		$search =~ s/\s+$//;
+		$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";
 	} else {
 		$sth->execute();
 	}
@@ -132,18 +177,43 @@
 		unshift @rows, $row;
 	}
 
-	my @msgs;
+	my @msgs = (
+		"Showing " . ($#rows + 1) . " messages..."
+	);
+
+	if ($context) {
+		my @ids = @rows;
+		@rows = ();
+
+		my $last_to = 0;
+
+		my $sth = $dbh->prepare( $sql_message . qq{ where id >= ? and id < ? } );
+		foreach my $row_id (sort { $a->{id} <=> $b->{id} } @ids) {
+			my $id = $row_id->{id} || die "can't find id in row";
+	
+			my ($from, $to) = ($id - $context, $id + $context);
+			$from = $last_to if ($from < $last_to);
+			$last_to = $to;
+			$sth->execute( $from, $to );
+
+			#warn "## id: $id from: $from to: $to returned: ", $sth->rows, "\n";
+
+			while (my $row = $sth->fetchrow_hashref) {
+				push @rows, $row;
+			}
+
+		}
+	}
 
 	foreach my $row (@rows) {
 
 		$row->{time} =~ s#\.\d+##;
 
-		my $t;
-		$t = $row->{date} . ' ' if ($last_row->{date} ne $row->{date});
-		$t .= $row->{time};
-
 		my $msg = '';
 
+		$msg = sprintf($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});
 		} else {
@@ -152,12 +222,32 @@
 
 		my $append = 1;
 
-		if ($last_row->{nick} ne $row->{nick}) {
-			$msg .= sprintf($args->{fmt}->{nick}, $row->{nick});
+		my $nick = $row->{nick};
+		if ($nick =~ s/^_*(.*?)_*$/$1/) {
+			$row->{nick} = $nick;
+		}
+
+		if ($last_row->{nick} ne $nick) {
+			# obfu way to find format for me_nick if needed or fallback to default
+			my $fmt = $row->{me} ? ( $args->{fmt}->{me_nick} || $args->{fmt}->{nick} ) : $args->{fmt}->{nick};
+			$fmt ||= '%s';
+
+			$nick = $args->{filter}->{nick}->($nick) if (ref($args->{filter}->{nick}) eq 'CODE');
+
+			$msg .= sprintf( $fmt, $nick );
 			$append = 0;
 		}
 
-		$msg .= sprintf($args->{fmt}->{message}, $row->{message});
+		$args->{fmt}->{message} ||= '%s';
+		if (ref($args->{filter}->{message}) eq 'CODE') {
+			$msg .= sprintf($args->{fmt}->{message},
+				$args->{filter}->{message}->(
+					$row->{message}
+				)
+			);
+		} else {
+			$msg .= sprintf($args->{fmt}->{message}, $row->{message});
+		}
 
 		if ($append && @msgs) {
 			$msgs[$#msgs] .= " " . $msg;
@@ -196,16 +286,31 @@
 		my $channel = $_[ARG1]->[0];
 		my $msg = $_[ARG2];
 
-		from_to($msg, 'UTF-8', 'ISO-8859-2');
+		from_to($msg, 'UTF-8', $ENCODING);
 
 		print "$channel: <$nick> $msg\n";
-		$sth->execute($channel, $nick, $msg);
+		$sth->execute($channel, 0, $nick, $msg);
+		add_tag( id => $dbh->last_insert_id(undef,undef,"log",undef),
+			message => $msg);
+    },
+    irc_ctcp_action => sub {
+		my $kernel = $_[KERNEL];
+		my $nick = (split /!/, $_[ARG0])[0];
+		my $channel = $_[ARG1]->[0];
+		my $msg = $_[ARG2];
+
+		from_to($msg, 'UTF-8', $ENCODING);
+
+		print "$channel ***$nick $msg\n";
+		$sth->execute($channel, 1, $nick, $msg);
+		add_tag( id => $dbh->last_insert_id(undef,undef,"log",undef),
+			message => $msg);
     },
 	irc_msg => sub {
 		my $kernel = $_[KERNEL];
 		my $nick = (split /!/, $_[ARG0])[0];
 		my $msg = $_[ARG2];
-		from_to($msg, 'UTF-8', 'ISO-8859-2');
+		from_to($msg, 'UTF-8', $ENCODING);
 
 		my $res = "unknown command '$msg', try /msg $NICK help!";
 		my @out;
@@ -240,19 +345,22 @@
 
 			foreach my $res (get_from_log( limit => $1 )) {
 				print "last: $res\n";
-				from_to($res, 'ISO-8859-2', 'UTF-8');
+				from_to($res, $ENCODING, 'UTF-8');
 				$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res );
 			}
 
 			$res = '';
 
-		} elsif ($msg =~ m/^(search|grep)\s+(.*)$/i) {
+		} elsif ($msg =~ m/^(search|grep)\s+(.*)\s*$/i) {
 
 			my $what = $2;
 
-			foreach my $res (get_from_log( limit => 20, search => "%${what}%" )) {
+			foreach my $res (get_from_log(
+					limit => 20,
+					search => $what,
+				)) {
 				print "search [$what]: $res\n";
-				from_to($res, 'ISO-8859-2', 'UTF-8');
+				from_to($res, $ENCODING, 'UTF-8');
 				$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res );
 			}
 
@@ -262,7 +370,7 @@
 
 		if ($res) {
 			print ">> [$nick] $res\n";
-			from_to($res, 'ISO-8859-2', 'UTF-8');
+			from_to($res, $ENCODING, 'UTF-8');
 			$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res );
 		}
 
@@ -375,39 +483,143 @@
    },
   );
 
+# 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#\b([\w-_]+)//##s) {
+		my $tag = $1;
+		next if (! $tag || $tag =~ m/https?:/i);
+		push @{ $tags->{$tag} }, $arg->{id};
+	}
+}
+
+=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;
+
 # http server
 
 my $httpd = POE::Component::Server::HTTP->new(
-	Port => 8000,
+	Port => $NICK =~ m/-dev/ ? 8001 : 8000,
 	ContentHandler => { '/' => \&root_handler },
 	Headers        => { Server => 'irc-logger' },
 );
 
+my %escape = ('<'=>'&lt;', '>'=>'&gt;', '&'=>'&amp;', '"'=>'&quot;');
+my $escape_re  = join '|' => keys %escape;
+
 my $style = <<'_END_OF_STYLE_';
+p { margin: 0; padding: 0.1em; }
 .time, .channel { color: #808080; font-size: 60%; }
-.nick { color: #0000ff; font-size: 80%; }
+.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; }
+.col-0 { background: #ffff66 }
+.col-1 { background: #a0ffff }
+.col-2 { background: #99ff99 }
+.col-3 { background: #ff9999 }
+.col-4 { background: #ff66ff }
 _END_OF_STYLE_
 
+my $max_color = 4;
+
+my %nick_enumerator;
+
 sub root_handler {
 	my ($request, $response) = @_;
 	$response->code(RC_OK);
-	$response->content_type('text/html');
+	$response->content_type("text/html; charset=$ENCODING");
+
+	my $q;
+
+	if ( $request->method eq 'POST' ) {
+		$q = new CGI::Simple( $request->content );
+	} elsif ( $request->uri =~ /\?(.+)$/ ) {
+		$q = new CGI::Simple( $1 );
+	} else {
+		$q = new CGI::Simple;
+	}
+
+	my $search = $q->param('search') || $q->param('grep') || '';
+
 	$response->content(
-		qq{<html><head><title>$NICK</title><style type="text/css">$style</style></head><body>} .
-	        "irc-logger url: " . $request->uri . '<br/>' .
-		join("<br/>",
+		qq{<html><head><title>$NICK</title><style type="text/css">$style} .
+		$cloud->css .
+		qq{</style></head><body>} .
+		qq{
+		<form method="post" class="search">
+		<input type="text" name="search" value="$search" size="10">
+		<input type="submit" value="search">
+		</form>
+		} .
+		qq{<div>} . $cloud->html(500) . qq{</div>} .
+		qq{<p>} .
+		join("</p><p>",
 			get_from_log(
-				limit => 100,
+				limit => $q->param('last') || 100,
+				search => $search || undef,
+				tag => $q->param('tag') || undef,
 				fmt => {
+					date => '<hr size="1" style="clear: both;"/><div class="date">%s</div> ',
 					time => '<span class="time">%s</span> ',
 					time_channel => '<span class="channel">%s %s</span> ',
-					nick => '<span class="nick">%s:</span> ',
+					nick => '%s:&nbsp;',
+					me_nick => '***%s&nbsp;',
 					message => '<span class="message">%s</span>',
 				},
+				filter => {
+					message => sub {
+						my $m = shift || return;
+						$m =~ s/($escape_re)/$escape{$1}/gs;
+						$m =~ s#($RE{URI}{HTTP})#<a href="$1">$1</a>#gs;
+						return $m;
+					},
+					nick => sub {
+						my $n = shift || return;
+						if (! $nick_enumerator{$n})  {
+							my $max = scalar keys %nick_enumerator;
+							$nick_enumerator{$n} = $max + 1;
+						}
+						return '<span class="nick col-' .
+							( $nick_enumerator{$n} % $max_color ) .
+							'">' . $n . '</span>';
+					},
+				},
 			)
 		) .
-		qq{</body></html>}
+		qq{</p></body></html>}
 	);
 	return RC_OK;
 }