10 |
|
|
11 |
./irc-logger.pl |
./irc-logger.pl |
12 |
|
|
13 |
|
=head2 Options |
14 |
|
|
15 |
|
=over 4 |
16 |
|
|
17 |
|
=item --import-dircproxy=filename |
18 |
|
|
19 |
|
Import log from C<dircproxy> to C<irc-logger> database |
20 |
|
|
21 |
=head1 DESCRIPTION |
=head1 DESCRIPTION |
22 |
|
|
23 |
log all conversation on irc channel |
log all conversation on irc channel |
50 |
my $ENCODING = 'ISO-8859-2'; |
my $ENCODING = 'ISO-8859-2'; |
51 |
my $TIMESTAMP = '%Y-%m-%d %H:%M:%S'; |
my $TIMESTAMP = '%Y-%m-%d %H:%M:%S'; |
52 |
|
|
53 |
|
my $sleep_on_error = 5; |
54 |
|
|
55 |
## END CONFIG |
## END CONFIG |
56 |
|
|
57 |
|
|
64 |
use CGI::Simple; |
use CGI::Simple; |
65 |
use HTML::TagCloud; |
use HTML::TagCloud; |
66 |
use POSIX qw/strftime/; |
use POSIX qw/strftime/; |
67 |
|
use HTML::CalendarMonthSimple; |
68 |
|
use Getopt::Long; |
69 |
|
use DateTime; |
70 |
|
|
71 |
|
my $import_dircproxy; |
72 |
|
GetOptions( |
73 |
|
'import-dircproxy:s' => \$import_dircproxy, |
74 |
|
); |
75 |
|
|
76 |
my $dbh = DBI->connect($DSN,"","", { RaiseError => 1, AutoCommit => 1 }) || die $DBI::errstr; |
my $dbh = DBI->connect($DSN,"","", { RaiseError => 1, AutoCommit => 1 }) || die $DBI::errstr; |
77 |
|
|
102 |
|
|
103 |
my $sth = $dbh->prepare(qq{ |
my $sth = $dbh->prepare(qq{ |
104 |
insert into log |
insert into log |
105 |
(channel, me, nick, message) |
(channel, me, nick, message, time) |
106 |
values (?,?,?,?) |
values (?,?,?,?,?) |
107 |
}); |
}); |
108 |
|
|
109 |
my $tags; |
my $tags; |
140 |
sub get_from_log { |
sub get_from_log { |
141 |
my $args = {@_}; |
my $args = {@_}; |
142 |
|
|
|
$args->{limit} ||= 10; |
|
|
|
|
143 |
$args->{fmt} ||= { |
$args->{fmt} ||= { |
144 |
date => '[%s] ', |
date => '[%s] ', |
145 |
time => '{%s} ', |
time => '{%s} ', |
172 |
|
|
173 |
$sql .= " where message ilike ? or nick ilike ? " if ($args->{search}); |
$sql .= " where message ilike ? or nick ilike ? " if ($args->{search}); |
174 |
$sql .= " where id in (" . join(",", @{ $tags->{ $args->{tag} } }) . ") " if ($args->{tag} && $tags->{ $args->{tag} }); |
$sql .= " where id in (" . join(",", @{ $tags->{ $args->{tag} } }) . ") " if ($args->{tag} && $tags->{ $args->{tag} }); |
175 |
|
$sql .= " where date(time) = ? " if ($args->{date}); |
176 |
$sql .= " order by log.time desc"; |
$sql .= " order by log.time desc"; |
177 |
$sql .= " limit " . $args->{limit}; |
$sql .= " limit " . $args->{limit} if ($args->{limit}); |
178 |
|
|
179 |
my $sth = $dbh->prepare( $sql ); |
my $sth = $dbh->prepare( $sql ); |
180 |
if (my $search = $args->{search}) { |
if (my $search = $args->{search}) { |
185 |
} elsif (my $tag = $args->{tag}) { |
} elsif (my $tag = $args->{tag}) { |
186 |
$sth->execute(); |
$sth->execute(); |
187 |
warn "tag '$tag' returned ", $sth->rows, " results ", $context || '', "\n"; |
warn "tag '$tag' returned ", $sth->rows, " results ", $context || '', "\n"; |
188 |
|
} elsif (my $date = $args->{date}) { |
189 |
|
$sth->execute($date); |
190 |
|
warn "found ", $sth->rows, " messages for date $date ", $context || '', "\n"; |
191 |
} else { |
} else { |
192 |
$sth->execute(); |
$sth->execute(); |
193 |
} |
} |
232 |
} |
} |
233 |
} |
} |
234 |
|
|
235 |
|
# sprintf which can take coderef as first parametar |
236 |
|
sub cr_sprintf { |
237 |
|
my $fmt = shift || return; |
238 |
|
if (ref($fmt) eq 'CODE') { |
239 |
|
$fmt->(@_); |
240 |
|
} else { |
241 |
|
sprintf($fmt, @_); |
242 |
|
} |
243 |
|
} |
244 |
|
|
245 |
foreach my $row (@rows) { |
foreach my $row (@rows) { |
246 |
|
|
247 |
$row->{time} =~ s#\.\d+##; |
$row->{time} =~ s#\.\d+##; |
248 |
|
|
249 |
my $msg = ''; |
my $msg = ''; |
250 |
|
|
251 |
$msg = sprintf($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}); |
252 |
my $t = $row->{time}; |
my $t = $row->{time}; |
253 |
|
|
254 |
if ($last_row->{channel} ne $row->{channel}) { |
if ($last_row->{channel} ne $row->{channel}) { |
255 |
$msg .= sprintf($args->{fmt}->{time_channel}, $t, $row->{channel}); |
$msg .= cr_sprintf($args->{fmt}->{time_channel}, $t, $row->{channel}); |
256 |
} else { |
} else { |
257 |
$msg .= sprintf($args->{fmt}->{time}, $t); |
$msg .= cr_sprintf($args->{fmt}->{time}, $t); |
258 |
} |
} |
259 |
|
|
260 |
my $append = 1; |
my $append = 1; |
271 |
|
|
272 |
$nick = $args->{filter}->{nick}->($nick) if (ref($args->{filter}->{nick}) eq 'CODE'); |
$nick = $args->{filter}->{nick}->($nick) if (ref($args->{filter}->{nick}) eq 'CODE'); |
273 |
|
|
274 |
$msg .= sprintf( $fmt, $nick ); |
$msg .= cr_sprintf( $fmt, $nick ); |
275 |
$append = 0; |
$append = 0; |
276 |
} |
} |
277 |
|
|
278 |
$args->{fmt}->{message} ||= '%s'; |
$args->{fmt}->{message} ||= '%s'; |
279 |
if (ref($args->{filter}->{message}) eq 'CODE') { |
if (ref($args->{filter}->{message}) eq 'CODE') { |
280 |
$msg .= sprintf($args->{fmt}->{message}, |
$msg .= cr_sprintf($args->{fmt}->{message}, |
281 |
$args->{filter}->{message}->( |
$args->{filter}->{message}->( |
282 |
$row->{message} |
$row->{message} |
283 |
) |
) |
284 |
); |
); |
285 |
} else { |
} else { |
286 |
$msg .= sprintf($args->{fmt}->{message}, $row->{message}); |
$msg .= cr_sprintf($args->{fmt}->{message}, $row->{message}); |
287 |
} |
} |
288 |
|
|
289 |
if ($append && @msgs) { |
if ($append && @msgs) { |
298 |
return @msgs; |
return @msgs; |
299 |
} |
} |
300 |
|
|
301 |
|
# tags support |
302 |
|
|
303 |
|
my $cloud = HTML::TagCloud->new; |
304 |
|
|
305 |
|
=head2 add_tag |
306 |
|
|
307 |
|
add_tag( id => 42, message => 'irc message' ); |
308 |
|
|
309 |
|
=cut |
310 |
|
|
311 |
|
sub add_tag { |
312 |
|
my $arg = {@_}; |
313 |
|
|
314 |
|
return unless ($arg->{id} && $arg->{message}); |
315 |
|
|
316 |
|
my $m = $arg->{message}; |
317 |
|
from_to('UTF-8', 'iso-8859-2', $m) if (is_utf8($m)); |
318 |
|
|
319 |
|
while ($m =~ s#$tag_regex##s) { |
320 |
|
my $tag = $1; |
321 |
|
next if (! $tag || $tag =~ m/https?:/i); |
322 |
|
push @{ $tags->{$tag} }, $arg->{id}; |
323 |
|
#warn "+tag $tag: $arg->{id}\n"; |
324 |
|
$cloud->add($tag, "?tag=$tag", scalar @{$tags->{$tag}} + 1); |
325 |
|
} |
326 |
|
} |
327 |
|
|
328 |
|
=head2 seed_tags |
329 |
|
|
330 |
|
Read all tags from database and create in-memory cache for tags |
331 |
|
|
332 |
|
=cut |
333 |
|
|
334 |
|
sub seed_tags { |
335 |
|
my $sth = $dbh->prepare(qq{ select id,message from log where message like '%//%' }); |
336 |
|
$sth->execute; |
337 |
|
while (my $row = $sth->fetchrow_hashref) { |
338 |
|
add_tag( %$row ); |
339 |
|
} |
340 |
|
|
341 |
|
foreach my $tag (keys %$tags) { |
342 |
|
$cloud->add($tag, "?tag=$tag", scalar @{$tags->{$tag}} + 1); |
343 |
|
} |
344 |
|
} |
345 |
|
|
346 |
|
seed_tags; |
347 |
|
|
348 |
|
|
349 |
|
=head2 save_message |
350 |
|
|
351 |
|
save_message( |
352 |
|
channel => '#foobar', |
353 |
|
me => 0, |
354 |
|
nick => 'dpavlin', |
355 |
|
msg => 'test message', |
356 |
|
time => '2006-06-25 18:57:18', |
357 |
|
); |
358 |
|
|
359 |
|
C<time> is optional, it will use C<< now() >> if it's not available. |
360 |
|
|
361 |
|
C<me> if not specified will be C<0> (not C</me> message) |
362 |
|
|
363 |
|
=cut |
364 |
|
|
365 |
|
sub save_message { |
366 |
|
my $a = {@_}; |
367 |
|
$a->{me} ||= 0; |
368 |
|
$a->{time} ||= strftime($TIMESTAMP,localtime()); |
369 |
|
|
370 |
|
print |
371 |
|
$a->{time}, " ", |
372 |
|
$a->{channel}, " ", |
373 |
|
$a->{me} ? "***" . $a->{nick} : "<" . $a->{nick} . ">", |
374 |
|
" " . $a->{msg} . "\n"; |
375 |
|
|
376 |
|
from_to($a->{msg}, 'UTF-8', $ENCODING); |
377 |
|
|
378 |
|
$sth->execute($a->{channel}, $a->{me}, $a->{nick}, $a->{msg}, $a->{time}); |
379 |
|
add_tag( id => $dbh->last_insert_id(undef,undef,"log",undef), |
380 |
|
message => $a->{msg}); |
381 |
|
} |
382 |
|
|
383 |
|
if ($import_dircproxy) { |
384 |
|
open(my $l, $import_dircproxy) || die "can't open $import_dircproxy: $!"; |
385 |
|
warn "importing $import_dircproxy...\n"; |
386 |
|
my $tz_offset = 2 * 60 * 60; # TZ GMT+2 |
387 |
|
while(<$l>) { |
388 |
|
chomp; |
389 |
|
if (/^@(\d+)\s(\S+)\s(.+)$/) { |
390 |
|
my ($time, $nick, $msg) = ($1,$2,$3); |
391 |
|
|
392 |
|
my $dt = DateTime->from_epoch( epoch => $time + $tz_offset ); |
393 |
|
|
394 |
|
my $me = 0; |
395 |
|
$me = 1 if ($nick =~ m/^\[\S+]/); |
396 |
|
$nick =~ s/^[\[<]([^!]+).*$/$1/; |
397 |
|
|
398 |
|
$msg =~ s/^ACTION\s+// if ($me); |
399 |
|
|
400 |
|
save_message( |
401 |
|
channel => $CHANNEL, |
402 |
|
me => $me, |
403 |
|
nick => $nick, |
404 |
|
msg => $msg, |
405 |
|
time => $dt->ymd . " " . $dt->hms, |
406 |
|
) if ($nick !~ m/^-/); |
407 |
|
|
408 |
|
} else { |
409 |
|
warn "can't parse: $_\n"; |
410 |
|
} |
411 |
|
} |
412 |
|
close($l); |
413 |
|
warn "import over\n"; |
414 |
|
exit; |
415 |
|
} |
416 |
|
|
417 |
|
|
418 |
|
# |
419 |
|
# POE handing part |
420 |
|
# |
421 |
|
|
422 |
my $SKIPPING = 0; # if skipping, how many we've done |
my $SKIPPING = 0; # if skipping, how many we've done |
423 |
my $SEND_QUEUE; # cache |
my $SEND_QUEUE; # cache |
442 |
my $channel = $_[ARG1]->[0]; |
my $channel = $_[ARG1]->[0]; |
443 |
my $msg = $_[ARG2]; |
my $msg = $_[ARG2]; |
444 |
|
|
445 |
from_to($msg, 'UTF-8', $ENCODING); |
save_message( channel => $channel, me => 0, nick => $nick, msg => $msg); |
|
|
|
|
print "$channel: <$nick> $msg\n"; |
|
|
$sth->execute($channel, 0, $nick, $msg); |
|
|
add_tag( id => $dbh->last_insert_id(undef,undef,"log",undef), |
|
|
message => $msg); |
|
446 |
}, |
}, |
447 |
irc_ctcp_action => sub { |
irc_ctcp_action => sub { |
448 |
my $kernel = $_[KERNEL]; |
my $kernel = $_[KERNEL]; |
450 |
my $channel = $_[ARG1]->[0]; |
my $channel = $_[ARG1]->[0]; |
451 |
my $msg = $_[ARG2]; |
my $msg = $_[ARG2]; |
452 |
|
|
453 |
from_to($msg, 'UTF-8', $ENCODING); |
save_message( channel => $channel, me => 1, nick => $nick, msg => $msg); |
|
|
|
|
print "$channel ***$nick $msg\n"; |
|
|
$sth->execute($channel, 1, $nick, $msg); |
|
|
add_tag( id => $dbh->last_insert_id(undef,undef,"log",undef), |
|
|
message => $msg); |
|
454 |
}, |
}, |
455 |
irc_msg => sub { |
irc_msg => sub { |
456 |
my $kernel = $_[KERNEL]; |
my $kernel = $_[KERNEL]; |
478 |
my $nr = $1 || 10; |
my $nr = $1 || 10; |
479 |
|
|
480 |
my $sth = $dbh->prepare(qq{ |
my $sth = $dbh->prepare(qq{ |
481 |
select nick,count(*) from log group by nick order by count desc limit $nr |
select |
482 |
|
nick, |
483 |
|
count(*) as count, |
484 |
|
sum(length(message)) as len |
485 |
|
from log |
486 |
|
group by nick |
487 |
|
order by len desc,count desc |
488 |
|
limit $nr |
489 |
}); |
}); |
490 |
$sth->execute(); |
$sth->execute(); |
491 |
$res = "Top $nr users: "; |
$res = "Top $nr users: "; |
492 |
my @users; |
my @users; |
493 |
while (my $row = $sth->fetchrow_hashref) { |
while (my $row = $sth->fetchrow_hashref) { |
494 |
push @users,$row->{nick} . ': ' . $row->{count}; |
push @users,$row->{nick} . ': ' . $row->{count} . '/' . $row->{len} . '=' . sprintf("%.2f", $row->{len}/$row->{count}); |
495 |
} |
} |
496 |
$res .= join(" | ", @users); |
$res .= join(" | ", @users); |
497 |
} elsif ($msg =~ m/^last.*?\s*(\d*)/i) { |
} elsif ($msg =~ m/^last.*?\s*(\d*)/i) { |
498 |
|
|
499 |
foreach my $res (get_from_log( limit => $1 )) { |
foreach my $res (get_from_log( limit => ($1 || 100) )) { |
500 |
print "last: $res\n"; |
print "last: $res\n"; |
501 |
from_to($res, $ENCODING, 'UTF-8'); |
from_to($res, $ENCODING, 'UTF-8'); |
502 |
$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res ); |
$_[KERNEL]->post( $IRC_ALIAS => privmsg => $nick, $res ); |
542 |
warn "## indetify $NICK\n"; |
warn "## indetify $NICK\n"; |
543 |
$_[KERNEL]->post( $IRC_ALIAS => privmsg => 'nickserv', "IDENTIFY $NICK" ); |
$_[KERNEL]->post( $IRC_ALIAS => privmsg => 'nickserv', "IDENTIFY $NICK" ); |
544 |
}, |
}, |
545 |
|
irc_disconnected => sub { |
546 |
|
warn "## disconnected, reconnecting again\n"; |
547 |
|
$_[KERNEL]->post($IRC_ALIAS => connect => $CONNECT); |
548 |
|
}, |
549 |
|
irc_socketerr => sub { |
550 |
|
warn "## socket error... sleeping for $sleep_on_error seconds and retry"; |
551 |
|
sleep($sleep_on_error); |
552 |
|
$_[KERNEL]->post($IRC_ALIAS => connect => $CONNECT); |
553 |
|
}, |
554 |
# irc_433 => sub { |
# irc_433 => sub { |
555 |
# print "# irc_433: ",$_[ARG1], "\n"; |
# print "# irc_433: ",$_[ARG1], "\n"; |
556 |
# warn "## indetify $NICK\n"; |
# warn "## indetify $NICK\n"; |
628 |
}, |
}, |
629 |
); |
); |
630 |
|
|
|
# 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; |
|
|
|
|
631 |
# http server |
# http server |
632 |
|
|
633 |
my $httpd = POE::Component::Server::HTTP->new( |
my $httpd = POE::Component::Server::HTTP->new( |
653 |
.col-4 { background: #ff66ff } |
.col-4 { background: #ff66ff } |
654 |
a:link.tag, a:visited.tag { border: 1px dashed #ccc; backgound: #ccc; text-decoration: none } |
a:link.tag, a:visited.tag { border: 1px dashed #ccc; backgound: #ccc; text-decoration: none } |
655 |
a:hover.tag { border: 1px solid #eee } |
a:hover.tag { border: 1px solid #eee } |
656 |
|
hr { border: 1px dashed #ccc; height: 1px; clear: both; } |
657 |
_END_OF_STYLE_ |
_END_OF_STYLE_ |
658 |
|
|
659 |
my $max_color = 4; |
my $max_color = 4; |
677 |
|
|
678 |
my $search = $q->param('search') || $q->param('grep') || ''; |
my $search = $q->param('search') || $q->param('grep') || ''; |
679 |
|
|
680 |
$response->content( |
my $html = |
681 |
qq{<html><head><title>$NICK</title><style type="text/css">$style} . |
qq{<html><head><title>$NICK</title><style type="text/css">$style} . |
682 |
$cloud->css . |
$cloud->css . |
683 |
qq{</style></head><body>} . |
qq{</style></head><body>} . |
688 |
</form> |
</form> |
689 |
} . |
} . |
690 |
$cloud->html(500) . |
$cloud->html(500) . |
691 |
qq{<p>} . |
qq{<p>}; |
692 |
join("</p><p>", |
if ($request->url =~ m#/history#) { |
693 |
|
my $sth = $dbh->prepare(qq{ |
694 |
|
select date(time) as date,count(*) as nr |
695 |
|
from log |
696 |
|
group by date(time) |
697 |
|
order by date(time) desc |
698 |
|
}); |
699 |
|
$sth->execute(); |
700 |
|
my ($l_yyyy,$l_mm) = (0,0); |
701 |
|
my $cal; |
702 |
|
while (my $row = $sth->fetchrow_hashref) { |
703 |
|
# this is probably PostgreSQL specific, expects ISO date |
704 |
|
my ($yyyy,$mm,$dd) = split(/-/, $row->{date}); |
705 |
|
if ($yyyy != $l_yyyy || $mm != $l_mm) { |
706 |
|
$html .= $cal->as_HTML() if ($cal); |
707 |
|
$cal = new HTML::CalendarMonthSimple('month'=>$mm,'year'=>$yyyy); |
708 |
|
$cal->border(2); |
709 |
|
($l_yyyy,$l_mm) = ($yyyy,$mm); |
710 |
|
} |
711 |
|
$cal->setcontent($dd, qq{ |
712 |
|
<a href="/?date=$row->{date}">$row->{nr}</a> |
713 |
|
}); |
714 |
|
} |
715 |
|
$html .= $cal->as_HTML() if ($cal); |
716 |
|
|
717 |
|
} else { |
718 |
|
$html .= join("</p><p>", |
719 |
get_from_log( |
get_from_log( |
720 |
limit => $q->param('last') || 100, |
limit => $q->param('last') || $q->param('date') ? undef : 100, |
721 |
search => $search || undef, |
search => $search || undef, |
722 |
tag => $q->param('tag') || undef, |
tag => $q->param('tag') || undef, |
723 |
|
date => $q->param('date') || undef, |
724 |
fmt => { |
fmt => { |
725 |
date => '<hr size="1" style="clear: both;"/><div class="date">%s</div> ', |
date => sub { |
726 |
|
my $date = shift || return; |
727 |
|
qq{<hr/><div class="date"><a href="/?date=$date">$date</a></div>}; |
728 |
|
}, |
729 |
time => '<span class="time">%s</span> ', |
time => '<span class="time">%s</span> ', |
730 |
time_channel => '<span class="channel">%s %s</span> ', |
time_channel => '<span class="channel">%s %s</span> ', |
731 |
nick => '%s: ', |
nick => '%s: ', |
752 |
}, |
}, |
753 |
}, |
}, |
754 |
) |
) |
755 |
) . |
); |
756 |
qq{</p></body></html>} |
} |
757 |
); |
|
758 |
|
$html .= qq{</p> |
759 |
|
<hr/> |
760 |
|
<p>See <a href="/history">history</a> of all messages.</p> |
761 |
|
</body></html>}; |
762 |
|
|
763 |
|
$response->content( $html ); |
764 |
return RC_OK; |
return RC_OK; |
765 |
} |
} |
766 |
|
|