いい加減しつこいですが、もうちょっとだけ改善しました。
今回の変更点は、以下です。
- コメントのイメージの扱いが間違っていたので修正
- コミュニティのアンケート(のコメント)に対応(アンケートに答えていない場合の挙動は未確認)
- コミュニティのトピックのコメントに、コミュニティ名を追加
しかし、アンケートのページには何故か日付が表示されないんですね。これだと、module: Freshがうまく機能しないのではないかしらん。
今回から、パッチを添付します。
Mixi.patch-20070108.tar.gz
WWW::Mixiのパッチ
*** /usr/local/lib/perl5/site_perl/5.8.8/WWW/Mixi.pm.ORIG Sat Jan 6 21:54:02 2007
--- /usr/local/lib/perl5/site_perl/5.8.8/WWW/Mixi.pm Mon Jan 8 15:56:01 2007
***************
*** 1422,1451 ****
my $base = $res->base->as_string;
my $content = $res->content;
my @items = ();
my $re_date = '<td rowspan="3" width="110" bgcolor="#ffd8b0" align="center" valign="top" nowrap>(\d{4})年(\d{2})月(\d{2})日<br>(\d{1,2}):(\d{2})</td>';
my $re_subj = '<td bgcolor="#fff4e0"> (.+?)</td>';
my $re_desc = '</table>(.+?)</td>';
my $re_c_date = '<td rowspan="2" width="110" bgcolor="#f2ddb7" align="center" nowrap>\n(\d{4})年(\d{2})月(\d{2})日<br>\n(\d{1,2}):(\d{2})';
my $re_c_desc = '<td class="h120">(.+?)\n</td>';
my $re_link = '<a href="?(.+?)"?>(.*?)<\/a>';
! if ($content =~ s/<!-- TOPIC: start -->.*?${re_date}.*?${re_subj}.*?${re_link}(.*?)${re_desc}(.*?)$//is) {
! my ($time, $subj, $link, $name, $imgs, $desc, $comm) = (sprintf('%04d/%02d/%02d %02d:%02d', $1,$2,$3,$4,$5), $6, $7, $8, $9, $10, $11);
($desc, $subj) = map { s/[\r\n]+//g; s/<br>/\n/g; $_ = $self->rewrite($_); } ($desc, $subj);
! my $item = { 'time' => $time, 'description' => $desc, 'subject' => $subj, 'link' => $res->request->uri->as_string, 'images' => [], 'comments' => [] , 'name' => $name, 'name_link' => $self->absolute_url($link, $base)};
foreach my $image ($imgs =~ /<td width=130[^<>]*>(.*?)<\/td>/g) {
next unless ($image =~ /<a [^<>]*'show_picture.pl\?img_src=(.*?)'[^<>]*><img src=([^ ]*) border=0>/);
push(@{$item->{'images'}}, {'link' => $self->absolute_url($1, $base), 'thumb_link' => $self->absolute_url($2, $base)});
}
! while ($comm =~ s/.*?${re_c_date}.*?${re_link}.*?${re_c_desc}.*?<\/table>//is){
! my ($time, $link, $name, $desc) = (sprintf('%04d/%02d/%02d %02d:%02d', $1,$2,$3,$4,$5), $6, $7, $8);
! ($name, $desc) = map { s/[\r\n]+//g; s/<br>/\n/g; $_ = $self->rewrite($_); } ($name, $desc);
! push(@{$item->{'comments'}}, {'time' => $time, 'link' => $self->absolute_url($link, $base), 'name' => $name, 'description' => $desc});
}
push(@items, $item);
}
return @items;
}
sub parse_view_diary {
my $self = shift;
my $res = (@_) ? shift : $self->response();
--- 1422,1491 ----
my $base = $res->base->as_string;
my $content = $res->content;
my @items = ();
+ my $re_comm = '<b>(\[.*?\]) トピック</b>';
my $re_date = '<td rowspan="3" width="110" bgcolor="#ffd8b0" align="center" valign="top" nowrap>(\d{4})年(\d{2})月(\d{2})日<br>(\d{1,2}):(\d{2})</td>';
my $re_subj = '<td bgcolor="#fff4e0"> (.+?)</td>';
my $re_desc = '</table>(.+?)</td>';
my $re_c_date = '<td rowspan="2" width="110" bgcolor="#f2ddb7" align="center" nowrap>\n(\d{4})年(\d{2})月(\d{2})日<br>\n(\d{1,2}):(\d{2})';
+ my $re_c_no = '<b>(.*?[0-9]+?)</b>';
my $re_c_desc = '<td class="h120">(.+?)\n</td>';
my $re_link = '<a href="?(.+?)"?>(.*?)<\/a>';
! if ($content =~ s/<!-- TOPIC: start -->.*?${re_comm}.*?${re_date}.*?${re_subj}.*?${re_link}(.*?)${re_desc}(.*?)$//is) {
! my ($community, $time, $subj, $link, $name, $imgs, $desc, $comm) = ($1, sprintf('%04d/%02d/%02d %02d:%02d', $2,$3,$4,$5,$6), $7, $8, $9, $10, $11, $12);
($desc, $subj) = map { s/[\r\n]+//g; s/<br>/\n/g; $_ = $self->rewrite($_); } ($desc, $subj);
! my $item = { 'time' => $time, 'description' => $desc, 'subject' => $subj, 'link' => $res->request->uri->as_string, 'images' => [], 'comments' => [] , 'name' => $name, 'name_link' => $self->absolute_url($link, $base), 'community_name' => $community};
foreach my $image ($imgs =~ /<td width=130[^<>]*>(.*?)<\/td>/g) {
next unless ($image =~ /<a [^<>]*'show_picture.pl\?img_src=(.*?)'[^<>]*><img src=([^ ]*) border=0>/);
push(@{$item->{'images'}}, {'link' => $self->absolute_url($1, $base), 'thumb_link' => $self->absolute_url($2, $base)});
}
! while ($comm =~ s/.*?${re_c_date}.*?${re_c_no}.*?${re_link}.*?${re_c_desc}.*?<\/table>//is){
! my ($time, $number, $link, $name, $desc) = (sprintf('%04d/%02d/%02d %02d:%02d', $1,$2,$3,$4,$5), $6, $7, $8, $9);
! # ($name, $desc) = map { s/[\r\n]+//g; s/<br>/\n/g; $_ = $self->rewrite($_); } ($name, $desc);
! my $comment = {'time' => $time, 'link' => $self->absolute_url($link, $base), 'name' => $name, 'description' => $desc, 'number' => $number, 'images' => []};
! push(@{$item->{'comments'}}, $comment);
! foreach my $image ($desc =~ /<td width="?130"?[^<>]*>(.*?)<\/td>/g) {
! next unless ($image =~ /<img src=([^ ]*) border="?0"?>/);
! push(@{$comment->{'images'}}, {'link' => $self->absolute_url($1, $base), 'thumb_link' => $self->absolute_url($1, $base)});
! }
}
push(@items, $item);
}
return @items;
}
+ sub parse_view_enquete {
+ my $self = shift;
+ my $res = (@_) ? shift : $self->response();
+ return unless ($res and $res->is_success);
+ my $base = $res->base->as_string;
+ my $content = $res->content;
+ my @items = ();
+ my $re_comm = '<b>(\[.*?\]).*?</b>';
+ my $re_subj = '<td bgcolor=#ffffff width=530>(.+?)</td>';
+ my $re_desc = '</table>(.+?)</td>';
+ my $re_c_date = '<td rowspan="2" width="110" bgcolor="#f2ddb7" align="center" nowrap>\n(\d{4})年(\d{2})月(\d{2})日<br>(\d{1,2}):(\d{2})';
+ my $re_c_no = '<b>(.*?[0-9]+?)</b>';
+ my $re_c_desc = '<td class="h120">(.+?)</td>';
+ my $re_link = '<a href="?(.+?)"?>(.*?)</a>';
+ if ($content =~ s/<td align=center colspan=2>.*?${re_comm}.*?${re_subj}.*?${re_desc}.*?${re_link}(.*?)$//is) {
+ my ($community, $subj, $desc, $link, $name, $comm) = ($1, $2, $3, $4, $5, $6);
+ # ($desc, $subj) = map { s/[\r\n]+//g; s/<br>/\n/g; $_ = $self->rewrite($_); } ($desc, $subj);
+ my $item = { 'description' => $desc, 'subject' => $subj, 'link' => $res->request->uri->as_string, 'comments' => [] , 'name' => $name, 'name_link' => $self->absolute_url($link, $base), 'community_name' => $community};
+ while ($comm =~ s/.*?${re_c_date}.*?${re_c_no}.*?${re_link}.*?${re_c_desc}.*?<\/table>//is){
+ my ($time, $number, $link, $name, $desc) = (sprintf('%04d/%02d/%02d %02d:%02d', $1,$2,$3,$4,$5), $6, $7, $8, $9);
+ # ($name, $desc) = map { s/[\r\n]+//g; s/<br>/\n/g; $_ = $self->rewrite($_); } ($name, $desc);
+ my $comment = {'time' => $time, 'link' => $self->absolute_url($link, $base), 'name' => $name, 'description' => $desc, 'number' => $number, 'images' => []};
+ push(@{$item->{'comments'}}, $comment);
+ foreach my $image ($desc =~ /<td width="?130"?[^<>]*>(.*?)<\/td>/g) {
+ next unless ($image =~ /<img src=([^ ]*) border="?0"?>/);
+ push(@{$comment->{'images'}}, {'link' => $self->absolute_url($1, $base), 'thumb_link' => $self->absolute_url($1, $base)});
+ }
+ }
+ push(@items, $item);
+ }
+ return @items;
+ }
+
sub parse_view_diary {
my $self = shift;
my $res = (@_) ? shift : $self->response();
***************
*** 1473,1479 ****
push(@{$item->{'comments'}}, {'time' => $time, 'link' => $self->absolute_url($link, $base), 'name' => $name, 'description' => $desc});
}
push(@items, $item);
! }
return @items;
}
--- 1513,1519 ----
push(@{$item->{'comments'}}, {'time' => $time, 'link' => $self->absolute_url($link, $base), 'name' => $name, 'description' => $desc});
}
push(@items, $item);
! }
return @items;
}
***************
*** 1582,1588 ****
my $label_time = "(?:\Q日 付\E|\Q日 付\E)";
my $label_name = "(?:\Q差出人\E|\Q宛 先\E)";
my $label_subj = "(?:\Q件 名\E|\Q件 名\E)";
! my $time = sprintf('%04d/%02d/%02d %02d:%02d', $1, $2, $3, $4, $5) if ($content =~ /<$td>$s<font(?:$attr)*>$label_time<\/font>$s:$s(\d{4})年(\d{2})月(\d{2})日$s(\d{2}):(\d{2})<\/td>/is);
my $subj = $self->rewrite($1) if ($content =~ /<$td>$s<font(?:$attr)*>$label_subj<\/font>$s:$s($str)<\/td>/is);
my $desc = $self->rewrite($1) if ($content =~ /<td(?:$attr)*CLASS=h120(?:$attr)*>$s($str)<\/td>/is);
my $image = $self->absolute_url($1, $base) if ($content =~ /<$td><a(?:$attr)*><img(?:$attr)*src=["']?([^"'\s<>]+)["'](?:$attr)*><\/a><\/td>/is);
--- 1622,1628 ----
my $label_time = "(?:\Q日 付\E|\Q日 付\E)";
my $label_name = "(?:\Q差出人\E|\Q宛 先\E)";
my $label_subj = "(?:\Q件 名\E|\Q件 名\E)";
! my $time = sprintf('%04d/%02d/%02d %02d:%02d', $1, $2, $3, $4, $5) if ($content =~ /<$td>$s<font(?:$attr)*>$label_time<\/font>$s:$s(\d{4})年(\d{2})月(\d{2})日$s(\d{2})時(\d{2})分$s<\/td>/is);
my $subj = $self->rewrite($1) if ($content =~ /<$td>$s<font(?:$attr)*>$label_subj<\/font>$s:$s($str)<\/td>/is);
my $desc = $self->rewrite($1) if ($content =~ /<td(?:$attr)*CLASS=h120(?:$attr)*>$s($str)<\/td>/is);
my $image = $self->absolute_url($1, $base) if ($content =~ /<$td><a(?:$attr)*><img(?:$attr)*src=["']?([^"'\s<>]+)["'](?:$attr)*><\/a><\/td>/is);
***************
*** 2246,2251 ****
--- 2286,2298 ----
my $url = shift or return;
$self->set_response($url, @_) or return undef;
return $self->parse_view_bbs();
+ }
+
+ sub get_view_enquete {
+ my $self = shift;
+ my $url = shift or return;
+ $self->set_response($url, @_) or return undef;
+ return $self->parse_view_enquete();
}
sub get_view_community {
CustomFeed::Mixiのパッチ
*** /usr/local/lib/perl5/site_perl/5.8.8/Plagger/Plugin/CustomFeed/Mixi.pm.ORIG Fri Jan 5 23:14:14 2007
--- /usr/local/lib/perl5/site_perl/5.8.8/Plagger/Plugin/CustomFeed/Mixi.pm Mon Jan 8 13:13:19 2007
***************
*** 48,53 ****
--- 48,62 ----
get_list => 'parse_show_calendar',
get_detail => 'get_view_event',
},
+ Bbs => {
+ start_url => 'http://mixi.jp/new_bbs.pl',
+ title => 'コミュニティ最新書き込み',
+ get_list => 'parse_new_bbs',
+ get_detail => 'dummy',
+ get_detail_bbs => 'get_view_bbs',
+ get_detail_event => 'get_view_event',
+ get_detail_enquete => 'get_view_enquete',
+ },
};
sub plugin_id {
***************
*** 134,148 ****
my $i = 0;
my $blocked = 0;
for my $msg (@msgs) {
next if $type eq 'FriendDiary' and not $msg->{image}; # external blog
last if $i++ >= $items;
my $entry = Plagger::Entry->new;
! $entry->title( decode('euc-jp', $msg->{subject}) );
$entry->link($msg->{link});
$entry->author( decode('euc-jp', $msg->{name}) );
$entry->date( Plagger::Date->parse($format, $msg->{time}) );
if ($self->conf->{show_icon} && !$blocked && defined $MAP->{$type}->{icon_re}) {
my $owner_id = ($msg->{link} =~ $MAP->{$type}->{icon_re})[0];
--- 143,165 ----
my $i = 0;
my $blocked = 0;
+ my $item;
for my $msg (@msgs) {
next if $type eq 'FriendDiary' and not $msg->{image}; # external blog
last if $i++ >= $items;
my $entry = Plagger::Entry->new;
! my $title = "";
! if ($msg->{community_name}) {
! $title = decode('euc-jp', $msg->{subject}) . " ";
! }
! $entry->title( $title . decode('euc-jp', $msg->{subject}) );
$entry->link($msg->{link});
$entry->author( decode('euc-jp', $msg->{name}) );
$entry->date( Plagger::Date->parse($format, $msg->{time}) );
+ if ($entry->date) {
+ $entry->date->set_time_zone('Asia/Tokyo');
+ }
if ($self->conf->{show_icon} && !$blocked && defined $MAP->{$type}->{icon_re}) {
my $owner_id = ($msg->{link} =~ $MAP->{$type}->{icon_re})[0];
***************
*** 170,182 ****
}
}
if ($self->conf->{fetch_body} && !$blocked && $msg->{link} =~ /view_/ && defined $MAP->{$type}->{get_detail}) {
$context->log(info => "Fetch body from $msg->{link}");
! my $item = $self->cache->get_callback(
"item-$msg->{link}",
sub {
Time::HiRes::sleep( $self->conf->{fetch_body_interval} || 1.5 );
! my $meth = $MAP->{$type}->{get_detail};
my($item) = $self->{mixi}->$meth($msg->{link});
if ($meth eq 'get_view_diary') {
--- 187,203 ----
}
}
+ $item = 0;
+
if ($self->conf->{fetch_body} && !$blocked && $msg->{link} =~ /view_/ && defined $MAP->{$type}->{get_detail}) {
$context->log(info => "Fetch body from $msg->{link}");
! $item = $self->cache->get_callback(
"item-$msg->{link}",
sub {
Time::HiRes::sleep( $self->conf->{fetch_body_interval} || 1.5 );
! # my $meth = $MAP->{$type}->{get_detail};
! my $meth = get_meth($MAP->{$type}, $msg->{link});
!
my($item) = $self->{mixi}->$meth($msg->{link});
if ($meth eq 'get_view_diary') {
***************
*** 189,195 ****
if ($item) {
my $body = decode('euc-jp', $item->{description});
$body =~ s!(\r\n?|\n)!<br />!g;
! for my $image (@{ $item->{images} }) {
$body .= qq(<div><a href="$image->{link}"><img src="$image->{thumb_link}" style="border:0" /></a></div>);
my $enclosure = Plagger::Enclosure->new;
$enclosure->url( URI->new($image->{thumb_link}) );
--- 210,217 ----
if ($item) {
my $body = decode('euc-jp', $item->{description});
$body =~ s!(\r\n?|\n)!<br />!g;
! my $image;
! for $image (@{ $item->{images} }) {
$body .= qq(<div><a href="$image->{link}"><img src="$image->{thumb_link}" style="border:0" /></a></div>);
my $enclosure = Plagger::Enclosure->new;
$enclosure->url( URI->new($image->{thumb_link}) );
***************
*** 197,205 ****
--- 219,238 ----
$enclosure->is_inline(1);
$entry->add_enclosure($enclosure);
}
+ if ($image = $item->{image}) {
+ $body .= qq(<div><a href="$image"><img src="$image" style="border:0" /></a></div>);
+ my $enclosure = Plagger::Enclosure->new;
+ $enclosure->url( URI->new($image) );
+ $enclosure->auto_set_type;
+ $enclosure->is_inline(1);
+ $entry->add_enclosure($enclosure);
+ }
$entry->body($body);
$entry->date( Plagger::Date->parse($format, $item->{time}) );
+ if ($entry->date) {
+ $entry->date->set_time_zone('Asia/Tokyo');
+ }
} else {
$context->log(warn => "Fetch body failed. You might be blocked?");
$blocked++;
***************
*** 207,212 ****
--- 240,248 ----
}
$feed->add_entry($entry);
+ if ($self->conf->{fetch_body} && $item) {
+ add_comments($feed, $item);
+ }
}
$context->update->add($feed);
***************
*** 221,226 ****
--- 257,305 ----
}
return \@images;
+ }
+ sub get_meth {
+ my($type, $link) = @_;
+
+ my $meth = "";
+ if ($link =~ /view_bbs/) {
+ $meth = $type->{get_detail_bbs};
+ } elsif ($link =~ /view_event/) {
+ $meth = $type->{get_detail_event};
+ } elsif ($link =~ /view_enquete/) {
+ $meth = $type->{get_detail_enquete};
+ } else {
+ $meth = $type->{get_detail};
+ }
+ return $meth;
+ }
+
+ sub add_comments {
+ my ($feed, $item) = @_;
+ my $format = DateTime::Format::Strptime->new(pattern => '%Y/%m/%d %H:%M');
+ if ($item->{'comments'}) {
+ for my $comment (@{$item->{'comments'}}) {
+ my $entry = Plagger::Entry->new;
+ $entry->title(decode('euc-jp', $item->{'community_name'}) . " " . decode('euc-jp', $item->{'subject'}) . ": " . $comment->{'number'});
+ $entry->link($item->{'link'});
+ $entry->author( decode('euc-jp', $comment->{'name'}) );
+ $entry->date( Plagger::Date->parse($format, $comment->{'time'}) );
+ if ($entry->date) {
+ $entry->date->set_time_zone('Asia/Tokyo');
+ }
+ my $body = decode('euc-jp', $comment->{'description'});
+ for my $image (@{ $comment->{images} }) {
+ $body .= qq(<div><a href="$image->{link}"><img src="$image->{thumb_link}" style="border:0" /></a></div>);
+ my $enclosure = Plagger::Enclosure->new;
+ $enclosure->url( URI->new($image->{thumb_link}) );
+ $enclosure->auto_set_type;
+ $enclosure->is_inline(1);
+ $entry->add_enclosure($enclosure);
+ }
+ $entry->body( $body );
+ $feed->add_entry($entry);
+ }
+ }
}
1;