root/branches/feature-server/plagger/lib/Plagger/Plugin/Publish/CHTML.pm

Revision 1009 (checked in by miyagawa, 3 years ago)

r2680@rock (orig r938): miyagawa | 2006-06-09 10:28:17 +0900
add podtrac TruePermalink?. via http://d.hatena.ne.jp/mryfmo/20060608
r2681@rock (orig r939): miyagawa | 2006-06-09 10:30:48 +0900
TruePermalink?: add feedburner podcast redirector. Refs #226
r2682@rock (orig r940): miyagawa | 2006-06-09 16:11:35 +0900
use Last-Modified header to populate entry date, even if handler can't find one.
via http://subtech.g.hatena.ne.jp/otsune/20060608/norkdailymemo
r2683@rock (orig r941): miyagawa | 2006-06-09 16:12:52 +0900
take off utf-8 flag when taking digest value
r2684@rock (orig r942): miyagawa | 2006-06-09 17:04:38 +0900

Publish
CHTML: Don't die if body contains non-sjis mappable characters
r2685@rock (orig r943): miyagawa | 2006-06-09 17:26:01 +0900
defaults to cp932 would be better
r2686@rock (orig r944): miyagawa | 2006-06-09 17:37:37 +0900

r2687@rock (orig r945): miyagawa | 2006-06-09 18:48:15 +0900
add pya.cc upgrader via http://subtech.g.hatena.ne.jp/otsune/20060608/pya2feed
r2688@rock (orig r946): miyagawa | 2006-06-09 21:21:47 +0900
CustomFeed?
2chSearch
r2689@rock (orig r947): miyagawa | 2006-06-09 21:26:31 +0900
oops, remove </b>
r2690@rock (orig r948): miyagawa | 2006-06-09 21:44:42 +0900
fix date if it found true entry
r2691@rock (orig r949): miyagawa | 2006-06-09 21:59:05 +0900
need quotes
r2692@rock (orig r950): miyagawa | 2006-06-09 22:06:35 +0900
Planet: Scrubber support back inlib/Plagger/Plugin/Publish/Planet.pm
r2693@rock (orig r951): miyagawa | 2006-06-09 22:08:01 +0900
oops
r2694@rock (orig r952): otsune | 2006-06-09 22:11:04 +0900
fix extract http://pyc.cc/

r2695@rock (orig r953): otsune | 2006-06-09 22:12:28 +0900
add EntryFulltext? for seesaa blog

r2696@rock (orig r954): otsune | 2006-06-09 23:27:11 +0900
fix %3A

r2697@rock (orig r955): miyagawa | 2006-06-10 02:26:28 +0900
MixiDiarySearch?: decode keyword query
r2698@rock (orig r956): miyagawa | 2006-06-10 02:53:41 +0900
TruePermalink? enbug stuff. Use permlalink to find handlers
r2699@rock (orig r957): otsune | 2006-06-10 03:08:33 +0900
add EntryFulltext? http://headlines.yahoo.co.jp/

r2700@rock (orig r958): otsune | 2006-06-10 04:38:27 +0900
add Apple KB and TIL document

r2701@rock (orig r959): otsune | 2006-06-10 04:43:22 +0900
oops.

r2702@rock (orig r960): miyagawa | 2006-06-10 23:07:48 +0900
set Bloglines n=100
r2703@rock (orig r961): miyagawa | 2006-06-11 01:35:38 +0900
MixiDiarySearch?: allow no_photo.gif
r2704@rock (orig r962): miyagawa | 2006-06-11 01:45:53 +0900
2chSearh: Fix error handling
r2705@rock (orig r963): miyagawa | 2006-06-11 02:07:11 +0900
added takesako-san for his patch
r2706@rock (orig r964): otsune | 2006-06-11 05:59:58 +0900
modified Chugoku SHinbun, add EFT for http://www.zianplus.net/

r2707@rock (orig r965): otsune | 2006-06-11 10:17:02 +0900
add pMachine ExpressionEngine? http://www.pmachine.com/

r2708@rock (orig r966): youpy | 2006-06-11 12:38:21 +0900
fix regexp

r2709@rock (orig r967): otsune | 2006-06-12 04:09:24 +0900
fix extract regexp

r2710@rock (orig r968): otsune | 2006-06-12 04:13:19 +0900
update regexp

r2711@rock (orig r969): otsune | 2006-06-12 04:29:18 +0900
support http://www.mainichi-msn.co.jp/photo/etc/photo_feature/

r2712@rock (orig r970): otsune | 2006-06-12 06:08:15 +0900
fix wordpress.
Add mainichi-msn Photo and separate handle.
Add http://www.actiblog.com/

r2713@rock (orig r971): otsune | 2006-06-12 07:02:23 +0900
refine livedoorblog.pl
fix miss.

r2714@rock (orig r972): miyagawa | 2006-06-12 13:25:28 +0900
extract_title should be case insensitive. via http://d.hatena.ne.jp/sfujiwara/20060611/1150051152
r2715@rock (orig r973): miyagawa | 2006-06-12 13:39:12 +0900
rewrite config doesn't die even if it can't rewrite because of permission problem
r2716@rock (orig r974): miyagawa | 2006-06-12 13:43:25 +0900
skip all livedoorkeyword link
r2719@rock (orig r975): otsune | 2006-06-12 14:50:19 +0900
fix misc regexp

r2720@rock (orig r976): miyagawa | 2006-06-12 15:44:57 +0900
support handle only in livedoorblog.pl to work with aggregated feeds
r2721@rock (orig r977): miyagawa | 2006-06-12 18:22:40 +0900
TruePermalink? for blogpeople redirector
r2722@rock (orig r978): otsune | 2006-06-12 22:14:03 +0900
opps 'Unmatched ( in regex;'

r2723@rock (orig r979): youpy | 2006-06-13 10:21:42 +0900
add mailman upgrader


r2724@rock (orig r980): youpy | 2006-06-13 10:28:19 +0900
fix handle regexp


r2727@rock (orig r983): miyagawa | 2006-06-13 19:00:22 +0900
Subscription
Planet: add feedster.jp
r2728@rock (orig r984): miyagawa | 2006-06-13 19:06:06 +0900
use lang/all on feedster.jp
r2734@rock (orig r985): otsune | 2006-06-13 22:11:21 +0900
fix regexp

r2735@rock (orig r986): miyagawa | 2006-06-14 00:34:01 +0900
new plugin Notify
Beep
r2736@rock (orig r987): miyagawa | 2006-06-14 00:34:40 +0900
planet: remove unnecessary bit
r2737@rock (orig r988): miyagawa | 2006-06-14 00:35:03 +0900
update example to use sixapart-std
r2738@rock (orig r989): otsune | 2006-06-14 02:55:47 +0900
remove icon_re. RecentComment? can't get it

r2745@rock (orig r990): miyagawa | 2006-06-14 12:07:29 +0900
t/core is for developer test and not needed for installers
r2746@rock (orig r991): miyagawa | 2006-06-14 12:49:00 +0900
support mixi_tos_paranoia mode
r2747@rock (orig r992): miyagawa | 2006-06-14 13:10:40 +0900
title would be ok
r2792@rock (orig r993): miyagawa | 2006-06-16 15:04:12 +0900
  • New plugin Subscription::Bookmarks (and its IE subclass) to read IE favorites.
r2793@rock (orig r994): miyagawa | 2006-06-16 15:11:52 +0900
added TODO as comment
r2794@rock (orig r995): youpy | 2006-06-17 20:36:18 +0900
add Plugin::Subscription::Bookmarks
Safari


r2795@rock (orig r996): youpy | 2006-06-17 21:39:18 +0900
add tag support by folder name


r2796@rock (orig r997): youpy | 2006-06-18 15:41:59 +0900
use $uri->file when scheme is 'file'


r2797@rock (orig r998): youpy | 2006-06-18 15:42:56 +0900
add Plugin::Subscription::Bookmarks
Mozilla


r2798@rock (orig r999): miyagawa | 2006-06-19 15:23:13 +0900
bump URI
Fetch req
r2800@rock (orig r1000): miyagawa | 2006-06-22 00:26:46 +0900
dependency for Bookmarks
Safari. 1000th commit!
r2801@rock (orig r1001): miyagawa | 2006-06-22 00:30:57 +0900
fix config rewriting bug when the password contains regexp metachars. via http://d.hatena.ne.jp/sfujiwara/20060621/1150899012
r2802@rock (orig r1002): otsune | 2006-06-22 00:54:24 +0900
add http://www.computerworld.jp/ http://autopage.teacup.com/
fix headlines_yahoo_jp (Thanks woremacx)
fix goo blog

r2803@rock (orig r1003): miyagawa | 2006-06-22 01:10:00 +0900
import drawnboy's EntryFullText? yamls via http://svn.nowherenear.net/repos/public/misc/eft/
r2804@rock (orig r1004): miyagawa | 2006-06-22 01:10:39 +0900
update AUTHOR
r2805@rock (orig r1005): s_nobu | 2006-06-22 06:17:15 +0900
require HTML
Entities for enclosure support.

r2807@rock (orig r1006): miyagawa | 2006-06-22 15:46:30 +0900
URI
Fetch 0.07 is broken (i was a moron), reverting back to 0.06 for now
r2808@rock (orig r1007): miyagawa | 2006-06-22 16:04:48 +0900
packaging 0.7.3
Line 
1 package Plagger::Plugin::Publish::CHTML;
2 use strict;
3 use base qw( Plagger::Plugin );
4
5 use Encode;
6 use Digest::MD5 qw(md5_hex);
7 use File::Path;
8
9 sub register {
10     my($self, $context) = @_;
11     $context->register_hook(
12         $self,
13         'publish.feed' => \&feed,
14         'publish.finalize' => \&finalize,
15     );
16     $self->chtml_init($context);
17 }
18
19 sub chtml_init {
20     my ($self, $context) = @_;
21     $self->{context} = $context;
22     $self->conf->{encoding} ||= 'cp932';
23     $self->{id} = time;
24     @{$self->{feeds}} = ();
25     unless ($self->conf->{work}) {
26         $context->error("Can't parse value in work");
27     }
28     $self->conf->{title} ||= __PACKAGE__;
29     $self->conf->{mobile_gw} = undef unless $self->conf->{mobile_gw} =~ m{^https?://.*?/}i;
30 }
31
32 sub id { shift->{id} }
33 sub context { shift->{context} }
34 sub work { shift->conf->{work} }
35
36 sub add {
37     my($self, $feed) = @_;
38     push @{ $self->{feeds} }, $feed;
39 }
40
41 sub feeds {
42     my $self = shift;
43     wantarray ? @{ $self->{feeds} } : $self->{feeds};
44 }
45
46 sub feed {
47     my($self, $context, $args) = @_;
48
49     my $feed = $args->{feed} or return;
50     my $feed_path = $self->work . '/feeds/' . $feed->id;
51     my $publish_path = "$feed_path/" . $self->id;
52
53     mkpath($publish_path);
54     foreach my $entry ($feed->entries) {
55         my $entry_id = md5_hex($entry->permalink);
56         $self->write("$publish_path/$entry_id.html",
57                      $self->entry_templatize($feed, $entry));
58         $entry->{feed2entry_link} = $self->id . "/$entry_id.html";
59     }
60
61     $self->write("$publish_path.html",
62                  $self->feed_templatize($feed, $self->earlier($feed_path)),
63                  "$feed_path/index.html");
64
65     $self->add(+{
66         feed_link => './feeds/' . $feed->id . '/' . $self->id . '.html',
67         title  => $feed->title || '(no-title)',
68         lastdate => $feed->entries->[-1]->date,
69         count => scalar(@{$feed->entries}),
70     });
71 }
72
73 sub finalize {
74     my($self, $context) = @_;
75
76     return unless @{$self->feeds};
77     $self->write($self->work . '/' . $self->id . '.html',
78                  $self->index_templatize($self->earlier($self->work)),
79                  $self->work . '/index.html');
80 }
81
82 sub entry_templatize {
83     my($self, $feed, $entry) = @_;
84     $self->templatize('chtml_entry.tt', {
85         conf => $self->conf,
86         feed => $feed,
87         entry => $entry,
88         strip_html => sub {
89             my $html = shift;
90             $html =~ s|\s{2,}||og;
91             $html =~ s|<[bh]r.*?>|\n|ogi;
92             $html =~ s|<.*?>||og;
93             $html;
94         }});
95 }
96
97 sub feed_templatize {
98     my($self, $feed, $earlier) = @_;
99     $self->templatize('chtml_feed.tt', {
100         conf => $self->conf,
101         feed => $feed,
102         earlier => $earlier,
103     });
104 }
105
106 sub index_templatize {
107     my($self, $earlier) = @_;
108     $self->templatize('chtml_index.tt', {
109         conf => $self->conf,
110         feeds => [ $self->feeds ],
111         earlier => $earlier,
112     });
113 }
114
115 sub templatize {
116     my $self = shift;
117     my $tt = $self->context->template();
118     $tt->process(shift, shift, \my $out) or $self->context->error($tt->error);
119     $out;
120 }
121
122 sub write {
123     my ($self, $file, $chtml, $symlink) = @_;
124     open my $out, ">:encoding($self->{conf}->{encoding})", $file or $self->context->error("$file: $!");
125     local $PerlIO::encoding::fallback = Encode::FB_HTMLCREF;
126     print $out $chtml;
127     close $out;
128     $self->symlink($file, $symlink) if $symlink;
129 }
130
131 sub symlink {
132     my ($self, $old, $new) = @_;
133     unlink $new if -e $new;
134     symlink $old, $new;
135 }
136
137 sub earlier {
138     my ($self, $path) = @_;
139     my $earlier;
140     my $file = "$path/earlier";
141     if (open my $in, $file) {
142         $earlier = <$in>;
143         close $in;
144     }
145     open my $out, ">$file" or $self->context->error("$file: $!");
146     print $out $self->id;
147     close $out;
148     $earlier;
149 }
150 1;
Note: See TracBrowser for help on using the browser.