webgui/lib/WebGUI/Asset/Wobject/SyndicatedContent.pm
2006-01-02 03:12:54 +00:00

647 lines
22 KiB
Perl

package WebGUI::Asset::Wobject::SyndicatedContent;
#-------------------------------------------------------------------
# WebGUI is Copyright 2001-2005 Plain Black Corporation.
#-------------------------------------------------------------------
# Please read the legal notices (docs/legal.txt) and the license
# (docs/license.txt) that came with this distribution before using
# this software.
#-------------------------------------------------------------------
# http://www.plainblack.com info@plainblack.com
#-------------------------------------------------------------------
use HTML::Entities;
use strict;
use Storable;
use Tie::CPHash;
use Tie::IxHash;
use WebGUI::Cache;
use WebGUI::DateTime;
use WebGUI::HTMLForm;
use WebGUI::HTML;
use WebGUI::International;
use WebGUI::Privilege;
use WebGUI::Session;
use WebGUI::Asset::Wobject;
use XML::RSSLite;
use XML::RSS::Creator;
use LWP::UserAgent;
use WebGUI::ErrorHandler;
use WebGUI::URL;
use WebGUI::HTTP;
use WebGUI::Style;
use POSIX qw/floor/;
my $hasEncode=1;
eval ' use Encode qw(from_to); '; $hasEncode=0 if $@;
our @ISA = qw(WebGUI::Asset::Wobject);
=head1 NAME
Package WebGUI::Asset::Wobject::SyndicatedContent
=head1 DESCRIPTION
Displays items and channels from RSS feeds.
=head1 SYNOPSIS
use WebGUI::Asset::Wobject::SyndicatedWobject;
=head1 METHODS
These methods are available from this class:
=cut
#-------------------------------------------------------------------
=head2 definition ( definition )
Defines the properties of this asset.
=head3 definition
A hash reference passed in from a subclass definition.
=cut
sub definition {
my $class = shift;
my $definition = shift;
my %properties;
tie %properties, 'Tie::IxHash';
%properties = (
templateId =>{
tab=>"display",
fieldType=>'template',
defaultValue=>'PBtmpl0000000000000065',
namespace=>'SyndicatedContent',
label=>WebGUI::International::get(72,'Asset_SyndicatedContent'),
hoverHelp=>WebGUI::International::get('72 description','Asset_SyndicatedContent')
},
rssUrl=>{
tab=>"properties",
defaultValue=>undef,
fieldType=>'textarea',
label=>WebGUI::International::get(1,'Asset_SyndicatedContent'),
hoverHelp=>WebGUI::International::get('1 description','Asset_SyndicatedContent')
},
maxHeadlines=>{
tab=>"properties",
fieldType=>'integer',
defaultValue=>10,
label=>WebGUI::International::get(3,'Asset_SyndicatedContent'),
hoverHelp=>WebGUI::International::get('3 description','Asset_SyndicatedContent')
},
displayMode=>{
tab=>"display",
fieldType=>'selectBox',
defaultValue=>'interleaved',
options=>{
'interleaved'=>WebGUI::International::get('interleaved','Asset_SyndicatedContent'),
'grouped'=>WebGUI::International::get('grouped','Asset_SyndicatedContent'),
},
sortByValue=>1,
label=>WebGUI::International::get('displayModeLabel','Asset_SyndicatedContent'),
hoverHelp=>WebGUI::International::get('displayModeLabel description','Asset_SyndicatedContent'),
subtext=>WebGUI::International::get('displayModeSubtext','Asset_SyndicatedContent')
},
hasTerms=>{
tab=>"properties",
fieldType=>'text',
defaultValue=>'',
label=>WebGUI::International::get('hasTermsLabel','Asset_SyndicatedContent'),
hoverHelp=>WebGUI::International::get('hasTermsLabel description','Asset_SyndicatedContent'),
maxlength=>255
}
);
push(@{$definition}, {
assetName=>WebGUI::International::get('assetName','Asset_SyndicatedContent'),
uiLevel=>6,
autoGenerateForms=>1,
icon=>'syndicatedContent.gif',
tableName=>'SyndicatedContent',
className=>'WebGUI::Asset::Wobject::SyndicatedContent',
properties=>\%properties
});
return $class->SUPER::definition($definition);
}
#-------------------------------------------------------------------
# strip all html tags from the given data structure. This is important to
# prevent cross site scripting attacks
#my $_stripped_html = {};
sub _strip_html {
#my ($data) = @_;
if (ref($_[0]) eq 'HASH') {
keys(%{$_[0]});
while (my ($name, $val) = each (%{$_[0]})) {
$_[0]->{$name} = _strip_html($val);
}
} elsif (ref($_[0]) eq 'ARRAY') {
for (my $i = 0; $i < @{$_[0]}; $i++) {
$_[0]->[$i] = _strip_html($_[0]->[$i]);
}
} else {
if ($_[0]) {
$_[0] =~ s/\&lt;/</g;
$_[0] =~ s/\&gt;/>/g;
$_[0] = WebGUI::HTML::filter($_[0], 'all');
##Unencode double encoded entities. This is usually done
##by passing XML::RSSLite an already encoded entity.
$_[0] =~ s/\&amp;(?=(#[0-9]+|#x[0-9a-fA-F]+|\w+);)/&/g;
}
}
return $_[0];
}
#-------------------------------------------------------------------
# horrible kludge to find the channel or item record
# in the varying kinds of rss structures returned by RSSLite
sub _find_record {
my ($data, $regex) = @_;
if (ref($data) eq 'HASH') {
# reset the hash before calling each()
keys(%{$data});
while (my ($name, $val) = each(%{$data})) {
if ($name =~ $_[1]) {
if ((((ref($val) eq 'HASH') &&
($val->{link} || $val->{title} ||
$val->{description})) ||
((ref($val) eq 'ARRAY') && @{$val} &&
(ref($val->[0]) eq 'HASH') &&
($val->[0]->{link} ||
$val->[0]->{title} ||
$val->[0]->{description})))) {
return $val;
}
}
if (my $record = _find_record($val, $regex)) {
return $record;
}
}
}
return undef;
}
#-------------------------------------------------------------------
# Copy the guid field to the link field if the guid looks like a link.
# This is a kludge that gets around the fact that some folks use the link
# field as the link to the story while others use it as the link
# to the story about which the story is written. The webuig templates seem
# to assume the former, so we should use the guid instead of the link, b/c
# the guid, if it is a link, always means the former.
# Also copy the first few words of the description into the title if
# there is no title
sub _normalize_items {
#my ($items) = @_;
# max number of words to take from description to fill in an empty
# title
my $max_words = 10;
for my $item (@{$_[0]}) {
if ($item->{guid} && ($item->{guid} =~ /^http:\/\//i)) {
$item->{link} = $item->{guid};
}
if (!$item->{title}) {
my @description_words = split(/\s/, $item->{description});
if (@description_words <= $max_words) {
$item->{title} = $item->{description};
} else {
$item->{title} = join(' ', @description_words[0..$max_words-1]) .
' ...';
}
}
# IE doesn't recognize &apos;
$item->{title} =~ s/&apos;/\'/;
$item->{description} =~ s/&apos;/\'/;
}
}
#-------------------------------------------------------------------
sub _get_rss_data {
my $url = shift;
my $cache = WebGUI::Cache->new('url:' . $url, 'RSS');
my $rss_serial = $cache->get;
my $rss = {};
if ($rss_serial) {
$rss = Storable::thaw($rss_serial);
} else {
my $ua = LWP::UserAgent->new(timeout => 5);
my $response = $ua->get($url);
if (!$response->is_success()) {
$self->session->errorHandler->warn("Error retrieving url '$url': " .
$response->status_line());
return undef;
}
my $xml = $response->content();
# Convert encoding if needed / Perl 5.8.0 or up required.
if ($] >= 5.008 && $hasEncode) {
$xml =~ /<\?xml.*?encoding=['"](\S+)['"]/i;
my $xmlEncoding = $1 || 'utf8';
my $encoding = 'utf8';
if (lc($xmlEncoding) ne lc($encoding)) {
eval { from_to($xml, $xmlEncoding, $encoding) };
$self->session->errorHandler->warn($@) if ($@);
}
}
my $rss_lite = {};
eval {
XML::RSSLite::parseXML($rss_lite, \$xml);
};
if ($@) {
$self->session->errorHandler->warn("error parsing rss for url $url :".$@);
#Returning undef on a parse failure is a change from previous behaviour,
#but it SHOULDN'T have a major effect.
return undef;
}
# make sure that the {channel} points to the channel
# description record and that {items} points to the list
# of items. without this voodoo, different versions of
# rss return the data in different places in the data
# structure.
$rss_lite = {channel => $rss_lite};
if (!($rss->{channel} =
_find_record($rss_lite, qr/^channel$/))) {
$self->session->errorHandler->warn("unable to find channel info for url $url");
}
if (!($rss->{items} = _find_record($rss_lite, qr/^items?$/))) {
$self->session->errorHandler->warn("unable to find item info for url $url");
$rss->{items} = [];
}
_strip_html($rss);
$rss->{items} = [ $rss->{items} ] unless (ref $rss->{items} eq 'ARRAY');
_normalize_items($rss->{items});
#Assign dates "globally" rather than when seen in a viewed feed.
#This is important because we can "filter" now and want to ensure we keep order
#correctly as new items appear.
_assign_rss_dates($rss->{items});
#Default to an hour timeout
$cache->set(Storable::freeze($rss), 3600);
}
return $rss;
}
#-------------------------------------------------------------------
# rss items don't have a standard date, so timestamp them the first time
# we see them and use that timestamp as the date. Periodically nuke the
# whole database to keep the thing from growing too large
sub _assign_rss_dates {
my ($items) = @_;
for my $item (@{$items}) {
my $key = 'dates:' . ($item->{guid} || $item->{title} ||
$item->{description} || $item->{link});
my $cache = WebGUI::Cache->new($key, 'RSS');
if (my $date = $cache->get()) {
$item->{date} = $date;
} else {
$item->{date} = time();
$cache->set($item->{date}, '1 year');
}
}
@{$items} = sort { $b->{date} <=> $a->{date} } @{$items};
}
#-------------------------------------------------------------------
# $items is the hashref to put items into.
# $rss_feeds is an arrayref of all the feeds in this wobject
# The only difference between an "interleaved" feed and a grouped feed
# is the order the items are output.
sub _create_grouped_items{
my($items,$rss_feeds,$maxHeadlines,$hasTermsRegex)=@_;
_create_interleaved_items($items,$rss_feeds,$maxHeadlines,$hasTermsRegex);
@$items=sort{$a->{'site_title'} cmp $b->{'site_title'}} @$items;
#Loop through the items and output the "site_
my $siteTitleTracker;
foreach(@$items){
if($siteTitleTracker ne $_->{site_title}){
$_->{new_rss_site}=1;
}
$siteTitleTracker=$_->{site_title};
}
}
#-------------------------------------------------------------------
# Loop through the feeds for this wobject
# and push in the items in "interleaved mode"
# No need to return because we're doing everything by reference.
sub _create_interleaved_items{
my($items,$rss_feeds,$maxHeadlines,$hasTermsRegex)=@_;
my $items_remain = 1;
while((@$items < $maxHeadlines) && $items_remain){
foreach my $rss(@$rss_feeds){
$items_remain=0;
if(defined $rss->{items}
&& @$items < $maxHeadlines
&& (my $item = shift @{$rss->{items}})
){
$item->{site_title}=$rss->{channel}->{title};
$item->{site_link}=$rss->{channel}->{link};
if(! $hasTermsRegex || _check_hasTerms($item,$hasTermsRegex)){
push @{$items},$item;
}
if (@{$rss->{items}}) {
$items_remain = 1;
}
}
}
}
}
#-------------------------------------------------------------------
# Uses the regex constructed in _get_items (with the terms defaulting to OR)
# to see if the title or description associated with this item match the kinds
# of items we're looking for.
#
sub _check_hasTerms{
my($item,$hasTermsRegex)=@_;
my $to_check=$item->{title}.$item->{description};
if( $to_check =~ /$hasTermsRegex/gism){
return 1;
} else {
return 0;
}
}
#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
sub _make_regex{
my $terms=shift;
my @terms=split(/,/,$terms);
return join('|',@terms);
}
#-------------------------------------------------------------------
# So- We're going to manage an "aggregate cache" that represents
# the rendering of the cumulative feeds in a Syndicated Wobject,
# but let each feed "fend for itself" based on URL in the cache.
#
# This means we can set up the hourly task to get and cache each
# individual feed WITHOUT having to re-request (undoubtedly the slowest
# part of every RSS parsing action is the network traffic) each feed
# when we re-render each aggregrate representation.
#
# If, however, a feed expires between hourly tasks, it will be re-requested and
# parsed per the usual. BUT, if a feed ever goes un-requested for more than an hour,
# then it's retrieval schedule will be taken over by the hourly task, and we'll
# be pre-seeding the RSS object cache automatically.
#
# Having the caching set up this way means we can re-use the same raw feed all over the site without
# having each wobject request it separately, ASSUMING the URL is the same.
#
# All the values that may have an effect on the composition of items
# are included in the cache key for the aggregate representation.
sub _get_items {
my $self = shift;
my $urls = shift;
my $maxHeadlines = shift;
my $displayMode=$self->getValue('displayMode');
my $hasTermsRegex=_make_regex($self->getValue('hasTerms'));
my $maxHeadlines=$self->getValue('maxHeadlines');
my $key=join(':',('aggregate', $displayMode,$hasTermsRegex,$maxHeadlines,$self->get('rssUrl')));
my $cache = WebGUI::Cache->new($key, 'RSS');
my $items = Storable::thaw($cache->get());
my @rss_feeds;
if (!$items) {
$items = [];
for my $url (@{$urls}) {
my $rss_info=_get_rss_data($url);
push(@rss_feeds, $rss_info) if($rss_info);
}
#Sort feeds in order by channel title.
#@rss_feeds=sort{$a->{channel}->{title} cmp $b->{channel}->{title}} @rss_feeds;
if ($displayMode eq 'grouped') {
_create_grouped_items($items,\@rss_feeds,$maxHeadlines,$hasTermsRegex);
} else {
_create_interleaved_items($items,\@rss_feeds,$maxHeadlines,$hasTermsRegex);
}
#@{$items} = sort { $b->{date} <=> $a->{date} } @{$items};
$cache->set(Storable::freeze($items), 3600);
}
#So return the item loop and the first RSS feed, because
#when we're parsing a single feed we can use that feed's title and
#description for channel.title, channel.link, and channel.description
return ($items,\@rss_feeds);
}
#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
=head2 view()
Returns the rendered output of the wobject.
=cut
sub view {
my $self = shift;
my $rssFlavor = shift;
$self->logView() if ($self->session->setting->get("passiveProfilingEnabled"));
my $maxHeadlines = $self->get('maxHeadlines') || 1000000;
my @urls = split(/\s+/,$self->get('rssUrl'));
return $self->processTemplate({},$self->get('templateId')) unless (scalar(@urls));
my $title=$self->get('title');
#We came into this subroutine as
my $rssObject=($rssFlavor) ? XML::RSS::Creator->new(version=>$rssFlavor) : undef;
my %var;
my($item_loop,$rss_feeds)=$self->_get_items(\@urls, $maxHeadlines,$rssObject);
if(@$rss_feeds > 1){
#If there is more than one (valid) feed in this wobject, put in the wobject description info.
$var{'channel.title'} = $title;
$var{'channel.description'} = $self->get('description');
} else {
#One feed. Put in the info from the feed.
$var{'channel.title'} = $rss_feeds->[0]->{channel}->{title} || $title;
$var{'channel.link'} = $rss_feeds->[0]->{channel}->{link};
$var{'channel.description'} = $rss_feeds->[0]->{channel}->{description};
}
$self->_createRSSURLs(\%var);
$var{item_loop} = $item_loop;
#Construct the title for the link.
_constructRSSHeadTitleLink(\%var,$title || $var{'channel.title'});
if ($rssObject) {
$self->_constructRSS($rssObject,\%var);
my $rss=$rssObject->as_string;
WebGUI::HTTP::setMimeType('text/xml');
#Looks like a kludge, but what this does is put in the proper
#XSLT stylesheet so the RSS doesn't look like total ass.
my $siteURL=$self->session->url->getSiteURL().$self->session->url->gateway();
$rss=~s|<\?xml version="1\.0" encoding="UTF\-8"\?>|<\?xml version="1\.0" encoding="UTF\-8"\?>\n<?xml\-stylesheet type="text/xsl" href="${siteURL}xslt/rss$rssFlavor.xsl"\?>\n|;
return $rss;
} else {
return $self->processTemplate(\%var,$self->get('templateId'));
}
}
#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
sub _constructRSSHeadTitleLink{
my($var,$rssTitle)=@_;
my $rssFeedSuffix=WebGUI::International::get('RSS Feed Title Suffix','Asset_SyndicatedContent');
my $title = ($rssTitle) ? ($rssTitle." ".$rssFeedSuffix) : $rssFeedSuffix;
$title =~ s/\"/&quot;/g;
$self->session->style->setLink($var->{'rss.url'},
{ rel=> 'alternate',
type=> 'application/rss+xml',
title=> ($rssTitle) ? ($rssTitle." ".$rssFeedSuffix) : $rssFeedSuffix }
);
}
#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
sub _constructRSS{
my($self,$rssObject,$var)=@_;
#They've chosen to emit this as an RSS feed, in one of the four flavors we support.
$rssObject->channel(
title=>$var->{'channel.title'} || $self->get('title'),
link=>$self->session->url->page('',1),
description=>$var->{'channel.description'} || ''
);
foreach my $item (@{$var->{item_loop}}) {
# I know this seems kludgy, but because XML::RSSLite parses
# feeds loosely, sometimes it returns a data structure when it shouldn't.
# So we're only pushing in attributes when they AREN'T a reference to
# a data structure.
my %attributes;
foreach my $attribute(keys %$item){
$attributes{$attribute}=$item->{$attribute} if (! ref($item->{$attribute}));
}
$rssObject->add_item(%attributes);
}
}
#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
sub _createRSSURLs{
my $self=shift;
my $var=shift;
foreach({ver=>'1.0',param=>'10'},{ver=>'0.9',param=>'090'},{ver=>'0.91',param=>'091'},{ver=>'2.0',param=>'20'}){
$var->{'rss.url.'.$_->{ver}}=$self->getUrl('func=viewRSS'.$_->{param});
}
$var->{'rss.url'}=$self->getUrl('func=viewRSS20');
}
#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
#=head2 www_edit()
#Sets parameters and returns a form to edit this wobject.
#=cut
#sub www_edit {
# my $self = shift;
# return WebGUI::Privilege::insufficient() unless $self->canEdit;
# $self->getAdminConsole->setHelp('syndicated content add/edit','Asset_SyndicatedContent');
# return $self->getAdminConsole->render($self->getEditForm->print,WebGUI::International::get('4','Asset_SyndicatedContent'));
#}
#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
=head2 www_viewRSS090()
Emit an RSS 0.9 feed.
=cut
sub www_viewRSS090{
my $self=shift;
return $self->view('0.9');
}
#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
=head2 www_viewRSS091()
Emit an RSS 0.91 feed.
=cut
sub www_viewRSS091{
my $self=shift;
return $self->view('0.91');
}
#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
=head2 www_viewRSS10()
Emit an RSS 1.0 feed.
=cut
sub www_viewRSS10{
my $self=shift;
return $self->view('1.0');
}
#+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
=head2 www_viewRSS20()
Emit an RSS 2.0 feed.
=cut
sub www_viewRSS20{
my $self=shift;
return $self->view('2.0');
}
1;