This commit is contained in:
logmanoriginal 2016-09-04 13:51:19 +02:00
commit 671703cd37
6 changed files with 225 additions and 126 deletions

View file

@ -1,40 +1,26 @@
<?php
class AcrimedBridge extends RssExpander{
class AcrimedBridge extends FeedExpander {
const MAINTAINER = "qwertygc";
const NAME = "Acrimed Bridge";
const URI = "http://www.acrimed.org/";
const DESCRIPTION = "Returns the newest articles.";
const MAINTAINER = "qwertygc";
const NAME = "Acrimed Bridge";
const URI = "http://www.acrimed.org/";
const DESCRIPTION = "Returns the newest articles.";
public function collectData(){
$this->collectExpandableDatas("http://www.acrimed.org/spip.php?page=backend");
}
public function collectData(){
protected function parseItem($newsItem){
$item = $this->parseRSS_2_0_Item($newsItem);
$this->collectExpandableDatas(static::URI.'spip.php?page=backend');
$hs = new HTMLSanitizer();
$articlePage = $this->getSimpleHTMLDOM($newsItem->link);
$article = $hs->sanitize($articlePage->find('article.article1', 0)->innertext);
$article = HTMLSanitizer::defaultImageSrcTo($article, "http://www.acrimed.org/");
$item['content'] = $article;
}
protected function parseRSSItem($newsItem) {
$hs = new HTMLSanitizer();
$namespaces = $newsItem->getNameSpaces(true);
$dc = $newsItem->children($namespaces['dc']);
$item = array();
$item['uri'] = trim($newsItem->link);
$item['title'] = trim($newsItem->title);
$item['timestamp'] = strtotime($dc->date);
$articlePage = $this->getSimpleHTMLDOM($newsItem->link);
$article = $hs->sanitize($articlePage->find('article.article1', 0)->innertext);
$article = HTMLSanitizer::defaultImageSrcTo($article, static::URI);
$item['content'] = $article;
return $item;
}
return $item;
}
public function getCacheDuration(){
return 4800; // 2 hours

View file

@ -0,0 +1,62 @@
<?php
class FeedExpanderExampleBridge extends FeedExpander {
const MAINTAINER = 'logmanoriginal';
const NAME = 'FeedExpander Example';
const URI = '#';
const DESCRIPTION = 'Example bridge to test FeedExpander';
const PARAMETERS = array(
'Feed' => array(
'version' => array(
'name' => 'Version',
'type' => 'list',
'required' => true,
'title' => 'Select your feed format/version',
'defaultValue' => 'RSS 2.0',
'values' => array(
'RSS 0.91' => 'rss_0_9_1',
'RSS 1.0' => 'rss_1_0',
'RSS 2.0' => 'rss_2_0',
'ATOM 1.0' => 'atom_1_0'
)
)
)
);
public function collectData(){
switch($this->getInput('version')){
case 'rss_0_9_1':
parent::collectExpandableDatas('http://static.userland.com/gems/backend/sampleRss.xml');
break;
case 'rss_1_0':
parent::collectExpandableDatas('http://feeds.nature.com/nature/rss/current?format=xml');
break;
case 'rss_2_0':
parent::collectExpandableDatas('http://feeds.rssboard.org/rssboard?format=xml');
break;
case 'atom_1_0':
parent::collectExpandableDatas('http://segfault.linuxmint.com/feed/atom/');
break;
default: $this->returnClientError('Unknown version ' . $this->getInput('version') . '!');
}
}
protected function parseItem($newsItem) {
switch($this->getInput('version')){
case 'rss_0_9_1':
return $this->parseRSS_0_9_1_Item($newsItem);
break;
case 'rss_1_0':
return $this->parseRSS_1_0_Item($newsItem);
break;
case 'rss_2_0':
return $this->parseRSS_2_0_Item($newsItem);
break;
case 'atom_1_0':
return $this->parseATOMItem($newsItem);
break;
default: $this->returnClientError('Unknown version ' . $this->getInput('version') . '!');
}
}
}

View file

@ -1,34 +1,22 @@
<?php
define("FREENEWS_RSS", 'http://feeds.feedburner.com/Freenews-Freebox?format=xml');
class FreenewsBridge extends RssExpander {
class FreenewsBridge extends FeedExpander {
const MAINTAINER = "mitsukarenai";
const NAME = "Freenews";
const URI = "http://freenews.fr";
const DESCRIPTION = "Un site d'actualité pour les freenautes (mais ne parlant pas que de la freebox). Ne rentrez pas d'id si vous voulez accéder aux actualités générales.";
const MAINTAINER = "mitsukarenai";
const NAME = "Freenews";
const URI = "http://freenews.fr";
const DESCRIPTION = "Un site d'actualité pour les freenautes (mais ne parlant pas que de la freebox). Ne rentrez pas d'id si vous voulez accéder aux actualités générales.";
public function collectData(){
parent::collectExpandableDatas(FREENEWS_RSS);
parent::collectExpandableDatas('http://feeds.feedburner.com/Freenews-Freebox?format=xml');
}
protected function parseRSSItem($newsItem) {
$item = array();
$item['title'] = trim($newsItem->title);
$this->debugMessage("item has for title \"".$item['title']."\"");
if(empty($newsItem->guid)) {
$item['uri'] = (string) $newsItem->link;
} else {
$item['uri'] = (string) $newsItem->guid;
}
// now load that uri from cache
$this->debugMessage("now loading page ".$item['uri']);
protected function parseItem($newsItem) {
$item = $this->parseRSS_2_0_Item($newsItem);
$articlePage = $this->get_cached($item['uri']);
$content = $articlePage->find('.post-container', 0);
$item['content'] = $content->innertext;
$item['author'] = $articlePage->find('a[rel=author]', 0)->innertext;
// format should parse 2014-03-25T16:21:20Z. But, according to http://stackoverflow.com/a/10478469, it is not that simple
$item['timestamp'] = $this->RSS_2_0_time_to_timestamp($newsItem);
return $item;
}
}

View file

@ -1,35 +1,19 @@
<?php
class Les400CulsBridge extends RssExpander{
const MAINTAINER = "unknown";
const NAME = "Les 400 Culs";
const URI = "http://sexes.blogs.liberation.fr/";
const DESCRIPTION = "La planete sexe vue par Agnes Girard via rss-bridge";
class Les400CulsBridge extends FeedExpander{
const MAINTAINER = "unknown";
const NAME = "Les 400 Culs";
const URI = "http://sexes.blogs.liberation.fr/";
const DESCRIPTION = "La planete sexe vue par Agnes Girard via rss-bridge";
public function collectData(){
$this->collectExpandableDatas(self::URI.'feeds/');
$this->collectExpandableDatas(self::URI . 'feeds/');
}
protected function parseRSSItem($newsItem) {
$item = array();
$item['title'] = trim((string) $newsItem->title);
$this->debugMessage("browsing item ".var_export($newsItem, true));
if(empty($newsItem->guid)) {
$item['uri'] = (string) $newsItem->link;
} else {
$item['uri'] = (string) $newsItem->guid;
}
// now load that uri from cache
$this->debugMessage("now loading page ".$item['uri']);
// $articlePage = $this->get_cached($item['uri']);
// $content = $articlePage->find('.post-container', 0);
$item['content'] = (string) $newsItem->description;
$item['author'] = (string) $newsItem->author;
$item['timestamp'] = $this->RSS_2_0_time_to_timestamp($newsItem);
return $item;
protected function parseItem($newsItem){
return $this->parseRSS_2_0_Item($newsItem);
}
public function getCacheDuration(){
return 7200; // 2h hours
}

View file

@ -1,5 +1,5 @@
<?php
class TheOatmealBridge extends RssExpander{
class TheOatmealBridge extends FeedExpander{
const MAINTAINER = "Riduidel";
const NAME = "The Oatmeal";
@ -10,44 +10,17 @@ class TheOatmealBridge extends RssExpander{
$this->collectExpandableDatas('http://feeds.feedburner.com/oatmealfeed');
}
protected function parseItem($newsItem) {
$item = $this->parseRSS_1_0_Item($newsItem);
/**
* Since the oatmeal produces a weird RSS feed, I have to fix it by loading the items separatly from the feed infos
*/
protected function collect_RSS_2_0_data($rssContent) {
$rssContent->registerXPathNamespace("dc", "http://purl.org/dc/elements/1.1/");
$rssHeaderContent = $rssContent->channel[0];
$this->debugMessage("RSS content is ===========\n".var_export($rssHeaderContent, true)."===========");
$this->load_RSS_2_0_feed_data($rssHeaderContent);
foreach($rssContent->item as $item) {
$this->debugMessage("parsing item ".var_export($item, true));
$this->items[] = $this->parseRSSItem($item);
}
}
protected function parseRSSItem($newsItem) {
$namespaces = $newsItem->getNameSpaces(true);
$dc = $newsItem->children($namespaces['dc']);
$rdf = $newsItem->children($namespaces['rdf']);
$item = array();
$item['title'] = trim($newsItem->title);
$this->debugMessage("browsing Oatmeal item ".var_export($newsItem, true));
$item['uri']=(string) $newsItem->attributes($namespaces['rdf'])->about;
// now load that uri from cache
$this->debugMessage("now loading page ".$item['uri']);
$articlePage = $this->get_cached($item['uri']);
$content = $articlePage->find('#comic', 0);
if($content==null) {
$content = $articlePage->find('#blog');
}
$item['content'] = $content->innertext;
if(is_null($content)) // load alternative
$content = $articlePage->find('#blog', 0);
if(!is_null($content))
$item['content'] = $content->innertext;
$this->debugMessage("dc content is ".var_export($dc, true));
$item['author'] = (string) $dc->creator;
$item['timestamp'] = DateTime::createFromFormat(DateTime::ISO8601, $dc->date)->getTimestamp();
$this->debugMessage("writtem by ".$item['author']." on ".$item['timestamp']);
return $item;
}