296 lines
9.9 KiB
PHP
Executable File
296 lines
9.9 KiB
PHP
Executable File
<?php
|
|
|
|
namespace Franzz\Objects;
|
|
use \Settings;
|
|
|
|
/**
|
|
* RSS Feed Class
|
|
* @author franzz
|
|
* @version 2.1
|
|
*/
|
|
class Feed extends PhpObject
|
|
{
|
|
const CHANNEL_TAGS = array('title', 'link', 'copyright', 'description', 'language', 'lastBuildDate', 'generator', 'webMaster');
|
|
const ITEM_TAGS = array('title', 'author', 'link', 'category', 'description', 'pubDate', 'guid');
|
|
|
|
private $asChannel;
|
|
private $asItems;
|
|
|
|
/**
|
|
* Constructor
|
|
* @param array $asChannel Description of the feed: fields 'title', 'link' (optional), 'copyright', 'description', 'language', 'webMaster'
|
|
* @param array $asItems Feed item data: fields 'title', 'author', 'link', 'category', 'description', 'pubDate', 'guid'
|
|
*/
|
|
public function __construct($asChannel=array(), $asItems=array())
|
|
{
|
|
parent::__construct(__CLASS__);
|
|
if(!empty($asChannel) && !array_key_exists('link', $asChannel))
|
|
{
|
|
$asChannel['link'] = $_SERVER['REQUEST_SCHEME'].'://'.$_SERVER['SERVER_NAME'].$_SERVER['SCRIPT_NAME'].'/rss';
|
|
}
|
|
$this->asChannel = $asChannel;
|
|
|
|
//Items
|
|
$this->removeItems();
|
|
array_walk($asItems, array($this, 'addItem'));
|
|
}
|
|
|
|
public function loadRss($sUrl, $sMode='curl')
|
|
{
|
|
$sRssContent = '';
|
|
switch($sMode) {
|
|
case 'curl':
|
|
$oCurl = curl_init();
|
|
curl_setopt($oCurl, CURLOPT_URL, $sUrl);
|
|
curl_setopt($oCurl, CURLOPT_HEADER, false);
|
|
curl_setopt($oCurl, CURLOPT_RETURNTRANSFER, true);
|
|
curl_setopt($oCurl, CURLOPT_USERAGENT, array_key_exists('HTTP_USER_AGENT', $_SERVER)?$_SERVER['HTTP_USER_AGENT']:'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:105.0) Gecko/20100101 Firefox/105.0');
|
|
$fCookieJar = @tempnam('/tmp','cookie');
|
|
curl_setopt($oCurl, CURLOPT_COOKIESESSION, true);
|
|
curl_setopt($oCurl, CURLOPT_COOKIEJAR, $fCookieJar);
|
|
curl_setopt($oCurl, CURLOPT_COOKIEFILE, $fCookieJar);
|
|
//curl_setopt($oCurl, CURLOPT_ENCODING, 'gzip');
|
|
$sRssContent = curl_exec($oCurl);
|
|
curl_close($oCurl);
|
|
break;
|
|
case 'w3m':
|
|
exec('w3m -dump '.$sUrl, $asOutput, $iResult);
|
|
$sRssContent = implode("\n", $asOutput);
|
|
break;
|
|
}
|
|
|
|
//Parse document encoding (useless)
|
|
$sEncoding = $this->getPregMatch("'encoding=[\'\"](.*?)[\'\"]'si", $sRssContent);
|
|
|
|
//Parse Channel info
|
|
if(preg_match("'<channel.*?>(.*?)</channel>'si", $sRssContent, $sChannelContent)) {
|
|
foreach(self::CHANNEL_TAGS as $sChannelTag)
|
|
{
|
|
$sTagContent = $this->getPregMatch("'<$sChannelTag.*?>(.*?)</$sChannelTag>'si", $sChannelContent[1]);
|
|
if($sTagContent != '') $this->asChannel[$sChannelTag] = $sTagContent;
|
|
}
|
|
}
|
|
//Parse Text Input info
|
|
/*preg_match("'<textinput(|[^>]*[^/])>(.*?)</textinput>'si", $sRssContent, $out_textinfo);
|
|
if (isset($out_textinfo[2])) {
|
|
foreach($this->textinputtags as $textinputtag) {
|
|
$temp = $this->getPregMatch("'<$textinputtag.*?>(.*?)</$textinputtag>'si", $out_textinfo[2]);
|
|
if ($temp != '') $result['textinput_'.$textinputtag] = $temp; // Set only if not empty
|
|
}
|
|
}*/
|
|
|
|
//Parse Image info
|
|
/*preg_match("'<image.*?>(.*?)</image>'si", $sRssContent, $out_imageinfo);
|
|
if (isset($out_imageinfo[1])) {
|
|
foreach($this->imagetags as $imagetag) {
|
|
$temp = $this->getPregMatch("'<$imagetag.*?>(.*?)</$imagetag>'si", $out_imageinfo[1]);
|
|
if ($temp != '') $result['image_'.$imagetag] = $temp; // Set only if not empty
|
|
}
|
|
}*/
|
|
|
|
//Parse Items
|
|
preg_match_all("'<item(| .*?)>(.*?)</item>'si", $sRssContent, $asItems);
|
|
//$i = 0;
|
|
foreach($asItems[2] as $asItem)
|
|
{
|
|
$asItemTags = array();
|
|
foreach(self::ITEM_TAGS as $sItemTag)
|
|
{
|
|
$sTagContent = $this->getPregMatch("'<$sItemTag.*?>(.*?)</$sItemTag>'si", $asItem);
|
|
if($sTagContent != '') $asItemTags[$sItemTag] = $sTagContent;
|
|
}
|
|
//Strip HTML tags and other bullshit from DESCRIPTION
|
|
//if($this->stripHTML && $this->asItems[$i]['description']) $this->asItems[$i]['description'] = strip_tags($this->unhtmlentities(strip_tags($this->asItems[$i]['description'])));
|
|
//Strip HTML tags and other bullshit from TITLE
|
|
//if($this->stripHTML && $this->asItems[$i]['title']) $this->asItems[$i]['title'] = strip_tags($this->unhtmlentities(strip_tags($this->asItems[$i]['title'])));
|
|
|
|
//Fix for author
|
|
if(!array_key_exists('author', $asItemTags))
|
|
{
|
|
$sTagContent = $this->getPregMatch("'<dc\:creator.*?>(.*?)</dc\:creator>'si", $asItem);
|
|
if($sTagContent != '') $asItemTags['author'] = $sTagContent;
|
|
}
|
|
|
|
$this->addItem($asItemTags);
|
|
}
|
|
}
|
|
|
|
public function addItem($asItem)
|
|
{
|
|
$this->asItems[] = $asItem;
|
|
return count($this->asItems) - 1;
|
|
}
|
|
|
|
public function changeItem($iItemId, $asAttributes) {
|
|
if(array_key_exists($iItemId, $this->asItems)) {
|
|
$this->asItems[$iItemId] = array_merge($this->asItems[$iItemId], $asAttributes);
|
|
}
|
|
}
|
|
|
|
public function removeItem($iItemId)
|
|
{
|
|
$bExist = array_key_exists($iItemId, $this->asItems);
|
|
if($bExist) unset($this->asItems[$iItemId]);
|
|
return $bExist;
|
|
}
|
|
|
|
public function removeItems()
|
|
{
|
|
$this->asItems = [];
|
|
}
|
|
|
|
public function filterItems($sField, $sRegex)
|
|
{
|
|
$this->asItems = array_filter($this->asItems, function($asItem) use ($sField, $sRegex) {
|
|
return preg_match($sRegex, $asItem[$sField]);
|
|
});
|
|
}
|
|
|
|
public function getItems()
|
|
{
|
|
return $this->asItems;
|
|
}
|
|
|
|
private function getGlobalPubDate()
|
|
{
|
|
$iGlobalPubDate = 0;
|
|
foreach($this->asItems as $asItem)
|
|
{
|
|
$iItemPubDate = strtotime($asItem['pubDate']);
|
|
if($iItemPubDate>$iGlobalPubDate)
|
|
{
|
|
$iGlobalPubDate = $iItemPubDate;
|
|
}
|
|
}
|
|
return self::cleanRss(self::getDate($iGlobalPubDate));
|
|
}
|
|
|
|
public function getFeed($bSetMime=true)
|
|
{
|
|
//Feed Channel
|
|
$sRssChannel = array_key_exists('title', $this->asChannel)?self::getHtml($this->asChannel['title'], 'title'):'';
|
|
$sRssChannel .= array_key_exists('link', $this->asChannel)?self::getHtml($this->asChannel['link'], 'link'):'';
|
|
$sRssChannel .= array_key_exists('copyright', $this->asChannel)?self::getHtml($this->asChannel['copyright'], 'copyright'):'';
|
|
$sRssChannel .= array_key_exists('description', $this->asChannel)?self::getHtml($this->asChannel['description'], 'description'):'';
|
|
$sRssChannel .= array_key_exists('link', $this->asChannel)?self::getHtml('', 'atom:link', '', '', array('href'=>$this->asChannel['link'], 'rel'=>'self', 'type'=>'application/atom+xml'), true):'';
|
|
$sRssChannel .= array_key_exists('language', $this->asChannel)?self::getHtml($this->asChannel['language'], 'language'):'';
|
|
$sRssChannel .= self::getHtml($this->getGlobalPubDate(), 'lastBuildDate');
|
|
$sRssChannel .= self::getHtml('Franzz RSS Feed Generator', 'generator');
|
|
$sRssChannel .= array_key_exists('webMaster', $this->asChannel)?self::getHtml($this->asChannel['webMaster'].' (Webmaster)', 'webMaster'):'';
|
|
|
|
//Feed Items
|
|
$asSortedItems = $this->rSortTimeMatrix($this->asItems, 'pubDate');
|
|
$sItems = implode("\n", array_map(array($this, 'buildItem'), $asSortedItems));
|
|
|
|
//Global Feed
|
|
$sFeed = '<?xml version="1.0" encoding="'.Settings::TEXT_ENC.'" ?>';
|
|
$sFeed .= self::getHtml(self::getHtml($sRssChannel.$sItems, 'channel'), 'rss', '', '', array('version'=>'2.0', 'xmlns:atom'=>'http://www.w3.org/2005/Atom'));
|
|
|
|
if($bSetMime) header('Content-type: application/rss+xml');
|
|
|
|
return $sFeed;
|
|
}
|
|
|
|
private static function getDate($sDate)
|
|
{
|
|
if(!is_numeric($sDate))
|
|
{
|
|
$sDate = strtotime($sDate);
|
|
}
|
|
return date('r', $sDate);
|
|
}
|
|
|
|
private function buildItem($asItem)
|
|
{
|
|
$sRssItem = self::getHtml(self::cleanRss($asItem['title']), 'title');
|
|
$sRssItem .= array_key_exists('author', $asItem)?self::getHtml(self::cleanRss($asItem['author']), 'author'):'';
|
|
$sRssItem .= array_key_exists('link', $asItem)?self::getHtml($asItem['link'], 'link'):'';
|
|
$sRssItem .= array_key_exists('category', $asItem)?self::getHtml($asItem['category'], 'category'):'';
|
|
$sRssItem .= self::getHtml(self::cleanRss($asItem['description']), 'description');
|
|
$sRssItem .= array_key_exists('pubDate', $asItem)?self::getHtml(self::getDate($asItem['pubDate']), 'pubDate'):'';
|
|
$sRssItem .= array_key_exists('guid', $asItem)?self::getHtml($asItem['guid'], 'guid', '', '', array('isPermaLink'=>'false')):'';
|
|
return self::getHtml($sRssItem, 'item');
|
|
}
|
|
|
|
private function getPregMatch($pattern, $subject)
|
|
{
|
|
preg_match($pattern, $subject, $out);
|
|
|
|
//if there is some result... process it and return it
|
|
if(isset($out[1]))
|
|
{
|
|
//If code page is set convert character encoding to required
|
|
//if ($this->cp != '') $out[1] = iconv($this->rsscp, $this->cp.'//TRANSLIT', $out[1]);
|
|
|
|
return str_replace(array('<![CDATA[', ']]>'), '', trim($out[1]));
|
|
}
|
|
else return '';
|
|
}
|
|
|
|
private static function getHtml($oText, $sTag, $sClass='', $sStyle='', $asExtraAttr=array(), $bAutoClose=false, $sInter='')
|
|
{
|
|
$sHtmlAttr = '';
|
|
if($sClass!='')
|
|
{
|
|
$asExtraAttr['class'] = $sClass;
|
|
}
|
|
if($sStyle!='')
|
|
{
|
|
$asExtraAttr['style'] = $sStyle;
|
|
}
|
|
foreach($asExtraAttr as $sAttrName=>$sAttrValue)
|
|
{
|
|
$sHtmlAttr .= ' '.$sAttrName.'="'.$sAttrValue.'"';
|
|
}
|
|
if($bAutoClose)
|
|
{
|
|
$sHtml = self::encapsulate('', "\n".'<'.$sTag.$sHtmlAttr, ' />');
|
|
}
|
|
else
|
|
{
|
|
$sHtml = self::encapsulate($oText, "\n".'<'.$sTag.$sHtmlAttr.'>', '</'.$sTag.'>', $sInter);
|
|
}
|
|
return $sHtml;
|
|
}
|
|
|
|
private static function cleanRss($oText)
|
|
{
|
|
if(!is_array($oText))
|
|
{
|
|
return htmlspecialchars($oText, ENT_QUOTES);
|
|
}
|
|
elseif(count($oText)>0)
|
|
{
|
|
$oTextKeys = array_map(array(self, 'cleanRss'), array_keys($oText));
|
|
$oTextValues = array_map(array(self, 'cleanRss'), array_values($oText));
|
|
return array_combine($oTextKeys, $oTextValues);
|
|
}
|
|
else
|
|
{
|
|
return $oText;
|
|
}
|
|
}
|
|
|
|
private static function encapsulate($oText, $sPre='', $sPost=false, $sInter='')
|
|
{
|
|
if($sPost===false)
|
|
{
|
|
$sPost = $sPre;
|
|
}
|
|
if(is_array($oText))
|
|
{
|
|
$oText = implode($sPost.$sInter.$sPre, $oText);
|
|
}
|
|
return $sPre.$oText.$sPost;
|
|
}
|
|
|
|
private static function rSortTimeMatrix($asMatrix, $sTimeCol)
|
|
{
|
|
$asResult = array();
|
|
foreach($asMatrix as $iRowId=>$asLine) $asKeys[$iRowId] = strtotime($asLine[$sTimeCol]);
|
|
arsort($asKeys);
|
|
foreach($asKeys as $iRowId=>$iTimeStamp) $asResult[$iRowId] = $asMatrix[$iRowId];
|
|
return $asResult;
|
|
}
|
|
}
|