full-text-rss/libraries/feedwriter/FeedWriter.php

575 lines
16 KiB
PHP
Raw Normal View History

2011-01-11 18:06:12 +00:00
<?php
2017-02-18 15:06:19 +00:00
define('RSS2', 1);
define('JSON', 2);
define('JSONP', 3);
2011-01-11 18:06:12 +00:00
/**
* Univarsel Feed Writer class
*
2014-09-15 20:24:06 +00:00
* Generate RSS2 or JSON (original: RSS 1.0, RSS2.0 and ATOM Feed)
2013-04-18 14:11:06 +00:00
*
2011-03-23 22:39:01 +00:00
* Modified for FiveFilters.org's Full-Text RSS project
2013-04-18 14:11:06 +00:00
* to allow for inclusion of hubs, JSON output.
* Stripped RSS1 and ATOM support.
2011-01-11 18:06:12 +00:00
*
* @package UnivarselFeedWriter
* @author Anis uddin Ahmad <anisniit@gmail.com>
* @link http://www.ajaxray.com/projects/rss
*/
class FeedWriter
{
2011-03-23 22:39:01 +00:00
private $self = null; // self URL - http://feed2.w3.org/docs/warning/MissingAtomSelfLink.html
2015-06-14 00:03:20 +00:00
private $alternate = array(); // alternate URL and title
private $related = array(); // related URL and title
2011-03-23 22:39:01 +00:00
private $hubs = array(); // PubSubHubbub hubs
2011-01-11 18:06:12 +00:00
private $channels = array(); // Collection of channel elements
private $items = array(); // Collection of items as object of FeedItem class.
private $data = array(); // Store some other version wise data
private $CDATAEncoding = array(); // The tag names which have to encoded as CDATA
2011-03-23 22:39:01 +00:00
private $xsl = null; // stylesheet to render RSS (used by Chrome)
2013-04-18 14:11:06 +00:00
private $json = null; // JSON object
2014-09-15 20:24:06 +00:00
private $simplejson = false;
2011-01-11 18:06:12 +00:00
private $version = null;
/**
* Constructor
*
2013-04-18 14:11:06 +00:00
* @param constant the version constant (RSS2 or JSON).
2011-01-11 18:06:12 +00:00
*/
function __construct($version = RSS2)
{
$this->version = $version;
// Setting default value for assential channel elements
$this->channels['title'] = $version . ' Feed';
$this->channels['link'] = 'http://www.ajaxray.com/blog';
//Tag names to encode in CDATA
$this->CDATAEncoding = array('description', 'content:encoded', 'content', 'subtitle', 'summary');
}
2013-04-18 14:11:06 +00:00
public function setFormat($format) {
$this->version = $format;
}
2011-01-11 18:06:12 +00:00
// Start # public functions ---------------------------------------------
2014-09-15 20:24:06 +00:00
public function enableSimpleJson($enable=true) {
$this->simplejson = $enable;
}
2011-01-11 18:06:12 +00:00
/**
* Set a channel element
* @access public
* @param srting name of the channel tag
* @param string content of the channel tag
* @return void
*/
public function setChannelElement($elementName, $content)
{
$this->channels[$elementName] = $content ;
}
/**
* Set multiple channel elements from an array. Array elements
* should be 'channelName' => 'channelContent' format.
*
* @access public
* @param array array of channels
* @return void
*/
public function setChannelElementsFromArray($elementArray)
{
2019-04-04 21:23:27 +00:00
if(!is_array($elementArray)) return;
2011-01-11 18:06:12 +00:00
foreach ($elementArray as $elementName => $content)
{
$this->setChannelElement($elementName, $content);
}
}
/**
2014-09-15 20:24:06 +00:00
* Generate the actual RSS/JSON file
2011-01-11 18:06:12 +00:00
*
* @access public
* @return void
*/
2014-09-15 20:24:06 +00:00
public function generateFeed()
2011-01-11 18:06:12 +00:00
{
2013-04-18 14:11:06 +00:00
if ($this->version == RSS2) {
header('Content-type: text/xml; charset=UTF-8');
2014-05-15 20:49:16 +00:00
// this line prevents Chrome 20 from prompting download
// used by Google: https://news.google.com/news/feeds?ned=us&topic=b&output=rss
header('X-content-type-options: nosniff');
2013-04-18 14:11:06 +00:00
} elseif ($this->version == JSON) {
header('Content-type: application/json; charset=UTF-8');
$this->json = new stdClass();
2014-05-15 20:49:16 +00:00
} elseif ($this->version == JSONP) {
header('Content-type: application/javascript; charset=UTF-8');
$this->json = new stdClass();
2013-04-18 14:11:06 +00:00
}
2011-01-11 18:06:12 +00:00
$this->printHead();
$this->printChannels();
$this->printItems();
$this->printTale();
2014-05-15 20:49:16 +00:00
if ($this->version == JSON || $this->version == JSONP) {
2014-09-15 20:24:06 +00:00
if (!$this->simplejson) {
echo json_encode($this->json);
} else {
$simplejson = new stdClass();
if (is_array($this->json->rss['channel']->item)) {
// get first item
$jsonitem = $this->json->rss['channel']->item[0];
} else {
$jsonitem = $this->json->rss['channel']->item;
}
// defaults
$simplejson->title = null;
$simplejson->excerpt = null;
$simplejson->date = null;
$simplejson->author = null;
$simplejson->language = null;
$simplejson->url = null;
$simplejson->effective_url = null;
2019-04-04 21:23:27 +00:00
$simplejson->domain = null;
$simplejson->word_count = null;
2017-02-18 15:06:19 +00:00
$simplejson->og_url = null;
$simplejson->og_title = null;
$simplejson->og_description = null;
$simplejson->og_image = null;
$simplejson->og_type = null;
2019-04-04 21:23:27 +00:00
$simplejson->twitter_card = null;
$simplejson->twitter_site = null;
$simplejson->twitter_creator = null;
$simplejson->twitter_image = null;
$simplejson->twitter_title = null;
$simplejson->twitter_description = null;
2014-09-15 20:24:06 +00:00
$simplejson->content = null;
// actual values
$simplejson->url = $jsonitem->link;
$simplejson->effective_url = $jsonitem->dc_identifier;
2019-04-04 21:23:27 +00:00
$simplejson->domain = strtolower(@parse_url($simplejson->effective_url, PHP_URL_HOST));
if (substr($simplejson->domain, 0, 4) === 'www.') {
$simplejson->domain = substr($simplejson->domain, 4);
}
2014-09-15 20:24:06 +00:00
if (isset($jsonitem->title)) $simplejson->title = $jsonitem->title;
if (isset($jsonitem->dc_language)) $simplejson->language = $jsonitem->dc_language;
if (isset($jsonitem->content_encoded)) {
$simplejson->content = $jsonitem->content_encoded;
2019-04-04 21:23:27 +00:00
// from http://php.net/manual/en/function.str-word-count.php#107363
$simplejson->word_count = count(preg_split('!\s+!', strip_tags($simplejson->content), -1, PREG_SPLIT_NO_EMPTY));
2014-09-15 20:24:06 +00:00
if (isset($jsonitem->description)) {
$simplejson->excerpt = $jsonitem->description;
}
} else {
$simplejson->content = $jsonitem->description;
}
if (isset($jsonitem->dc_creator)) {
$simplejson->author = $jsonitem->dc_creator;
}
if (isset($jsonitem->pubDate)) {
$simplejson->date = gmdate(DATE_ATOM, strtotime($jsonitem->pubDate));
}
2017-02-18 15:06:19 +00:00
if (isset($jsonitem->og_url)) $simplejson->og_url = $jsonitem->og_url;
if (isset($jsonitem->og_title)) $simplejson->og_title = $jsonitem->og_title;
if (isset($jsonitem->og_description)) $simplejson->og_description = $jsonitem->og_description;
if (isset($jsonitem->og_image)) $simplejson->og_image = $jsonitem->og_image;
if (isset($jsonitem->og_type)) $simplejson->og_type = $jsonitem->og_type;
2019-04-04 21:23:27 +00:00
if (isset($jsonitem->twitter_card)) $simplejson->twitter_card = $jsonitem->twitter_card;
if (isset($jsonitem->twitter_site)) $simplejson->twitter_site = $jsonitem->twitter_site;
if (isset($jsonitem->twitter_creator)) $simplejson->twitter_creator = $jsonitem->twitter_creator;
if (isset($jsonitem->twitter_image)) $simplejson->twitter_image = $jsonitem->twitter_image;
if (isset($jsonitem->twitter_title)) $simplejson->twitter_title = $jsonitem->twitter_title;
if (isset($jsonitem->twitter_description)) $simplejson->twitter_description = $jsonitem->twitter_description;
2014-09-15 20:24:06 +00:00
echo json_encode($simplejson);
}
2013-04-18 14:11:06 +00:00
}
2011-01-11 18:06:12 +00:00
}
2014-05-15 21:03:31 +00:00
public function &getItems()
{
return $this->items;
}
2011-01-11 18:06:12 +00:00
/**
* Create a new FeedItem.
*
* @access public
* @return object instance of FeedItem class
*/
public function createNewItem()
{
$Item = new FeedItem($this->version);
return $Item;
}
/**
* Add a FeedItem to the main class
*
* @access public
* @param object instance of FeedItem class
* @return void
*/
public function addItem($feedItem)
{
$this->items[] = $feedItem;
}
// Wrapper functions -------------------------------------------------------------------
/**
* Set the 'title' channel element
*
* @access public
* @param srting value of 'title' channel tag
* @return void
*/
public function setTitle($title)
{
$this->setChannelElement('title', $title);
}
2011-03-23 22:39:01 +00:00
/**
* Add a hub to the channel element
*
* @access public
* @param string URL
* @return void
*/
public function addHub($hub)
{
$this->hubs[] = $hub;
}
/**
* Set XSL URL
*
* @access public
* @param string URL
* @return void
*/
public function setXsl($xsl)
{
$this->xsl = $xsl;
2014-09-15 20:24:06 +00:00
}
/**
* Set TTL
*
* @access public
* @param int time to live (minutes)
* @return void
*/
public function setTtl($ttl)
{
$this->setChannelElement('ttl', (int)$ttl);
}
2011-03-23 22:39:01 +00:00
/**
* Set self URL
*
* @access public
* @param string URL
* @return void
*/
2015-06-14 00:03:20 +00:00
public function setSelf($url)
2011-03-23 22:39:01 +00:00
{
2015-06-14 00:03:20 +00:00
$this->self = $url;
}
/**
* Set alternate URL
*
* @access public
* @param string URL
* @param string title
* @return void
*/
public function setAlternate($url, $title)
{
$this->alternate = array('url'=>$url, 'title'=>$title);
}
/**
* Set related URL
*
* @access public
* @param string URL
* @param string title
* @return void
*/
public function setRelated($url, $title)
{
$this->related = array('url'=>$url, 'title'=>$title);
}
2011-03-23 22:39:01 +00:00
2011-01-11 18:06:12 +00:00
/**
* Set the 'description' channel element
*
* @access public
* @param srting value of 'description' channel tag
* @return void
*/
2014-09-15 20:24:06 +00:00
public function setDescription($description)
{
$this->setChannelElement('description', $description);
2011-01-11 18:06:12 +00:00
}
/**
* Set the 'link' channel element
*
* @access public
* @param srting value of 'link' channel tag
* @return void
*/
public function setLink($link)
{
$this->setChannelElement('link', $link);
}
/**
* Set the 'image' channel element
*
* @access public
* @param srting title of image
* @param srting link url of the imahe
* @param srting path url of the image
* @return void
*/
public function setImage($title, $link, $url)
{
$this->setChannelElement('image', array('title'=>$title, 'link'=>$link, 'url'=>$url));
}
// End # public functions ----------------------------------------------
// Start # private functions ----------------------------------------------
/**
* Prints the xml and rss namespace
*
* @access private
* @return void
*/
private function printHead()
{
2013-04-18 14:11:06 +00:00
if ($this->version == RSS2)
2011-01-11 18:06:12 +00:00
{
2013-04-18 14:11:06 +00:00
$out = '<?xml version="1.0" encoding="utf-8"?>'."\n";
2011-03-23 22:39:01 +00:00
if ($this->xsl) $out .= '<?xml-stylesheet type="text/xsl" href="'.htmlspecialchars($this->xsl).'"?>' . PHP_EOL;
2019-04-04 21:23:27 +00:00
//$out .= '<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:media="http://search.yahoo.com/mrss/" xmlns:og="http://ogp.me/ns#" xmlns:twitter="https://dev.twitter.com/cards/markup">' . PHP_EOL;
$out .= '<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:media="http://search.yahoo.com/mrss/">' . PHP_EOL;
2013-04-18 14:11:06 +00:00
echo $out;
2011-01-11 18:06:12 +00:00
}
2014-05-15 20:49:16 +00:00
elseif ($this->version == JSON || $this->version == JSONP)
2011-01-11 18:06:12 +00:00
{
2013-04-18 14:11:06 +00:00
$this->json->rss = array('@attributes' => array('version' => '2.0'));
2011-01-11 18:06:12 +00:00
}
}
/**
* Closes the open tags at the end of file
*
* @access private
* @return void
*/
private function printTale()
{
2013-04-18 14:11:06 +00:00
if ($this->version == RSS2)
2011-01-11 18:06:12 +00:00
{
2013-04-18 14:11:06 +00:00
echo '</channel>',PHP_EOL,'</rss>';
2011-01-11 18:06:12 +00:00
}
2013-04-18 14:11:06 +00:00
// do nothing for JSON
2011-01-11 18:06:12 +00:00
}
/**
* Creates a single node as xml format
*
* @access private
2013-04-18 14:11:06 +00:00
* @param string name of the tag
2011-01-11 18:06:12 +00:00
* @param mixed tag value as string or array of nested tags in 'tagName' => 'tagValue' format
* @param array Attributes(if any) in 'attrName' => 'attrValue' format
* @return string formatted xml tag
*/
private function makeNode($tagName, $tagContent, $attributes = null)
{
2013-04-18 14:11:06 +00:00
if ($this->version == RSS2)
2011-01-11 18:06:12 +00:00
{
2013-04-18 14:11:06 +00:00
$nodeText = '';
$attrText = '';
if (is_array($attributes))
2011-01-11 18:06:12 +00:00
{
2013-04-18 14:11:06 +00:00
foreach ($attributes as $key => $value)
{
2017-02-18 15:06:19 +00:00
//$attrText .= " $key=\"".htmlspecialchars($value, ENT_COMPAT, 'UTF-8', false)."\" ";
// TODO: replace HTML entities not supported in XML with UTF8 equivalent characters
$attrText .= " $key=\"".htmlspecialchars($value, ENT_COMPAT, 'UTF-8')."\" ";
2013-04-18 14:11:06 +00:00
}
}
$nodeText .= "<{$tagName}{$attrText}>";
if (is_array($tagContent))
{
foreach ($tagContent as $key => $value)
{
$nodeText .= $this->makeNode($key, $value);
}
2011-01-11 18:06:12 +00:00
}
2013-04-18 14:11:06 +00:00
else
{
//$nodeText .= (in_array($tagName, $this->CDATAEncoding))? $tagContent : htmlentities($tagContent);
2017-02-18 15:06:19 +00:00
//$nodeText .= htmlspecialchars($tagContent, ENT_COMPAT, 'UTF-8', false);
// TODO: replace HTML entities not supported in XML with UTF8 equivalent characters
$nodeText .= htmlspecialchars($tagContent, ENT_COMPAT, 'UTF-8');
2013-04-18 14:11:06 +00:00
}
//$nodeText .= (in_array($tagName, $this->CDATAEncoding))? "]]></$tagName>" : "</$tagName>";
$nodeText .= "</$tagName>";
return $nodeText . PHP_EOL;
2011-01-11 18:06:12 +00:00
}
2014-05-15 20:49:16 +00:00
elseif ($this->version == JSON || $this->version == JSONP)
2011-01-11 18:06:12 +00:00
{
2013-04-18 14:11:06 +00:00
$tagName = (string)$tagName;
$tagName = strtr($tagName, ':', '_');
$node = null;
if (!$tagContent && is_array($attributes) && count($attributes))
2011-01-11 18:06:12 +00:00
{
2013-04-18 14:11:06 +00:00
$node = array('@attributes' => $this->json_keys($attributes));
} else {
if (is_array($tagContent)) {
$node = $this->json_keys($tagContent);
} else {
$node = $tagContent;
}
2011-01-11 18:06:12 +00:00
}
2013-04-18 14:11:06 +00:00
return $node;
2011-01-11 18:06:12 +00:00
}
2013-04-18 14:11:06 +00:00
return ''; // should not get here
}
private function json_keys(array $array) {
$new = array();
foreach ($array as $key => $val) {
if (is_string($key)) $key = strtr($key, ':', '_');
if (is_array($val)) {
$new[$key] = $this->json_keys($val);
} else {
$new[$key] = $val;
}
}
return $new;
2011-01-11 18:06:12 +00:00
}
/**
* @desc Print channels
* @access private
* @return void
*/
private function printChannels()
{
//Start channel tag
2014-05-15 20:49:16 +00:00
if ($this->version == RSS2) {
echo '<channel>' . PHP_EOL;
// add hubs
foreach ($this->hubs as $hub) {
//echo $this->makeNode('link', '', array('rel'=>'hub', 'href'=>$hub, 'xmlns'=>'http://www.w3.org/2005/Atom'));
2015-06-14 00:03:20 +00:00
echo '<atom:link rel="hub" href="'.htmlspecialchars($hub).'" />' . PHP_EOL;
2014-05-15 20:49:16 +00:00
}
// add self
if (isset($this->self)) {
//echo $this->makeNode('link', '', array('rel'=>'self', 'href'=>$this->self, 'xmlns'=>'http://www.w3.org/2005/Atom'));
2015-06-14 00:03:20 +00:00
echo '<atom:link rel="self" href="'.htmlspecialchars($this->self).'" />' . PHP_EOL;
}
// add alternate
if (isset($this->alternate)) {
echo '<atom:link rel="alternate" title="'.htmlspecialchars($this->alternate['title']).'" href="'.htmlspecialchars($this->alternate['url']).'" />' . PHP_EOL;
2014-05-15 20:49:16 +00:00
}
2015-06-14 00:03:20 +00:00
// add related
if (isset($this->related)) {
echo '<atom:link rel="related" title="'.htmlspecialchars($this->related['title']).'" href="'.htmlspecialchars($this->related['url']).'" />' . PHP_EOL;
}
2014-05-15 20:49:16 +00:00
//Print Items of channel
foreach ($this->channels as $key => $value)
{
echo $this->makeNode($key, $value);
}
} elseif ($this->version == JSON || $this->version == JSONP) {
$this->json->rss['channel'] = (object)$this->json_keys($this->channels);
2011-01-11 18:06:12 +00:00
}
}
/**
* Prints formatted feed items
*
* @access private
* @return void
*/
private function printItems()
{
2014-05-15 20:56:02 +00:00
foreach ($this->items as $item) {
$itemElements = $item->getElements();
2011-01-11 18:06:12 +00:00
2013-04-18 14:11:06 +00:00
echo $this->startItem();
2014-05-15 20:49:16 +00:00
if ($this->version == JSON || $this->version == JSONP) {
2013-04-18 14:11:06 +00:00
$json_item = array();
}
2011-01-11 18:06:12 +00:00
2014-05-15 20:56:02 +00:00
foreach ($itemElements as $thisElement) {
foreach ($thisElement as $instance) {
if ($this->version == RSS2) {
2019-04-04 21:23:27 +00:00
// Let's not include twitter and open graph elements in regular RSS output
// These are aimed more at developers, and so JSON is more appropriate
if (preg_match('/^(twitter|og):/i', $instance['name'])) continue;
2014-05-15 20:56:02 +00:00
echo $this->makeNode($instance['name'], $instance['content'], $instance['attributes']);
} elseif ($this->version == JSON || $this->version == JSONP) {
$_json_node = $this->makeNode($instance['name'], $instance['content'], $instance['attributes']);
if (count($thisElement) > 1) {
$json_item[strtr($instance['name'], ':', '_')][] = $_json_node;
} else {
$json_item[strtr($instance['name'], ':', '_')] = $_json_node;
}
}
2013-04-18 14:11:06 +00:00
}
2011-01-11 18:06:12 +00:00
}
echo $this->endItem();
2014-05-15 20:49:16 +00:00
if ($this->version == JSON || $this->version == JSONP) {
2013-04-18 14:11:06 +00:00
if (count($this->items) > 1) {
2014-09-15 20:24:06 +00:00
$this->json->rss['channel']->item[] = (object)$json_item;
2013-04-18 14:11:06 +00:00
} else {
2014-09-15 20:24:06 +00:00
$this->json->rss['channel']->item = (object)$json_item;
2013-04-18 14:11:06 +00:00
}
}
2011-01-11 18:06:12 +00:00
}
}
/**
* Make the starting tag of channels
*
* @access private
* @return void
*/
2013-04-18 14:11:06 +00:00
private function startItem()
2011-01-11 18:06:12 +00:00
{
2013-04-18 14:11:06 +00:00
if ($this->version == RSS2)
2011-01-11 18:06:12 +00:00
{
echo '<item>' . PHP_EOL;
}
2013-04-18 14:11:06 +00:00
// nothing for JSON
2011-01-11 18:06:12 +00:00
}
/**
* Closes feed item tag
*
* @access private
* @return void
*/
private function endItem()
{
2013-04-18 14:11:06 +00:00
if ($this->version == RSS2)
2011-01-11 18:06:12 +00:00
{
echo '</item>' . PHP_EOL;
}
2013-04-18 14:11:06 +00:00
// nothing for JSON
2011-01-11 18:06:12 +00:00
}
// End # private functions ----------------------------------------------
2013-04-18 14:11:06 +00:00
}