2013-05-01 15:04:57 +02:00
|
|
|
<?php
|
|
|
|
class FeedParser {
|
|
|
|
private $doc;
|
|
|
|
private $error;
|
|
|
|
private $items;
|
|
|
|
private $link;
|
|
|
|
private $title;
|
|
|
|
private $type;
|
2013-05-01 17:08:04 +02:00
|
|
|
private $xpath;
|
2013-05-01 15:04:57 +02:00
|
|
|
|
|
|
|
const FEED_RDF = 0;
|
|
|
|
const FEED_RSS = 1;
|
|
|
|
const FEED_ATOM = 2;
|
|
|
|
|
|
|
|
function __construct($data) {
|
|
|
|
libxml_use_internal_errors(true);
|
|
|
|
libxml_clear_errors();
|
|
|
|
$this->doc = new DOMDocument();
|
|
|
|
$this->doc->loadXML($data);
|
|
|
|
$this->error = $this->format_error(libxml_get_last_error());
|
|
|
|
libxml_clear_errors();
|
|
|
|
|
|
|
|
$this->items = array();
|
|
|
|
}
|
|
|
|
|
|
|
|
function init() {
|
|
|
|
$root = $this->doc->firstChild;
|
2013-05-01 15:38:16 +02:00
|
|
|
$xpath = new DOMXPath($this->doc);
|
|
|
|
$xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
|
2013-05-01 17:08:04 +02:00
|
|
|
$this->xpath = $xpath;
|
2013-05-01 15:38:16 +02:00
|
|
|
|
|
|
|
$root = $xpath->query("(//atom:feed|//channel)")->item(0);
|
2013-05-01 15:04:57 +02:00
|
|
|
|
|
|
|
if ($root) {
|
|
|
|
switch ($root->tagName) {
|
2013-05-01 15:38:16 +02:00
|
|
|
case "channel":
|
2013-05-01 15:04:57 +02:00
|
|
|
$this->type = $this::FEED_RSS;
|
|
|
|
break;
|
|
|
|
case "feed":
|
|
|
|
$this->type = $this::FEED_ATOM;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
$this->error = "Unknown/unsupported feed type";
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
switch ($this->type) {
|
|
|
|
case $this::FEED_ATOM:
|
|
|
|
|
|
|
|
$title = $xpath->query("//atom:feed/atom:title")->item(0);
|
|
|
|
|
|
|
|
if ($title) {
|
|
|
|
$this->title = $title->nodeValue;
|
|
|
|
}
|
|
|
|
|
|
|
|
$link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0);
|
|
|
|
|
|
|
|
if ($link && $link->hasAttributes()) {
|
|
|
|
$this->link = $link->getAttribute("href");
|
|
|
|
}
|
|
|
|
|
|
|
|
$articles = $xpath->query("//atom:entry");
|
|
|
|
|
|
|
|
foreach ($articles as $article) {
|
|
|
|
array_push($this->items, new FeedItem_Atom($article));
|
|
|
|
}
|
|
|
|
|
|
|
|
break;
|
|
|
|
case $this::FEED_RSS:
|
2013-05-01 15:38:16 +02:00
|
|
|
|
|
|
|
$title = $xpath->query("//channel/title")->item(0);
|
|
|
|
|
|
|
|
if ($title) {
|
|
|
|
$this->title = $title->nodeValue;
|
|
|
|
}
|
|
|
|
|
|
|
|
$link = $xpath->query("//channel/link")->item(0);
|
|
|
|
|
|
|
|
if ($link && $link->hasAttributes()) {
|
|
|
|
$this->link = $link->getAttribute("href");
|
|
|
|
}
|
|
|
|
|
|
|
|
$articles = $xpath->query("//channel/item");
|
|
|
|
|
|
|
|
foreach ($articles as $article) {
|
|
|
|
array_push($this->items, new FeedItem_RSS($article));
|
|
|
|
}
|
|
|
|
|
2013-05-01 15:04:57 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
function format_error($error) {
|
|
|
|
if ($error) {
|
|
|
|
return sprintf("LibXML error %s at line %d (column %d): %s",
|
|
|
|
$error->code, $error->line, $error->column,
|
|
|
|
$error->message);
|
|
|
|
} else {
|
|
|
|
return "";
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
function error() {
|
|
|
|
return $this->error;
|
|
|
|
}
|
|
|
|
|
|
|
|
function get_link() {
|
|
|
|
return $this->link;
|
|
|
|
}
|
|
|
|
|
|
|
|
function get_title() {
|
|
|
|
return $this->title;
|
|
|
|
}
|
|
|
|
|
|
|
|
function get_items() {
|
|
|
|
return $this->items;
|
|
|
|
}
|
|
|
|
|
2013-05-01 17:08:04 +02:00
|
|
|
function get_links($rel) {
|
|
|
|
$rv = array();
|
|
|
|
|
|
|
|
switch ($this->type) {
|
|
|
|
case $this::FEED_ATOM:
|
|
|
|
$links = $this->xpath->query("//atom:feed/atom:link");
|
|
|
|
|
|
|
|
foreach ($links as $link) {
|
|
|
|
if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
|
|
|
|
array_push($rv, $link->getAttribute('href'));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case $this::FEED_RSS:
|
|
|
|
$links = $this->xpath->query("//channel/link");
|
2013-05-01 17:09:07 +02:00
|
|
|
foreach ($links as $link) {
|
|
|
|
if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
|
|
|
|
array_push($rv, $link->getAttribute('href'));
|
|
|
|
}
|
2013-05-01 17:08:04 +02:00
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
return $rv;
|
|
|
|
}
|
2013-05-01 15:04:57 +02:00
|
|
|
} ?>
|