ttrss/classes/feedparser.php

219 lines
5.0 KiB
PHP
Raw Normal View History

2013-05-01 13:04:57 +00:00
<?php
class FeedParser {
private $doc;
private $error;
private $items;
private $link;
private $title;
private $type;
2013-05-01 15:08:04 +00:00
private $xpath;
2013-05-01 13:04:57 +00:00
const FEED_RDF = 0;
const FEED_RSS = 1;
const FEED_ATOM = 2;
function __construct($data) {
libxml_use_internal_errors(true);
libxml_clear_errors();
$this->doc = new DOMDocument();
$this->doc->loadXML($data);
2013-05-17 16:09:43 +00:00
$error = libxml_get_last_error();
if ($error && $error->code == 9) {
libxml_clear_errors();
// we might want to try guessing input encoding here too
$data = iconv("UTF-8", "UTF-8//IGNORE", $data);
$this->doc = new DOMDocument();
$this->doc->loadXML($data);
$error = libxml_get_last_error();
}
$this->error = $this->format_error($error);
2013-05-01 13:04:57 +00:00
libxml_clear_errors();
$this->items = array();
}
function init() {
$root = $this->doc->firstChild;
2013-05-01 13:38:16 +00:00
$xpath = new DOMXPath($this->doc);
$xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
2013-05-15 05:09:25 +00:00
$xpath->registerNamespace('atom03', 'http://purl.org/atom/ns#');
$xpath->registerNamespace('media', 'http://search.yahoo.com/mrss/');
2013-05-01 16:30:52 +00:00
$xpath->registerNamespace('rdf', 'http://www.w3.org/1999/02/22-rdf-syntax-ns#');
$xpath->registerNamespace('slash', 'http://purl.org/rss/1.0/modules/slash/');
$xpath->registerNamespace('dc', 'http://purl.org/dc/elements/1.1/');
2013-05-01 18:05:59 +00:00
$xpath->registerNamespace('content', 'http://purl.org/rss/1.0/modules/content/');
2013-05-01 15:08:04 +00:00
$this->xpath = $xpath;
2013-05-01 13:38:16 +00:00
$root = $xpath->query("(//atom03:feed|//atom:feed|//channel|//rdf:rdf|//rdf:RDF)");
2013-05-01 13:04:57 +00:00
if ($root) {
$root = $root->item(0);
if ($root) {
switch (mb_strtolower($root->tagName)) {
case "rdf:rdf":
$this->type = $this::FEED_RDF;
break;
case "channel":
$this->type = $this::FEED_RSS;
break;
case "feed":
$this->type = $this::FEED_ATOM;
break;
default:
if( !isset($this->error) ){
$this->error = "Unknown/unsupported feed type";
}
return;
}
2013-05-01 13:04:57 +00:00
}
switch ($this->type) {
case $this::FEED_ATOM:
$title = $xpath->query("//atom:feed/atom:title")->item(0);
2013-05-15 05:09:25 +00:00
if (!$title)
$title = $xpath->query("//atom03:feed/atom03:title")->item(0);
2013-05-01 13:04:57 +00:00
if ($title) {
$this->title = $title->nodeValue;
}
$link = $xpath->query("//atom:feed/atom:link[not(@rel)]")->item(0);
2013-05-15 05:09:25 +00:00
if (!$link)
$link = $xpath->query("//atom03:feed/atom03:link[not(@rel)]")->item(0);
2013-05-01 13:04:57 +00:00
if ($link && $link->hasAttributes()) {
$this->link = $link->getAttribute("href");
}
$articles = $xpath->query("//atom:entry");
2013-05-15 05:09:25 +00:00
if (!$articles || $articles->length == 0)
$articles = $xpath->query("//atom03:entry");
2013-05-01 13:04:57 +00:00
foreach ($articles as $article) {
2013-07-09 14:28:15 +00:00
array_push($this->items, new FeedItem_Atom($article, $this->doc, $this->xpath));
2013-05-01 13:04:57 +00:00
}
break;
case $this::FEED_RSS:
2013-05-01 13:38:16 +00:00
$title = $xpath->query("//channel/title")->item(0);
if ($title) {
$this->title = $title->nodeValue;
}
$link = $xpath->query("//channel/link")->item(0);
if ($link) {
if ($link->getAttribute("href"))
$this->link = $link->getAttribute("href");
else if ($link->nodeValue)
$this->link = $link->nodeValue;
2013-05-01 13:38:16 +00:00
}
$articles = $xpath->query("//channel/item");
foreach ($articles as $article) {
array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
2013-05-01 13:38:16 +00:00
}
2013-05-01 13:04:57 +00:00
break;
2013-05-01 16:30:52 +00:00
case $this::FEED_RDF:
$xpath->registerNamespace('rssfake', 'http://purl.org/rss/1.0/');
$title = $xpath->query("//rssfake:channel/rssfake:title")->item(0);
if ($title) {
$this->title = $title->nodeValue;
}
$link = $xpath->query("//rssfake:channel/rssfake:link")->item(0);
if ($link) {
$this->link = $link->nodeValue;
}
$articles = $xpath->query("//rssfake:item");
foreach ($articles as $article) {
array_push($this->items, new FeedItem_RSS($article, $this->doc, $this->xpath));
}
break;
2013-05-01 13:04:57 +00:00
}
2013-05-01 16:30:52 +00:00
} else {
if( !isset($this->error) ){
$this->error = "Unknown/unsupported feed type";
}
2013-05-01 16:30:52 +00:00
return;
2013-05-01 13:04:57 +00:00
}
}
function format_error($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 $this->items;
}
2013-05-01 15:08:04 +00:00
function get_links($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:
2013-05-24 10:40:47 +00:00
$links = $this->xpath->query("//atom:link");
2013-05-01 15:09:07 +00:00
foreach ($links as $link) {
if (!$rel || $link->hasAttribute('rel') && $link->getAttribute('rel') == $rel) {
array_push($rv, $link->getAttribute('href'));
}
2013-05-01 15:08:04 +00:00
}
break;
}
return $rv;
}
2013-05-01 13:04:57 +00:00
} ?>