rss-bridge/bridges/YoutubeBridge.php

177 lines
5.9 KiB
PHP
Raw Normal View History

2013-08-15 12:05:58 +02:00
<?php
/**
* RssBridgeYoutube
2013-08-15 12:05:58 +02:00
* Returns the newest videos
* WARNING: to parse big playlists (over ~90 videos), you need to edit simple_html_dom.php:
2014-05-14 14:34:06 +02:00
* change: define('MAX_FILE_SIZE', 600000);
* into: define('MAX_FILE_SIZE', 900000); (or more)
2013-08-15 12:05:58 +02:00
*/
class YoutubeBridge extends BridgeAbstract {
public function loadMetadatas() {
$this->name = 'YouTube Bridge';
2016-08-11 21:42:49 +02:00
$this->uri = 'https://www.youtube.com/';
$this->description = 'Returns the 10 newest videos by username/channel/playlist or search';
$this->maintainer = 'mitsukarenai';
$this->update = '2016-08-17';
$this->parameters['By username'] =
'[
{
"type" : "text",
"identifier" : "u",
"name" : "username",
"exampleValue" : "test",
"required" : true
}
]';
$this->parameters['By channel id'] =
'[
{
"type" : "text",
"identifier" : "c",
"name" : "channel id",
"exampleValue" : "15",
"required" : true
}
]';
$this->parameters['By playlist Id'] =
'[
{
"type" : "text",
"identifier" : "p",
"name" : "playlist id",
"exampleValue" : "15"
}
]';
$this->parameters['Search result'] =
'[
{
"type" : "text",
"identifier" : "s",
"name" : "search keyword",
"exampleValue" : "test"
},
{
"type" : "number",
"identifier" : "pa",
"name" : "page",
"exampleValue" : "1"
}
]';
}
private function ytBridgeQueryVideoInfo($vid, &$author, &$desc, &$time) {
$html = $this->getSimpleHTMLDOM($this->uri."watch?v=$vid");
$author = $html->innertext;
$author = substr($author, strpos($author, '"author=') + 8);
$author = substr($author, 0, strpos($author, '\u0026'));
$desc = $html->find('div#watch-description-text', 0)->innertext;
$time = strtotime($html->find('meta[itemprop=datePublished]', 0)->getAttribute('content'));
}
private function ytBridgeAddItem($vid, $title, $author, $desc, $time) {
$item = new \Item();
$item->id = $vid;
$item->title = $title;
$item->author = $author;
$item->timestamp = $time;
$item->uri = $this->uri.'watch?v='.$vid;
$thumbnailUri = str_replace('/www.', '/img.', $this->uri).'vi/'.$vid.'/0.jpg';
$item->content = '<a href="'.$item->uri.'"><img src="'.$thumbnailUri.'" /></a><br />'.$desc;
$this->items[] = $item;
}
private function ytBridgeParseXmlFeed($xml) {
foreach ($xml->find('entry') as $element) {
$title = $this->ytBridgeFixTitle($element->find('title',0)->plaintext);
$author = $element->find('name', 0)->plaintext;
$desc = $element->find('media:description', 0)->innertext;
$vid = str_replace('yt:video:', '', $element->find('id', 0)->plaintext);
$time = strtotime($element->find('published', 0)->plaintext);
$this->ytBridgeAddItem($vid, $title, $author, $desc, $time);
}
$this->request = $this->ytBridgeFixTitle($xml->find('feed > title', 0)->plaintext);
}
2014-05-14 12:39:12 +02:00
private function ytBridgeParseHtmlListing($html, $element_selector, $title_selector) {
$limit = 10; $count = 0;
foreach ($html->find($element_selector) as $element) {
if ($count < $limit) {
$author = ''; $desc = ''; $time = 0;
$vid = str_replace('/watch?v=', '', $element->find('a', 0)->href);
$title = $this->ytBridgeFixTitle($element->find($title_selector, 0)->plaintext);
if ($title != '[Private Video]') {
$this->ytBridgeQueryVideoInfo($vid, $author, $desc, $time);
$this->ytBridgeAddItem($vid, $title, $author, $desc, $time);
$count++;
}
2014-06-20 17:00:36 +02:00
}
}
}
private function ytBridgeFixTitle($title) {
// convert both &#1234; and &quot; to UTF-8
return html_entity_decode($title,ENT_QUOTES,'UTF-8');
}
public function collectData(array $param) {
$xml = '';
$html = '';
$url_feed = '';
$url_listing = '';
2014-05-14 12:39:12 +02:00
if (isset($param['u'])) { /* User and Channel modes */
$this->request = $param['u'];
$url_feed = $this->uri.'feeds/videos.xml?user='.urlencode($this->request);
$url_listing = $this->uri.'user/'.urlencode($this->request).'/videos';
} else if (isset($param['c'])) {
$this->request = $param['c'];
$url_feed = $this->uri.'feeds/videos.xml?channel_id='.urlencode($this->request);
$url_listing = $this->uri.'channel/'.urlencode($this->request).'/videos';
}
if (!empty($url_feed) && !empty($url_listing)) {
if ($xml = $this->getSimpleHTMLDOM($url_feed)) {
$this->ytBridgeParseXmlFeed($xml);
} else if ($html = $this->getSimpleHTMLDOM($url_listing)) {
$this->ytBridgeParseHtmlListing($html, 'li.channels-content-item', 'h3');
} else $this->returnServerError("Could not request YouTube. Tried:\n - $url_feed\n - $url_listing");
}
else if (isset($param['p'])) { /* playlist mode */
2014-06-20 17:00:36 +02:00
$this->request = $param['p'];
$url_listing = $this->uri.'playlist?list='.urlencode($this->request);
$html = $this->getSimpleHTMLDOM($url_listing) or $this->returnServerError("Could not request YouTube. Tried:\n - $url_listing");
$this->ytBridgeParseHtmlListing($html, 'tr.pl-video', '.pl-video-title a');
$this->request = 'Playlist: '.str_replace(' - YouTube', '', $html->find('title', 0)->plaintext);
2014-06-20 17:04:27 +02:00
}
2013-08-15 12:05:58 +02:00
else if (isset($param['s'])) { /* search mode */
$this->request = $param['s']; $page = 1; if (isset($param['pa'])) $page = (int)preg_replace("/[^0-9]/",'', $param['pa']);
$url_listing = $this->uri.'results?search_query='.urlencode($this->request).'&page='.$page.'&filters=video&search_sort=video_date_uploaded';
$html = $this->getSimpleHTMLDOM($url_listing) or $this->returnServerError("Could not request YouTube. Tried:\n - $url_listing");
$this->ytBridgeParseHtmlListing($html, 'div.yt-lockup', 'h3');
$this->request = 'Search: '.str_replace(' - YouTube', '', $html->find('title', 0)->plaintext);
2014-06-20 17:00:36 +02:00
}
else { /* no valid mode */
$this->returnClientError("You must either specify either:\n - YouTube username (?u=...)\n - Channel id (?c=...)\n - Playlist id (?p=...)\n - Search (?s=...)");
}
}
2014-06-20 17:00:36 +02:00
public function getName(){
return (!empty($this->request) ? $this->request .' - ' : '') .'YouTube Bridge';
2014-06-20 17:00:36 +02:00
}
2013-08-15 12:05:58 +02:00
2014-06-20 17:00:36 +02:00
public function getCacheDuration(){
return 10800; // 3 hours
}
2013-08-15 12:05:58 +02:00
}