rss-bridge/bridges/GooglePlusPostBridge.php

132 lines
4.1 KiB
PHP
Raw Normal View History

2014-07-20 10:40:00 +02:00
<?php
class GooglePlusPostBridge extends BridgeAbstract
{
2014-07-20 16:28:51 +02:00
protected $_title;
protected $_url;
public function loadMetadatas() {
$this->maintainer = "Grummfy";
$this->name = "Google Plus Post Bridge";
$this->uri = "http://plus.google.com/";
$this->description = "Returns user public post (without API).";
$this->update = "2014-07-20";
$this->parameters[] =
'[
{
"name" : "username or Id",
"identifier" : "username"
}
]';
}
2014-07-20 16:28:51 +02:00
const GOOGLE_PLUS_BASE_URL = 'https://plus.google.com/';
2014-07-20 10:40:00 +02:00
public function collectData(array $param)
{
if (!isset($param['username']))
{
$this->returnError('You must specify a username (?username=...).', 400);
}
$this->request = $param['username'];
2014-07-20 16:28:51 +02:00
// get content parsed
// $html = file_get_html(__DIR__ . '/../posts2.html'
$html = file_get_html(self::GOOGLE_PLUS_BASE_URL . urlencode($this->request) . '/posts'
// force language
, false, stream_context_create(array('http'=> array(
'header' => 'Accept-Language: fr,fr-be,fr-fr;q=0.8,en;q=0.4,en-us;q=0.2;*' . "\r\n"
)))
) OR $this->returnError('No results for this query.', 404);
2014-07-20 10:40:00 +02:00
2014-07-20 16:28:51 +02:00
// get title, url, ... there is a lot of intresting stuff in meta
$this->_title = $html->find('meta[property]', 0)->getAttribute('content');
$this->_url = $html->find('meta[itemprop=url]', 0)->getAttribute('content');
// foreach ($html->find('meta') as $e)
// {
// $item = new \Item();
// $item->content = var_export($e->attr, true);
// $this->items[] = $item;
// }
// div[jsmodel=XNmfOc]
foreach($html->find('div.yt') as $post)
2014-07-20 10:40:00 +02:00
{
$item = new \Item();
2014-07-20 16:28:51 +02:00
// $item->content = $post->find('div.Al', 0)->innertext;
2014-07-20 10:40:00 +02:00
$item->username = $item->fullname = $post->find('header.lea h3 a', 0)->innertext;
$item->id = $post->getAttribute('id');
2014-07-20 16:28:51 +02:00
// $item->title = $item->fullname = $post->find('header.lea', 0)->plaintext;
$item->avatar = $post->find('div.ys img', 0)->src;
// var_dump((($post->find('a.o-U-s', 0)->getAllAttributes())));
$item->uri = self::GOOGLE_PLUS_BASE_URL . $post->find('a.o-U-s', 0)->href;
2014-07-20 16:28:51 +02:00
$item->timestamp = strtotime($post->find('a.o-U-s', 0)->plaintext);
2014-07-20 10:40:00 +02:00
$this->items[] = $item;
2014-07-20 16:28:51 +02:00
// hashtag to treat : https://plus.google.com/explore/tag
$hashtags = array();
foreach($post->find('a.d-s') as $hashtag)
{
$hashtags[ trim($hashtag->plaintext) ] = self::GOOGLE_PLUS_BASE_URL . $hashtag->href;
}
$item->content = '';
// avatar display
$item->content .= '<div style="float:left; margin: 0 0.5em 0.5em 0;"><a href="' . self::GOOGLE_PLUS_BASE_URL . urlencode($this->request);
$item->content .= '"><img align="top" alt="avatar" src="' . $item->avatar.'" />' . $item->username . '</a></div>';
$content = $post->find('div.Al', 0);
// alter link
// $content = $content->innertext;
// $content = str_replace('href="./', 'href="' . self::GOOGLE_PLUS_BASE_URL, $content);
// $content = str_replace('href="photos', 'href="' . self::GOOGLE_PLUS_BASE_URL . 'photos', $content);
// XXX ugly but I don't have any idea how to do a better stuff, str_replace on link doesn't work as expected and ask too many checks
foreach($content->find('a') as $link)
{
$hasHttp = strpos($link->href, 'http');
$hasDoubleSlash = strpos($link->href, '//');
if ((!$hasHttp && !$hasDoubleSlash)
|| (false !== $hasHttp && strpos($link->href, 'http') != 0)
|| (false === $hasHttp && false !== $hasDoubleSlash && $hasDoubleSlash != 0))
{
// skipp bad link, for some hashtag or other stuff
if (strpos($link->href, '/') == 0)
{
$link->href = substr($link->href, 1);
}
$link->href = self::GOOGLE_PLUS_BASE_URL . $link->href;
}
}
$content = $content->innertext;
$item->content .= '<div style="margin-top: -1.5em">' . $content . '</div>';
// extract plaintext
$item->content_simple = $post->find('div.Al', 0)->plaintext;
2014-07-20 10:40:00 +02:00
}
2014-07-20 16:28:51 +02:00
// $html->save(__DIR__ . '/../posts2.html');
2014-07-20 10:40:00 +02:00
}
public function getName()
{
2014-07-20 16:28:51 +02:00
return $this->_title ?: 'Google Plus Post Bridge';
2014-07-20 10:40:00 +02:00
}
public function getURI()
{
2014-07-20 16:28:51 +02:00
return $this->_url ?: 'http://plus.google.com/';
2014-07-20 10:40:00 +02:00
}
public function getCacheDuration()
{
2014-07-20 12:38:45 +02:00
return 1; // 600; // 10 minutes
2014-07-20 10:40:00 +02:00
}
}