array(
'name' => 'username or Id',
'required' => true
)
));
public function collectData(){
// get content parsed
$html = getSimpleHTMLDOMCached(self::URI . urlencode($this->getInput('username')) . '/posts',
// force language
84600,
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 returnServerError('No results for this query.');
// get title, url, ... there is a lot of intresting stuff in meta
$this->_title = $html->find('meta[property=og:title]', 0)->getAttribute('content');
$this->_url = $html->find('meta[property=og:url]', 0)->getAttribute('content');
// I don't even know where to start with this discusting html...
foreach($html->find('div[jsname=WsjYwc]') as $post){
$item = array();
$item['author'] = $item['fullname'] = $post->find('div div div div a', 0)->innertext;
$item['id'] = $post->find('div div div', 0)->getAttribute('id');
$item['avatar'] = $post->find('div img', 0)->src;
$item['uri'] = self::URI . $post->find('div div div a', 1)->href;
$item['timestamp'] = strtotime(
'+' . preg_replace('/[^0-9A-Za-z]/',
'',
$post->find('div div div a span', 1)->getAttribute('aria-label')));
// hashtag to treat : https://plus.google.com/explore/tag
// $hashtags = array();
// foreach($post->find('a.d-s') as $hashtag){
// $hashtags[trim($hashtag->plaintext)] = self::URI . $hashtag->href;
// }
$item['content'] = '';
// avatar display
$item['content'] .= '
';
$content = $post->find('div div[id^=body] div div', 0);
// extract plaintext
$item['content_simple'] = $content->plaintext;
$item['title'] = substr($item['content_simple'], 0, 72) . '...';
// 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::URI . $link->href;
}
}
$content = $content->innertext;
$item['content'] .= '