2010-07-23 01:41:45 -04:00
|
|
|
<?php
|
2016-11-27 17:41:55 -05:00
|
|
|
|
2017-01-09 07:14:25 -05:00
|
|
|
/**
|
2015-12-25 17:17:34 -05:00
|
|
|
* @file mod/parse_url.php
|
2016-11-25 10:59:31 -05:00
|
|
|
* @brief The parse_url module
|
2017-01-09 07:14:25 -05:00
|
|
|
*
|
2017-01-26 22:57:53 -05:00
|
|
|
* This module does parse an url for embeddable content (audio, video, image files or link)
|
|
|
|
* information and does format this information to BBCode
|
2017-01-09 07:14:25 -05:00
|
|
|
*
|
|
|
|
* @see ParseUrl::getSiteinfo() for more information about scraping embeddable content
|
2018-09-02 17:24:56 -04:00
|
|
|
*/
|
2017-04-30 00:07:00 -04:00
|
|
|
use Friendica\App;
|
2018-12-26 01:06:24 -05:00
|
|
|
use Friendica\Core\Hook;
|
2018-10-29 17:20:46 -04:00
|
|
|
use Friendica\Core\Logger;
|
2018-01-26 23:18:38 -05:00
|
|
|
use Friendica\Util\Network;
|
2018-01-04 12:03:15 -05:00
|
|
|
use Friendica\Util\ParseUrl;
|
2012-09-05 19:26:11 -04:00
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
function parse_url_content(App $a)
|
|
|
|
{
|
2016-11-23 19:11:22 -05:00
|
|
|
$text = null;
|
2018-09-02 17:24:56 -04:00
|
|
|
$str_tags = '';
|
2016-01-14 17:59:51 -05:00
|
|
|
|
2017-01-26 22:57:53 -05:00
|
|
|
$br = "\n";
|
2014-02-22 09:46:19 -05:00
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
if (!empty($_GET['binurl'])) {
|
|
|
|
$url = trim(hex2bin($_GET['binurl']));
|
2016-11-23 19:11:22 -05:00
|
|
|
} else {
|
2018-09-02 17:24:56 -04:00
|
|
|
$url = trim($_GET['url']);
|
2016-11-12 15:23:00 -05:00
|
|
|
}
|
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
if (!empty($_GET['title'])) {
|
|
|
|
$title = strip_tags(trim($_GET['title']));
|
2014-07-14 02:37:40 -04:00
|
|
|
}
|
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
if (!empty($_GET['description'])) {
|
|
|
|
$text = strip_tags(trim($_GET['description']));
|
2013-03-02 08:46:06 -05:00
|
|
|
}
|
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
if (!empty($_GET['tags'])) {
|
|
|
|
$arr_tags = ParseUrl::convertTagsToArray($_GET['tags']);
|
2016-11-23 19:11:22 -05:00
|
|
|
if (count($arr_tags)) {
|
2018-09-02 17:24:56 -04:00
|
|
|
$str_tags = $br . implode(' ', $arr_tags) . $br;
|
2015-04-05 14:40:31 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-11-23 19:11:22 -05:00
|
|
|
// Add url scheme if it is missing
|
|
|
|
$arrurl = parse_url($url);
|
2018-11-30 09:06:22 -05:00
|
|
|
if (empty($arrurl['scheme'])) {
|
|
|
|
if (!empty($arrurl['host'])) {
|
2018-09-02 17:24:56 -04:00
|
|
|
$url = 'http:' . $url;
|
2016-11-23 19:11:22 -05:00
|
|
|
} else {
|
2018-09-02 17:24:56 -04:00
|
|
|
$url = 'http://' . $url;
|
2015-03-07 17:14:26 -05:00
|
|
|
}
|
2013-02-24 06:54:53 -05:00
|
|
|
}
|
|
|
|
|
2018-10-29 17:20:46 -04:00
|
|
|
Logger::log($url);
|
2016-11-23 19:11:22 -05:00
|
|
|
|
2016-11-25 10:59:31 -05:00
|
|
|
// Check if the URL is an image, video or audio file. If so format
|
|
|
|
// the URL with the corresponding BBCode media tag
|
2016-11-23 19:11:22 -05:00
|
|
|
$redirects = 0;
|
|
|
|
// Fetch the header of the URL
|
2018-10-10 15:08:43 -04:00
|
|
|
$curlResponse = Network::curl($url, false, $redirects, ['novalidate' => true, 'nobody' => true]);
|
2018-09-02 17:24:56 -04:00
|
|
|
|
2018-10-10 15:08:43 -04:00
|
|
|
if ($curlResponse->isSuccess()) {
|
2016-11-23 19:11:22 -05:00
|
|
|
// Convert the header fields into an array
|
2018-01-15 08:05:12 -05:00
|
|
|
$hdrs = [];
|
2018-10-10 15:08:43 -04:00
|
|
|
$h = explode("\n", $curlResponse->getHeader());
|
2016-11-23 19:11:22 -05:00
|
|
|
foreach ($h as $l) {
|
2018-09-02 17:24:56 -04:00
|
|
|
$header = array_map('trim', explode(':', trim($l), 2));
|
2018-07-15 14:36:20 -04:00
|
|
|
if (count($header) == 2) {
|
2018-09-02 17:24:56 -04:00
|
|
|
list($k, $v) = $header;
|
2018-07-15 14:36:20 -04:00
|
|
|
$hdrs[$k] = $v;
|
|
|
|
}
|
2016-11-23 19:11:22 -05:00
|
|
|
}
|
2018-10-16 18:27:13 -04:00
|
|
|
$type = null;
|
2018-09-02 17:24:56 -04:00
|
|
|
if (array_key_exists('Content-Type', $hdrs)) {
|
|
|
|
$type = $hdrs['Content-Type'];
|
2016-11-23 19:11:22 -05:00
|
|
|
}
|
|
|
|
if ($type) {
|
2018-09-02 17:24:56 -04:00
|
|
|
if (stripos($type, 'image/') !== false) {
|
|
|
|
echo $br . '[img]' . $url . '[/img]' . $br;
|
|
|
|
exit();
|
2014-04-04 04:58:31 -04:00
|
|
|
}
|
2018-09-02 17:24:56 -04:00
|
|
|
if (stripos($type, 'video/') !== false) {
|
|
|
|
echo $br . '[video]' . $url . '[/video]' . $br;
|
|
|
|
exit();
|
2014-04-04 04:58:31 -04:00
|
|
|
}
|
2018-09-02 17:24:56 -04:00
|
|
|
if (stripos($type, 'audio/') !== false) {
|
|
|
|
echo $br . '[audio]' . $url . '[/audio]' . $br;
|
|
|
|
exit();
|
2012-07-12 02:20:27 -04:00
|
|
|
}
|
2015-03-07 17:14:26 -05:00
|
|
|
}
|
2012-07-12 02:20:27 -04:00
|
|
|
}
|
2012-07-11 19:17:33 -04:00
|
|
|
|
2018-10-17 15:05:45 -04:00
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
$template = '[bookmark=%s]%s[/bookmark]%s';
|
2010-07-23 01:41:45 -04:00
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
$arr = ['url' => $url, 'text' => ''];
|
2010-12-25 18:01:02 -05:00
|
|
|
|
2018-12-26 01:06:24 -05:00
|
|
|
Hook::callAll('parse_link', $arr);
|
2010-12-25 18:01:02 -05:00
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
if (strlen($arr['text'])) {
|
|
|
|
echo $arr['text'];
|
|
|
|
exit();
|
2010-12-25 18:01:02 -05:00
|
|
|
}
|
|
|
|
|
2018-07-31 01:54:25 -04:00
|
|
|
// If there is already some content information submitted we don't
|
2016-11-25 10:59:31 -05:00
|
|
|
// need to parse the url for content.
|
2018-07-31 01:54:25 -04:00
|
|
|
if (!empty($url) && !empty($title) && !empty($text)) {
|
2018-09-02 17:24:56 -04:00
|
|
|
$title = str_replace(["\r", "\n"], ['', ''], $title);
|
2011-09-20 01:21:55 -04:00
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
$text = '[quote]' . trim($text) . '[/quote]' . $br;
|
2011-09-20 01:21:55 -04:00
|
|
|
|
2016-11-23 19:11:22 -05:00
|
|
|
$result = sprintf($template, $url, ($title) ? $title : $url, $text) . $str_tags;
|
2011-09-20 01:21:55 -04:00
|
|
|
|
2018-10-29 17:20:46 -04:00
|
|
|
Logger::log('(unparsed): returns: ' . $result);
|
2011-09-20 01:21:55 -04:00
|
|
|
|
|
|
|
echo $result;
|
2018-09-02 17:24:56 -04:00
|
|
|
exit();
|
2011-09-20 01:21:55 -04:00
|
|
|
}
|
|
|
|
|
2016-11-25 10:59:31 -05:00
|
|
|
// Fetch the information directly from the webpage
|
2016-11-23 19:11:22 -05:00
|
|
|
$siteinfo = ParseUrl::getSiteinfo($url);
|
2011-09-20 01:21:55 -04:00
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
unset($siteinfo['keywords']);
|
2014-01-05 10:22:42 -05:00
|
|
|
|
2018-10-24 10:20:10 -04:00
|
|
|
// Bypass attachment if parse url for a comment
|
|
|
|
if (!empty($_GET['noAttachment'])) {
|
|
|
|
echo $br . '[url=' . $url . ']' . $siteinfo['title'] . '[/url]';
|
|
|
|
exit();
|
|
|
|
}
|
|
|
|
|
2016-11-23 19:11:22 -05:00
|
|
|
// Format it as BBCode attachment
|
2016-04-18 14:57:01 -04:00
|
|
|
$info = add_page_info_data($siteinfo);
|
|
|
|
|
|
|
|
echo $info;
|
|
|
|
|
2018-09-02 17:24:56 -04:00
|
|
|
exit();
|
2011-05-23 04:37:09 -04:00
|
|
|
}
|
2016-11-28 09:26:51 -05:00
|
|
|
|
|
|
|
/**
|
|
|
|
* @brief Legacy function to call ParseUrl::getSiteinfoCached
|
2017-01-09 07:14:25 -05:00
|
|
|
*
|
2016-11-28 09:26:51 -05:00
|
|
|
* Note: We have moved the function to ParseUrl.php. This function is only for
|
|
|
|
* legacy support and will be remove in the future
|
2017-01-09 07:14:25 -05:00
|
|
|
*
|
2016-11-28 09:26:51 -05:00
|
|
|
* @param type $url The url of the page which should be scraped
|
|
|
|
* @param type $no_guessing If true the parse doens't search for
|
|
|
|
* preview pictures
|
|
|
|
* @param type $do_oembed The false option is used by the function fetch_oembed()
|
|
|
|
* to avoid endless loops
|
2017-01-09 07:14:25 -05:00
|
|
|
*
|
2016-11-28 09:26:51 -05:00
|
|
|
* @return array which contains needed data for embedding
|
2017-01-09 07:14:25 -05:00
|
|
|
*
|
2016-11-28 09:26:51 -05:00
|
|
|
* @see ParseUrl::getSiteinfoCached()
|
2017-01-09 07:14:25 -05:00
|
|
|
*
|
2016-11-28 09:26:51 -05:00
|
|
|
* @todo Remove this function after all Addons has been changed to use
|
|
|
|
* ParseUrl::getSiteinfoCached
|
|
|
|
*/
|
2018-09-02 17:24:56 -04:00
|
|
|
function parseurl_getsiteinfo_cached($url, $no_guessing = false, $do_oembed = true)
|
|
|
|
{
|
2016-11-28 09:26:51 -05:00
|
|
|
$siteinfo = ParseUrl::getSiteinfoCached($url, $no_guessing, $do_oembed);
|
|
|
|
return $siteinfo;
|
|
|
|
}
|