friendica/src/Content/OEmbed.php

446 lines
13 KiB
PHP
Raw Normal View History

2018-01-10 14:36:02 +01:00
<?php
/**
2021-03-29 08:40:20 +02:00
* @copyright Copyright (C) 2010-2021, the Friendica project
2020-02-09 15:45:36 +01:00
*
* @license GNU AGPL version 3 or any later version
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU Affero General Public License as
* published by the Free Software Foundation, either version 3 of the
* License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Affero General Public License for more details.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <https://www.gnu.org/licenses/>.
*
2018-01-10 14:36:02 +01:00
*/
2020-02-09 15:45:36 +01:00
2018-01-10 14:36:02 +01:00
namespace Friendica\Content;
use DOMDocument;
use DOMNode;
use DOMText;
use DOMXPath;
use Exception;
use Friendica\Core\Cache\Duration;
use Friendica\Core\Hook;
use Friendica\Core\Renderer;
2020-11-19 20:34:48 +01:00
use Friendica\Database\Database;
use Friendica\Database\DBA;
use Friendica\DI;
use Friendica\Util\DateTimeFormat;
2018-01-27 05:09:48 +01:00
use Friendica\Util\Network;
2018-01-10 14:36:02 +01:00
use Friendica\Util\ParseUrl;
use Friendica\Util\Proxy as ProxyUtils;
use Friendica\Util\Strings;
2018-01-10 14:36:02 +01:00
/**
* Handles all OEmbed content fetching and replacement
*
* OEmbed is a standard used to allow an embedded representation of a URL on
* third party sites
*
* @see https://oembed.com
*/
class OEmbed
{
public static function replaceCallback($matches)
{
$embedurl = $matches[1];
$j = self::fetchURL($embedurl, !self::isAllowedURL($embedurl));
2018-01-10 14:36:02 +01:00
$s = self::formatObject($j);
return $s;
}
/**
2020-01-19 07:05:23 +01:00
* Get data from an URL to embed its content.
2018-01-10 14:36:02 +01:00
*
* @param string $embedurl The URL from which the data should be fetched.
* @param bool $no_rich_type If set to true rich type content won't be fetched.
2021-03-16 08:04:16 +01:00
* @param bool $use_parseurl Use the "ParseUrl" functionality to add additional data
2018-01-10 14:36:02 +01:00
*
* @return \Friendica\Object\OEmbed
2019-01-06 22:06:53 +01:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
2018-01-10 14:36:02 +01:00
*/
2021-03-16 08:04:16 +01:00
public static function fetchURL($embedurl, bool $no_rich_type = false, bool $use_parseurl = true)
2018-01-10 14:36:02 +01:00
{
$embedurl = trim($embedurl, '\'"');
2018-01-10 14:36:02 +01:00
2020-01-04 23:42:01 +01:00
$a = DI::app();
2018-01-10 14:36:02 +01:00
$cache_key = 'oembed:' . $a->videowidth . ':' . $embedurl;
$condition = ['url' => Strings::normaliseLink($embedurl), 'maxwidth' => $a->videowidth];
$oembed_record = DBA::selectFirst('oembed', ['content'], $condition);
if (DBA::isResult($oembed_record)) {
$json_string = $oembed_record['content'];
2018-01-10 14:36:02 +01:00
} else {
$json_string = DI::cache()->get($cache_key);
2018-01-10 14:36:02 +01:00
}
2018-01-10 14:36:02 +01:00
// These media files should now be caught in bbcode.php
// left here as a fallback in case this is called from another source
$noexts = ['mp3', 'mp4', 'ogg', 'ogv', 'oga', 'ogm', 'webm'];
2018-01-10 14:36:02 +01:00
$ext = pathinfo(strtolower($embedurl), PATHINFO_EXTENSION);
$oembed = new \Friendica\Object\OEmbed($embedurl);
2018-01-10 14:36:02 +01:00
if ($json_string) {
$oembed->parseJSON($json_string);
} else {
$json_string = '';
2018-01-10 14:36:02 +01:00
if (!in_array($ext, $noexts)) {
// try oembed autodiscovery
$html_text = DI::httpRequest()->fetch($embedurl, 15, 'text/*');
2018-01-10 14:36:02 +01:00
if ($html_text) {
$dom = new DOMDocument();
if (@$dom->loadHTML($html_text)) {
2018-01-10 14:36:02 +01:00
$xpath = new DOMXPath($dom);
foreach (
$xpath->query("//link[@type='application/json+oembed'] | //link[@type='text/json+oembed']")
as $link)
{
$href = $link->getAttributeNode('href')->nodeValue;
// Both Youtube and Vimeo output OEmbed endpoint URL with HTTP
// but their OEmbed endpoint is only accessible by HTTPS ¯\_(ツ)_/¯
$href = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'],
['https://www.youtube.com/', 'https://player.vimeo.com/'], $href);
$result = DI::httpRequest()->fetchFull($href . '&maxwidth=' . $a->videowidth);
if ($result->getReturnCode() === 200) {
$json_string = $result->getBody();
break;
}
2018-01-10 14:36:02 +01:00
}
}
}
}
$json_string = trim($json_string);
2018-01-10 14:36:02 +01:00
if (!$json_string || $json_string[0] != '{') {
$json_string = '{"type":"error"}';
2018-01-10 14:36:02 +01:00
}
$oembed->parseJSON($json_string);
2018-07-24 17:05:09 +02:00
if (!empty($oembed->type) && $oembed->type != 'error') {
DBA::insert('oembed', [
'url' => Strings::normaliseLink($embedurl),
'maxwidth' => $a->videowidth,
'content' => $json_string,
'created' => DateTimeFormat::utcNow()
2020-11-19 20:34:48 +01:00
], Database::INSERT_UPDATE);
$cache_ttl = Duration::DAY;
2018-07-24 17:05:09 +02:00
} else {
$cache_ttl = Duration::FIVE_MINUTES;
}
2018-01-10 14:36:02 +01:00
DI::cache()->set($cache_key, $json_string, $cache_ttl);
2018-01-10 14:36:02 +01:00
}
// Always embed the SSL version
if (!empty($oembed->html)) {
$oembed->html = str_replace(['http://www.youtube.com/', 'http://player.vimeo.com/'], ['https://www.youtube.com/', 'https://player.vimeo.com/'], $oembed->html);
2018-01-10 14:36:02 +01:00
}
// Improve the OEmbed data with data from OpenGraph, Twitter cards and other sources
2021-03-16 08:04:16 +01:00
if ($use_parseurl) {
2021-03-16 08:15:20 +01:00
$data = ParseUrl::getSiteinfoCached($embedurl, false);
2021-03-16 08:04:16 +01:00
if (($oembed->type == 'error') && empty($data['title']) && empty($data['text'])) {
return $oembed;
}
2018-01-10 14:36:02 +01:00
2021-03-16 08:04:16 +01:00
if ($no_rich_type || ($oembed->type == 'error')) {
$oembed->html = '';
$oembed->type = $data['type'];
2018-01-10 14:36:02 +01:00
2021-03-16 08:04:16 +01:00
if ($oembed->type == 'photo') {
$oembed->url = $data['url'];
}
2018-01-10 14:36:02 +01:00
}
2021-03-16 08:04:16 +01:00
if (!empty($data['title'])) {
$oembed->title = $data['title'];
}
2018-01-10 14:36:02 +01:00
2021-03-16 08:04:16 +01:00
if (!empty($data['text'])) {
$oembed->description = $data['text'];
}
2018-01-10 14:36:02 +01:00
2021-03-16 08:04:16 +01:00
if (!empty($data['publisher_name'])) {
$oembed->provider_name = $data['publisher_name'];
}
2021-03-16 08:04:16 +01:00
if (!empty($data['publisher_url'])) {
$oembed->provider_url = $data['publisher_url'];
}
2021-03-16 08:04:16 +01:00
if (!empty($data['author_name'])) {
$oembed->author_name = $data['author_name'];
}
2021-03-16 08:04:16 +01:00
if (!empty($data['author_url'])) {
$oembed->author_url = $data['author_url'];
}
2021-03-16 08:04:16 +01:00
if (!empty($data['images'])) {
$oembed->thumbnail_url = $data['images'][0]['src'];
$oembed->thumbnail_width = $data['images'][0]['width'];
$oembed->thumbnail_height = $data['images'][0]['height'];
}
2018-01-10 14:36:02 +01:00
}
Hook::callAll('oembed_fetch_url', $embedurl, $oembed);
2018-01-10 14:36:02 +01:00
return $oembed;
2018-01-10 14:36:02 +01:00
}
private static function formatObject(\Friendica\Object\OEmbed $oembed)
2018-01-10 14:36:02 +01:00
{
$ret = '<div class="oembed ' . $oembed->type . '">';
switch ($oembed->type) {
2018-01-10 14:36:02 +01:00
case "video":
if ($oembed->thumbnail_url) {
$tw = (isset($oembed->thumbnail_width) && intval($oembed->thumbnail_width)) ? $oembed->thumbnail_width : 200;
$th = (isset($oembed->thumbnail_height) && intval($oembed->thumbnail_height)) ? $oembed->thumbnail_height : 180;
2018-01-10 14:36:02 +01:00
// make sure we don't attempt divide by zero, fallback is a 1:1 ratio
$tr = (($th) ? $tw / $th : 1);
$th = 120;
$tw = $th * $tr;
$tpl = Renderer::getMarkupTemplate('oembed_video.tpl');
$ret .= Renderer::replaceMacros($tpl, [
'$embedurl' => $oembed->embed_url,
'$escapedhtml' => base64_encode($oembed->html),
2018-01-10 14:36:02 +01:00
'$tw' => $tw,
'$th' => $th,
'$turl' => $oembed->thumbnail_url,
]);
2018-01-10 14:36:02 +01:00
} else {
$ret = $oembed->html;
2018-01-10 14:36:02 +01:00
}
break;
2018-01-10 14:36:02 +01:00
case "photo":
$ret .= '<img width="' . $oembed->width . '" src="' . ProxyUtils::proxifyUrl($oembed->url) . '">';
2018-01-10 14:36:02 +01:00
break;
2018-01-10 14:36:02 +01:00
case "link":
break;
2018-01-10 14:36:02 +01:00
case "rich":
$ret .= ProxyUtils::proxifyHtml($oembed->html);
2018-01-10 14:36:02 +01:00
break;
}
// add link to source if not present in "rich" type
if ($oembed->type != 'rich' || !strpos($oembed->html, $oembed->embed_url)) {
2018-01-10 14:36:02 +01:00
$ret .= '<h4>';
if (!empty($oembed->title)) {
if (!empty($oembed->provider_name)) {
$ret .= $oembed->provider_name . ": ";
2018-01-10 14:36:02 +01:00
}
$ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
if (!empty($oembed->author_name)) {
$ret .= ' (' . $oembed->author_name . ')';
2018-01-10 14:36:02 +01:00
}
} elseif (!empty($oembed->provider_name) || !empty($oembed->author_name)) {
2018-01-10 14:36:02 +01:00
$embedlink = "";
if (!empty($oembed->provider_name)) {
$embedlink .= $oembed->provider_name;
2018-01-10 14:36:02 +01:00
}
if (!empty($oembed->author_name)) {
2018-01-10 14:36:02 +01:00
if ($embedlink != "") {
$embedlink .= ": ";
}
$embedlink .= $oembed->author_name;
2018-01-10 14:36:02 +01:00
}
if (trim($embedlink) == "") {
$embedlink = $oembed->embed_url;
2018-01-10 14:36:02 +01:00
}
$ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $embedlink . '</a>';
} else {
$ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->embed_url . '</a>';
2018-01-10 14:36:02 +01:00
}
$ret .= "</h4>";
} elseif (!strpos($oembed->html, $oembed->embed_url)) {
2018-01-10 14:36:02 +01:00
// add <a> for html2bbcode conversion
$ret .= '<a href="' . $oembed->embed_url . '" rel="oembed">' . $oembed->title . '</a>';
2018-01-10 14:36:02 +01:00
}
2018-01-11 23:34:56 +01:00
$ret .= '</div>';
return str_replace("\n", "", $ret);
2018-01-10 14:36:02 +01:00
}
public static function BBCode2HTML($text)
{
$stopoembed = DI::config()->get("system", "no_oembed");
2018-01-10 14:36:02 +01:00
if ($stopoembed == true) {
return preg_replace("/\[embed\](.+?)\[\/embed\]/is", "<!-- oembed $1 --><i>" . DI::l10n()->t('Embedding disabled') . " : $1</i><!-- /oembed $1 -->", $text);
2018-01-10 14:36:02 +01:00
}
return preg_replace_callback("/\[embed\](.+?)\[\/embed\]/is", ['self', 'replaceCallback'], $text);
}
/**
* Find <span class='oembed'>..<a href='url' rel='oembed'>..</a></span>
* and replace it with [embed]url[/embed]
2019-01-06 22:06:53 +01:00
*
* @param $text
* @return string
2018-01-10 14:36:02 +01:00
*/
public static function HTML2BBCode($text)
{
// start parser only if 'oembed' is in text
if (strpos($text, "oembed")) {
// convert non ascii chars to html entities
$html_text = mb_convert_encoding($text, 'HTML-ENTITIES', mb_detect_encoding($text));
// If it doesn't parse at all, just return the text.
2019-01-23 22:54:20 +01:00
$dom = @DOMDocument::loadHTML($html_text);
2018-01-10 14:36:02 +01:00
if (!$dom) {
return $text;
}
$xpath = new DOMXPath($dom);
$xattr = self::buildXPath("class", "oembed");
$entries = $xpath->query("//div[$xattr]");
$xattr = "@rel='oembed'"; //oe_build_xpath("rel","oembed");
foreach ($entries as $e) {
$href = $xpath->evaluate("a[$xattr]/@href", $e)->item(0)->nodeValue;
if (!is_null($href)) {
$e->parentNode->replaceChild(new DOMText("[embed]" . $href . "[/embed]"), $e);
}
}
return self::getInnerHTML($dom->getElementsByTagName("body")->item(0));
} else {
return $text;
}
}
/**
* Determines if rich content OEmbed is allowed for the provided URL
*
* @param string $url
* @return boolean
2019-01-06 22:06:53 +01:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
2018-01-10 14:36:02 +01:00
*/
public static function isAllowedURL($url)
{
if (!DI::config()->get('system', 'no_oembed_rich_content')) {
2018-01-10 14:36:02 +01:00
return true;
}
$domain = parse_url($url, PHP_URL_HOST);
if (empty($domain)) {
2018-01-10 14:36:02 +01:00
return false;
}
$str_allowed = DI::config()->get('system', 'allowed_oembed', '');
if (empty($str_allowed)) {
2018-01-10 14:36:02 +01:00
return false;
}
2018-01-10 14:36:02 +01:00
$allowed = explode(',', $str_allowed);
return Network::isDomainAllowed($domain, $allowed);
2018-01-10 14:36:02 +01:00
}
public static function getHTML($url, $title = null)
{
$o = self::fetchURL($url, !self::isAllowedURL($url));
2018-01-10 14:36:02 +01:00
if (!is_object($o) || property_exists($o, 'type') && $o->type == 'error') {
2018-01-10 14:36:02 +01:00
throw new Exception('OEmbed failed for URL: ' . $url);
}
if (!empty($title)) {
2018-01-10 14:36:02 +01:00
$o->title = $title;
}
$html = self::formatObject($o);
2018-01-10 14:36:02 +01:00
return $html;
}
/**
2020-01-19 07:05:23 +01:00
* Generates the iframe HTML for an oembed attachment.
2018-01-10 14:36:02 +01:00
*
* Width and height are given by the remote, and are regularly too small for
* the generated iframe.
*
* The width is entirely discarded for the actual width of the post, while fixed
* height is used as a starting point before the inevitable resizing.
*
* Since the iframe is automatically resized on load, there are no need for ugly
* and impractical scrollbars.
*
2019-01-06 22:06:53 +01:00
* @todo This function is currently unused until someone™ adds support for a separate OEmbed domain
2018-01-10 14:36:02 +01:00
*
* @param string $src Original remote URL to embed
* @param string $width
* @param string $height
* @return string formatted HTML
*
2019-01-06 22:06:53 +01:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
* @see oembed_format_object()
2018-01-10 14:36:02 +01:00
*/
private static function iframe($src, $width, $height)
{
if (!$height || strstr($height, '%')) {
$height = '200';
}
$width = '100%';
$src = DI::baseUrl() . '/oembed/' . Strings::base64UrlEncode($src);
return '<iframe onload="resizeIframe(this);" class="embed_rich" height="' . $height . '" width="' . $width . '" src="' . $src . '" allowfullscreen scrolling="no" frameborder="no">' . DI::l10n()->t('Embedded content') . '</iframe>';
2018-01-10 14:36:02 +01:00
}
/**
2020-01-19 07:05:23 +01:00
* Generates attribute search XPath string
*
2018-01-10 14:36:02 +01:00
* Generates an XPath query to select elements whose provided attribute contains
* the provided value in a space-separated list.
*
* @param string $attr Name of the attribute to seach
* @param string $value Value to search in a space-separated list
* @return string
*/
private static function buildXPath($attr, $value)
{
// https://www.westhoffswelt.de/blog/2009/6/9/select-html-elements-with-more-than-one-css-class-using-xpath
return "contains(normalize-space(@$attr), ' $value ') or substring(normalize-space(@$attr), 1, string-length('$value') + 1) = '$value ' or substring(normalize-space(@$attr), string-length(@$attr) - string-length('$value')) = ' $value' or @$attr = '$value'";
}
/**
* Returns the inner XML string of a provided DOMNode
*
* @param DOMNode $node
* @return string
*/
private static function getInnerHTML(DOMNode $node)
{
$innerHTML = '';
$children = $node->childNodes;
foreach ($children as $child) {
$innerHTML .= $child->ownerDocument->saveXML($child);
}
return $innerHTML;
}
}