2018-09-20 07:30:07 +02:00
|
|
|
<?php
|
|
|
|
/**
|
|
|
|
* @file src/Util/JsonLD.php
|
|
|
|
*/
|
|
|
|
namespace Friendica\Util;
|
|
|
|
|
|
|
|
use Friendica\Core\Cache;
|
2018-10-29 22:20:46 +01:00
|
|
|
use Friendica\Core\Logger;
|
2018-09-27 00:45:13 +02:00
|
|
|
use Exception;
|
2020-01-07 00:41:20 +01:00
|
|
|
use Friendica\DI;
|
2018-09-20 07:30:07 +02:00
|
|
|
|
|
|
|
/**
|
|
|
|
* @brief This class contain methods to work with JsonLD data
|
|
|
|
*/
|
|
|
|
class JsonLD
|
|
|
|
{
|
2018-09-27 00:02:14 +02:00
|
|
|
/**
|
|
|
|
* @brief Loader for LD-JSON validation
|
|
|
|
*
|
|
|
|
* @param $url
|
|
|
|
*
|
2019-01-06 22:06:53 +01:00
|
|
|
* @return mixed the loaded data
|
|
|
|
* @throws \JsonLdException
|
2018-09-27 00:02:14 +02:00
|
|
|
*/
|
2018-09-20 07:30:07 +02:00
|
|
|
public static function documentLoader($url)
|
|
|
|
{
|
|
|
|
$recursion = 0;
|
|
|
|
|
|
|
|
$x = debug_backtrace();
|
|
|
|
if ($x) {
|
|
|
|
foreach ($x as $n) {
|
|
|
|
if ($n['function'] === __FUNCTION__) {
|
|
|
|
$recursion ++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if ($recursion > 5) {
|
2019-02-28 10:07:24 +01:00
|
|
|
Logger::error('jsonld bomb detected at: ' . $url);
|
2018-09-20 07:30:07 +02:00
|
|
|
exit();
|
|
|
|
}
|
|
|
|
|
2020-01-07 00:45:49 +01:00
|
|
|
$result = DI::cache()->get('documentLoader:' . $url);
|
2018-09-20 07:30:07 +02:00
|
|
|
if (!is_null($result)) {
|
|
|
|
return $result;
|
|
|
|
}
|
|
|
|
|
|
|
|
$data = jsonld_default_document_loader($url);
|
2020-01-07 00:41:20 +01:00
|
|
|
DI::cache()->set('documentLoader:' . $url, $data, Cache::DAY);
|
2018-09-20 07:30:07 +02:00
|
|
|
return $data;
|
|
|
|
}
|
|
|
|
|
2018-09-27 00:02:14 +02:00
|
|
|
/**
|
|
|
|
* @brief Normalises a given JSON array
|
|
|
|
*
|
|
|
|
* @param array $json
|
|
|
|
*
|
2019-01-06 22:06:53 +01:00
|
|
|
* @return mixed|bool normalized JSON string
|
|
|
|
* @throws Exception
|
2018-09-27 00:02:14 +02:00
|
|
|
*/
|
2018-09-20 07:30:07 +02:00
|
|
|
public static function normalize($json)
|
|
|
|
{
|
|
|
|
jsonld_set_document_loader('Friendica\Util\JsonLD::documentLoader');
|
|
|
|
|
2018-09-22 00:31:33 +02:00
|
|
|
$jsonobj = json_decode(json_encode($json, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE));
|
2018-09-20 07:30:07 +02:00
|
|
|
|
2018-09-27 00:45:13 +02:00
|
|
|
try {
|
|
|
|
$normalized = jsonld_normalize($jsonobj, array('algorithm' => 'URDNA2015', 'format' => 'application/nquads'));
|
|
|
|
}
|
|
|
|
catch (Exception $e) {
|
2018-10-02 10:31:58 +02:00
|
|
|
$normalized = false;
|
2019-05-20 00:18:14 +02:00
|
|
|
$messages = [];
|
|
|
|
$currentException = $e;
|
|
|
|
do {
|
|
|
|
$messages[] = $currentException->getMessage();
|
|
|
|
} while($currentException = $currentException->getPrevious());
|
|
|
|
|
|
|
|
Logger::warning('JsonLD normalize error');
|
|
|
|
Logger::notice('JsonLD normalize error', ['messages' => $messages]);
|
|
|
|
Logger::info('JsonLD normalize error', ['trace' => $e->getTraceAsString()]);
|
|
|
|
Logger::debug('JsonLD normalize error', ['jsonobj' => $jsonobj]);
|
2018-09-27 00:45:13 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
return $normalized;
|
2018-09-20 07:30:07 +02:00
|
|
|
}
|
|
|
|
|
2018-09-27 00:02:14 +02:00
|
|
|
/**
|
|
|
|
* @brief Compacts a given JSON array
|
|
|
|
*
|
|
|
|
* @param array $json
|
|
|
|
*
|
2019-01-06 22:06:53 +01:00
|
|
|
* @return array Compacted JSON array
|
|
|
|
* @throws Exception
|
2018-09-27 00:02:14 +02:00
|
|
|
*/
|
2018-09-20 07:30:07 +02:00
|
|
|
public static function compact($json)
|
|
|
|
{
|
|
|
|
jsonld_set_document_loader('Friendica\Util\JsonLD::documentLoader');
|
|
|
|
|
2018-10-07 15:37:05 +02:00
|
|
|
$context = (object)['as' => 'https://www.w3.org/ns/activitystreams#',
|
|
|
|
'w3id' => 'https://w3id.org/security#',
|
2018-11-02 22:57:06 +01:00
|
|
|
'ldp' => (object)['@id' => 'http://www.w3.org/ns/ldp#', '@type' => '@id'],
|
2018-09-20 07:30:07 +02:00
|
|
|
'vcard' => (object)['@id' => 'http://www.w3.org/2006/vcard/ns#', '@type' => '@id'],
|
2018-11-02 22:57:06 +01:00
|
|
|
'dfrn' => (object)['@id' => 'http://purl.org/macgirvin/dfrn/1.0/', '@type' => '@id'],
|
2018-10-07 15:37:05 +02:00
|
|
|
'diaspora' => (object)['@id' => 'https://diasporafoundation.org/ns/', '@type' => '@id'],
|
2018-11-02 22:57:06 +01:00
|
|
|
'ostatus' => (object)['@id' => 'http://ostatus.org#', '@type' => '@id'],
|
2018-11-07 21:34:03 +01:00
|
|
|
'dc' => (object)['@id' => 'http://purl.org/dc/terms/', '@type' => '@id'],
|
2019-05-14 19:50:45 +02:00
|
|
|
'toot' => (object)['@id' => 'http://joinmastodon.org/ns#', '@type' => '@id'],
|
|
|
|
'litepub' => (object)['@id' => 'http://litepub.social/ns#', '@type' => '@id']];
|
2018-09-20 07:30:07 +02:00
|
|
|
|
2019-04-25 15:30:30 +02:00
|
|
|
// Preparation for adding possibly missing content to the context
|
|
|
|
if (!empty($json['@context']) && is_string($json['@context'])) {
|
|
|
|
$json['@context'] = [$json['@context']];
|
|
|
|
}
|
|
|
|
|
|
|
|
// Workaround for servers with missing context
|
2019-01-13 10:38:01 +01:00
|
|
|
// See issue https://github.com/nextcloud/social/issues/330
|
2019-01-13 19:03:13 +01:00
|
|
|
if (!empty($json['@context']) && is_array($json['@context'])) {
|
2019-01-13 17:00:27 +01:00
|
|
|
$json['@context'][] = 'https://w3id.org/security/v1';
|
|
|
|
}
|
2019-01-13 10:38:01 +01:00
|
|
|
|
2018-12-11 19:48:33 +01:00
|
|
|
// Trying to avoid memory problems with large content fields
|
|
|
|
if (!empty($json['object']['source']['content'])) {
|
|
|
|
$content = $json['object']['source']['content'];
|
|
|
|
$json['object']['source']['content'] = '';
|
|
|
|
}
|
2018-09-20 07:30:07 +02:00
|
|
|
|
2018-12-11 19:48:33 +01:00
|
|
|
$jsonobj = json_decode(json_encode($json, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE));
|
2018-10-13 15:17:10 +02:00
|
|
|
|
|
|
|
try {
|
|
|
|
$compacted = jsonld_compact($jsonobj, $context);
|
|
|
|
}
|
|
|
|
catch (Exception $e) {
|
|
|
|
$compacted = false;
|
2019-02-28 10:07:24 +01:00
|
|
|
Logger::error('compacting error');
|
|
|
|
// Sooner or later we should log some details as well - but currently this leads to memory issues
|
|
|
|
// Logger::log('compacting error:' . substr(print_r($e, true), 0, 10000), Logger::DEBUG);
|
2018-10-13 15:17:10 +02:00
|
|
|
}
|
2018-09-20 07:30:07 +02:00
|
|
|
|
2018-12-11 19:48:33 +01:00
|
|
|
$json = json_decode(json_encode($compacted, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE), true);
|
|
|
|
|
|
|
|
if (isset($json['as:object']['as:source']['as:content']) && !empty($content)) {
|
|
|
|
$json['as:object']['as:source']['as:content'] = $content;
|
|
|
|
}
|
|
|
|
|
|
|
|
return $json;
|
2018-09-20 07:30:07 +02:00
|
|
|
}
|
2018-09-20 07:37:01 +02:00
|
|
|
|
2018-10-07 15:37:05 +02:00
|
|
|
/**
|
|
|
|
* @brief Fetches an element array from a JSON array
|
|
|
|
*
|
|
|
|
* @param $array
|
|
|
|
* @param $element
|
|
|
|
* @param $key
|
|
|
|
*
|
2019-01-06 22:06:53 +01:00
|
|
|
* @return array fetched element
|
2018-10-07 15:37:05 +02:00
|
|
|
*/
|
|
|
|
public static function fetchElementArray($array, $element, $key = '@id')
|
|
|
|
{
|
|
|
|
if (empty($array)) {
|
|
|
|
return null;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!isset($array[$element])) {
|
|
|
|
return null;
|
|
|
|
}
|
|
|
|
|
|
|
|
// If it isn't an array yet, make it to one
|
|
|
|
if (!is_int(key($array[$element]))) {
|
|
|
|
$array[$element] = [$array[$element]];
|
|
|
|
}
|
|
|
|
|
|
|
|
$elements = [];
|
|
|
|
|
|
|
|
foreach ($array[$element] as $entry) {
|
|
|
|
if (!is_array($entry)) {
|
|
|
|
$elements[] = $entry;
|
2019-04-26 08:17:37 +02:00
|
|
|
} elseif (isset($entry[$key])) {
|
2018-10-07 15:37:05 +02:00
|
|
|
$elements[] = $entry[$key];
|
2018-10-10 21:15:16 +02:00
|
|
|
} elseif (!empty($entry) || !is_array($entry)) {
|
2018-10-07 15:37:05 +02:00
|
|
|
$elements[] = $entry;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return $elements;
|
|
|
|
}
|
|
|
|
|
2018-09-27 00:02:14 +02:00
|
|
|
/**
|
|
|
|
* @brief Fetches an element from a JSON array
|
|
|
|
*
|
|
|
|
* @param $array
|
|
|
|
* @param $element
|
|
|
|
* @param $key
|
|
|
|
* @param $type
|
|
|
|
* @param $type_value
|
|
|
|
*
|
2019-01-06 22:06:53 +01:00
|
|
|
* @return string fetched element
|
2018-09-27 00:02:14 +02:00
|
|
|
*/
|
2018-10-07 15:37:05 +02:00
|
|
|
public static function fetchElement($array, $element, $key = '@id', $type = null, $type_value = null)
|
2018-09-20 07:37:01 +02:00
|
|
|
{
|
|
|
|
if (empty($array)) {
|
2018-10-07 15:37:05 +02:00
|
|
|
return null;
|
2018-09-20 07:37:01 +02:00
|
|
|
}
|
|
|
|
|
2018-10-07 15:37:05 +02:00
|
|
|
if (!isset($array[$element])) {
|
|
|
|
return null;
|
2018-09-20 07:37:01 +02:00
|
|
|
}
|
|
|
|
|
2018-10-07 15:37:05 +02:00
|
|
|
if (!is_array($array[$element])) {
|
2018-09-20 07:37:01 +02:00
|
|
|
return $array[$element];
|
|
|
|
}
|
|
|
|
|
2018-10-07 15:37:05 +02:00
|
|
|
if (is_null($type) || is_null($type_value)) {
|
|
|
|
$element_array = self::fetchElementArray($array, $element, $key);
|
|
|
|
if (is_null($element_array)) {
|
|
|
|
return null;
|
2018-09-20 07:37:01 +02:00
|
|
|
}
|
|
|
|
|
2018-10-07 15:37:05 +02:00
|
|
|
return array_shift($element_array);
|
2018-09-20 07:37:01 +02:00
|
|
|
}
|
|
|
|
|
2018-10-07 15:37:05 +02:00
|
|
|
$element_array = self::fetchElementArray($array, $element);
|
|
|
|
if (is_null($element_array)) {
|
|
|
|
return null;
|
2018-09-20 07:37:01 +02:00
|
|
|
}
|
|
|
|
|
2018-10-07 15:37:05 +02:00
|
|
|
foreach ($element_array as $entry) {
|
|
|
|
if (isset($entry[$key]) && isset($entry[$type]) && ($entry[$type] == $type_value)) {
|
|
|
|
return $entry[$key];
|
|
|
|
}
|
|
|
|
}
|
2018-09-20 07:37:01 +02:00
|
|
|
|
2018-10-07 15:37:05 +02:00
|
|
|
return null;
|
2018-09-20 07:37:01 +02:00
|
|
|
}
|
2018-09-20 07:30:07 +02:00
|
|
|
}
|