src/Network/HTTPClient.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2021, the Friendica project
   4  *
   5  * @license GNU APGL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Network;
  23
  24 use DOMDocument;
  25 use DomXPath;
  26 use Friendica\Core\Config\IConfig;
  27 use Friendica\Core\System;
  28 use Friendica\Util\Network;
  29 use Friendica\Util\Profiler;
  30 use GuzzleHttp\Client;
  31 use GuzzleHttp\Cookie\FileCookieJar;
  32 use GuzzleHttp\Exception\RequestException;
  33 use GuzzleHttp\Exception\TransferException;
  34 use GuzzleHttp\RequestOptions;
  35 use Psr\Http\Message\ResponseInterface;
  36 use Psr\Log\LoggerInterface;
  37
  38 /**
  39  * Performs HTTP requests to a given URL
  40  */
  41 class HTTPClient implements IHTTPClient
  42 {
  43         /** @var LoggerInterface */
  44         private $logger;
  45         /** @var Profiler */
  46         private $profiler;
  47         /** @var IConfig */
  48         private $config;
  49         /** @var string */
  50         private $userAgent;
  51         /** @var Client */
  52         private $client;
  53
  54         public function __construct(LoggerInterface $logger, Profiler $profiler, IConfig $config, string $userAgent, Client $client)
  55         {
  56                 $this->logger    = $logger;
  57                 $this->profiler  = $profiler;
  58                 $this->config    = $config;
  59                 $this->userAgent = $userAgent;
  60                 $this->client    = $client;
  61         }
  62
  63         protected function request(string $method, string $url, array $opts = [])
  64         {
  65                 $this->profiler->startRecording('network');
  66
  67                 if (Network::isLocalLink($url)) {
  68                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
  69                 }
  70
  71                 if (strlen($url) > 1000) {
  72                         $this->logger->debug('URL is longer than 1000 characters.', ['url' => $url, 'callstack' => System::callstack(20)]);
  73                         $this->profiler->stopRecording();
  74                         return CurlResult::createErrorCurl(substr($url, 0, 200));
  75                 }
  76
  77                 $parts2     = [];
  78                 $parts      = parse_url($url);
  79                 $path_parts = explode('/', $parts['path'] ?? '');
  80                 foreach ($path_parts as $part) {
  81                         if (strlen($part) <> mb_strlen($part)) {
  82                                 $parts2[] = rawurlencode($part);
  83                         } else {
  84                                 $parts2[] = $part;
  85                         }
  86                 }
  87                 $parts['path'] = implode('/', $parts2);
  88                 $url           = Network::unparseURL($parts);
  89
  90                 if (Network::isUrlBlocked($url)) {
  91                         $this->logger->info('Domain is blocked.', ['url' => $url]);
  92                         $this->profiler->stopRecording();
  93                         return CurlResult::createErrorCurl($url);
  94                 }
  95
  96                 $conf = [];
  97
  98                 if (!empty($opts['cookiejar'])) {
  99                         $jar = new FileCookieJar($opts['cookiejar']);
 100                         $conf[RequestOptions::COOKIES] = $jar;
 101                 }
 102
 103                 if (!empty($opts['accept_content'])) {
 104                         array_push($curlOptions[CURLOPT_HTTPHEADER], 'Accept: ' . $opts['accept_content']);
 105                 }
 106
 107                 if (!empty($opts['header'])) {
 108                         $curlOptions[CURLOPT_HTTPHEADER] = array_merge($opts['header'], $curlOptions[CURLOPT_HTTPHEADER]);
 109                 }
 110
 111                 $curlOptions[CURLOPT_USERAGENT] = $this->userAgent;
 112
 113                 if (!empty($opts['headers'])) {
 114                         $this->logger->notice('Wrong option \'headers\' used.');
 115                         $curlOptions[CURLOPT_HTTPHEADER] = array_merge($opts['headers'], $curlOptions[CURLOPT_HTTPHEADER]);
 116                 }
 117
 118                 if (!empty($opts['timeout'])) {
 119                         $curlOptions[CURLOPT_TIMEOUT] = $opts['timeout'];
 120                 }
 121
 122                 $onHeaders = function (ResponseInterface $response) use ($opts) {
 123                         if (!empty($opts['content_length']) &&
 124                                 $response->getHeaderLine('Content-Length') > $opts['content_length']) {
 125                                 throw new TransferException('The file is too big!');
 126                         }
 127                 };
 128
 129                 try {
 130                         $response = $this->client->$method($url, [
 131                                 'on_headers' => $onHeaders,
 132                                 'curl'       => $curlOptions,
 133                         ]);
 134                         return new GuzzleResponse($response, $url);
 135                 } catch (TransferException $exception) {
 136                         if ($exception instanceof RequestException &&
 137                                 $exception->hasResponse()) {
 138                                 return new GuzzleResponse($exception->getResponse(), $url, $exception->getCode(), '');
 139                         } else {
 140                                 return new CurlResult($url, '', ['http_code' => $exception->getCode()], $exception->getCode(), '');
 141                         }
 142                 } finally {
 143                         $this->profiler->stopRecording();
 144                 }
 145         }
 146
 147         /** {@inheritDoc}
 148          *
 149          * @throws HTTPException\InternalServerErrorException
 150          */
 151         public function head(string $url, array $opts = [])
 152         {
 153                 return $this->request('head', $url, $opts);
 154         }
 155
 156         /**
 157          * {@inheritDoc}
 158          */
 159         public function get(string $url, array $opts = [])
 160         {
 161                 return $this->request('get', $url, $opts);
 162         }
 163
 164         /**
 165          * {@inheritDoc}
 166          *
 167          * @param int $redirects The recursion counter for internal use - default 0
 168          *
 169          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 170          */
 171         public function post(string $url, $params, array $headers = [], int $timeout = 0, &$redirects = 0)
 172         {
 173                 $this->profiler->startRecording('network');
 174
 175                 if (Network::isLocalLink($url)) {
 176                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
 177                 }
 178
 179                 if (Network::isUrlBlocked($url)) {
 180                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 181                         $this->profiler->stopRecording();
 182                         return CurlResult::createErrorCurl($url);
 183                 }
 184
 185                 $ch = curl_init($url);
 186
 187                 if (($redirects > 8) || (!$ch)) {
 188                         $this->profiler->stopRecording();
 189                         return CurlResult::createErrorCurl($url);
 190                 }
 191
 192                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 193
 194                 curl_setopt($ch, CURLOPT_HEADER, true);
 195                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 196                 curl_setopt($ch, CURLOPT_POST, 1);
 197                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 198                 curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent);
 199
 200                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 201                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 202                 }
 203
 204                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 205
 206                 if (intval($timeout)) {
 207                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 208                 } else {
 209                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 210                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 211                 }
 212
 213                 if (!empty($headers)) {
 214                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 215                 }
 216
 217                 $check_cert = $this->config->get('system', 'verifyssl');
 218                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 219
 220                 if ($check_cert) {
 221                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 222                 }
 223
 224                 $proxy = $this->config->get('system', 'proxy');
 225
 226                 if (!empty($proxy)) {
 227                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 228                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 229                         $proxyuser = $this->config->get('system', 'proxyuser');
 230                         if (!empty($proxyuser)) {
 231                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 232                         }
 233                 }
 234
 235                 // don't let curl abort the entire application
 236                 // if it throws any errors.
 237
 238                 $s = @curl_exec($ch);
 239
 240                 $curl_info = curl_getinfo($ch);
 241
 242                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 243
 244                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 245                         $redirects++;
 246                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 247                         curl_close($ch);
 248                         $this->profiler->stopRecording();
 249                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 250                 }
 251
 252                 curl_close($ch);
 253
 254                 $this->profiler->stopRecording();
 255
 256                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 257                 if ($curlResponse->getReturnCode() == 417) {
 258                         $redirects++;
 259
 260                         if (empty($headers)) {
 261                                 $headers = ['Expect:'];
 262                         } else {
 263                                 if (!in_array('Expect:', $headers)) {
 264                                         array_push($headers, 'Expect:');
 265                                 }
 266                         }
 267                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 268                         return $this->post($url, $params, $headers, $redirects, $timeout);
 269                 }
 270
 271                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 272
 273                 return $curlResponse;
 274         }
 275
 276         /**
 277          * {@inheritDoc}
 278          */
 279         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 280         {
 281                 if (Network::isLocalLink($url)) {
 282                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
 283                 }
 284
 285                 if (Network::isUrlBlocked($url)) {
 286                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 287                         return $url;
 288                 }
 289
 290                 if (Network::isRedirectBlocked($url)) {
 291                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 292                         return $url;
 293                 }
 294
 295                 $url = Network::stripTrackingQueryParams($url);
 296
 297                 if ($depth > 10) {
 298                         return $url;
 299                 }
 300
 301                 $url = trim($url, "'");
 302
 303                 $this->profiler->startRecording('network');
 304
 305                 $ch = curl_init();
 306                 curl_setopt($ch, CURLOPT_URL, $url);
 307                 curl_setopt($ch, CURLOPT_HEADER, 1);
 308                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 309                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 310                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 311                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 312                 curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent);
 313
 314                 curl_exec($ch);
 315                 $curl_info = @curl_getinfo($ch);
 316                 $http_code = $curl_info['http_code'];
 317                 curl_close($ch);
 318
 319                 $this->profiler->stopRecording();
 320
 321                 if ($http_code == 0) {
 322                         return $url;
 323                 }
 324
 325                 if (in_array($http_code, ['301', '302'])) {
 326                         if (!empty($curl_info['redirect_url'])) {
 327                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 328                         } elseif (!empty($curl_info['location'])) {
 329                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 330                         }
 331                 }
 332
 333                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 334                 if (!$fetchbody) {
 335                         return $this->finalUrl($url, ++$depth, true);
 336                 }
 337
 338                 // if the file is too large then exit
 339                 if ($curl_info["download_content_length"] > 1000000) {
 340                         return $url;
 341                 }
 342
 343                 // if it isn't a HTML file then exit
 344                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 345                         return $url;
 346                 }
 347
 348                 $this->profiler->startRecording('network');
 349
 350                 $ch = curl_init();
 351                 curl_setopt($ch, CURLOPT_URL, $url);
 352                 curl_setopt($ch, CURLOPT_HEADER, 0);
 353                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 354                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 355                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 356                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 357                 curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent);
 358
 359                 $body = curl_exec($ch);
 360                 curl_close($ch);
 361
 362                 $this->profiler->stopRecording();
 363
 364                 if (trim($body) == "") {
 365                         return $url;
 366                 }
 367
 368                 // Check for redirect in meta elements
 369                 $doc = new DOMDocument();
 370                 @$doc->loadHTML($body);
 371
 372                 $xpath = new DomXPath($doc);
 373
 374                 $list = $xpath->query("//meta[@content]");
 375                 foreach ($list as $node) {
 376                         $attr = [];
 377                         if ($node->attributes->length) {
 378                                 foreach ($node->attributes as $attribute) {
 379                                         $attr[$attribute->name] = $attribute->value;
 380                                 }
 381                         }
 382
 383                         if (@$attr["http-equiv"] == 'refresh') {
 384                                 $path = $attr["content"];
 385                                 $pathinfo = explode(";", $path);
 386                                 foreach ($pathinfo as $value) {
 387                                         if (substr(strtolower($value), 0, 4) == "url=") {
 388                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 389                                         }
 390                                 }
 391                         }
 392                 }
 393
 394                 return $url;
 395         }
 396
 397         /**
 398          * {@inheritDoc}
 399          */
 400         public function fetch(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '')
 401         {
 402                 $ret = $this->fetchFull($url, $timeout, $accept_content, $cookiejar);
 403
 404                 return $ret->getBody();
 405         }
 406
 407         /**
 408          * {@inheritDoc}
 409          */
 410         public function fetchFull(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '')
 411         {
 412                 return $this->get(
 413                         $url,
 414                         [
 415                                 'timeout'        => $timeout,
 416                                 'accept_content' => $accept_content,
 417                                 'cookiejar'      => $cookiejar
 418                         ]
 419                 );
 420         }
 421 }