src/Network/HTTPRequest.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\App;
  27 use Friendica\Core\Config\IConfig;
  28 use Friendica\Core\System;
  29 use Friendica\Util\Network;
  30 use Friendica\Util\Profiler;
  31 use Psr\Log\LoggerInterface;
  32
  33 /**
  34  * Performs HTTP requests to a given URL
  35  */
  36 class HTTPRequest implements IHTTPRequest
  37 {
  38         /** @var LoggerInterface */
  39         private $logger;
  40         /** @var Profiler */
  41         private $profiler;
  42         /** @var IConfig */
  43         private $config;
  44         /** @var string */
  45         private $baseUrl;
  46
  47         public function __construct(LoggerInterface $logger, Profiler $profiler, IConfig $config, App\BaseURL $baseUrl)
  48         {
  49                 $this->logger   = $logger;
  50                 $this->profiler = $profiler;
  51                 $this->config   = $config;
  52                 $this->baseUrl  = $baseUrl->get();
  53         }
  54
  55         /** {@inheritDoc}
  56          *
  57          * @throws HTTPException\InternalServerErrorException
  58          */
  59         public function head(string $url, array $opts = [])
  60         {
  61                 $opts['nobody'] = true;
  62
  63                 return $this->get($url, $opts);
  64         }
  65
  66         /**
  67          * {@inheritDoc}
  68          *
  69          * @param int $redirects The recursion counter for internal use - default 0
  70          *
  71          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  72          */
  73         public function get(string $url, array $opts = [], &$redirects = 0)
  74         {
  75                 $this->profiler->startRecording('network');
  76
  77                 if (Network::isLocalLink($url)) {
  78                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
  79                 }
  80
  81                 if (strlen($url) > 1000) {
  82                         $this->logger->debug('URL is longer than 1000 characters.', ['url' => $url, 'callstack' => System::callstack(20)]);
  83                         $this->profiler->stopRecording();
  84                         return CurlResult::createErrorCurl(substr($url, 0, 200));
  85                 }
  86
  87                 $parts2     = [];
  88                 $parts      = parse_url($url);
  89                 $path_parts = explode('/', $parts['path'] ?? '');
  90                 foreach ($path_parts as $part) {
  91                         if (strlen($part) <> mb_strlen($part)) {
  92                                 $parts2[] = rawurlencode($part);
  93                         } else {
  94                                 $parts2[] = $part;
  95                         }
  96                 }
  97                 $parts['path'] = implode('/', $parts2);
  98                 $url           = Network::unparseURL($parts);
  99
 100                 if (Network::isUrlBlocked($url)) {
 101                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 102                         $this->profiler->stopRecording();
 103                         return CurlResult::createErrorCurl($url);
 104                 }
 105
 106                 $ch = @curl_init($url);
 107
 108                 if (($redirects > 8) || (!$ch)) {
 109                         $this->profiler->stopRecording();
 110                         return CurlResult::createErrorCurl($url);
 111                 }
 112
 113                 @curl_setopt($ch, CURLOPT_HEADER, true);
 114
 115                 if (!empty($opts['cookiejar'])) {
 116                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
 117                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
 118                 }
 119
 120                 // These settings aren't needed. We're following the location already.
 121                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 122                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 123
 124                 if (!empty($opts['accept_content'])) {
 125                         curl_setopt(
 126                                 $ch,
 127                                 CURLOPT_HTTPHEADER,
 128                                 ['Accept: ' . $opts['accept_content']]
 129                         );
 130                 }
 131
 132                 if (!empty($opts['header'])) {
 133                         curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['header']);
 134                 }
 135
 136                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 137                 @curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 138
 139                 $range = intval($this->config->get('system', 'curl_range_bytes', 0));
 140
 141                 if ($range > 0) {
 142                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 143                 }
 144
 145                 // Without this setting it seems as if some webservers send compressed content
 146                 // This seems to confuse curl so that it shows this uncompressed.
 147                 /// @todo  We could possibly set this value to "gzip" or something similar
 148                 curl_setopt($ch, CURLOPT_ENCODING, '');
 149
 150                 if (!empty($opts['headers'])) {
 151                         $this->logger->notice('Wrong option \'headers\' used.');
 152                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 153                 }
 154
 155                 if (!empty($opts['nobody'])) {
 156                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 157                 }
 158
 159                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 160
 161                 if (!empty($opts['timeout'])) {
 162                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 163                 } else {
 164                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 165                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 166                 }
 167
 168                 // by default we will allow self-signed certs
 169                 // but you can override this
 170
 171                 $check_cert = $this->config->get('system', 'verifyssl');
 172                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 173
 174                 if ($check_cert) {
 175                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 176                 }
 177
 178                 $proxy = $this->config->get('system', 'proxy');
 179
 180                 if (!empty($proxy)) {
 181                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 182                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 183                         $proxyuser = $this->config->get('system', 'proxyuser');
 184
 185                         if (!empty($proxyuser)) {
 186                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 187                         }
 188                 }
 189
 190                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 191                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 192                 }
 193
 194                 $s         = @curl_exec($ch);
 195                 $curl_info = @curl_getinfo($ch);
 196
 197                 // Special treatment for HTTP Code 416
 198                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 199                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 200                         @curl_setopt($ch, CURLOPT_RANGE, '');
 201                         $s         = @curl_exec($ch);
 202                         $curl_info = @curl_getinfo($ch);
 203                 }
 204
 205                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 206
 207                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 208                         $redirects++;
 209                         $this->logger->notice('Curl redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 210                         @curl_close($ch);
 211                         $this->profiler->stopRecording();
 212                         return $this->get($curlResponse->getRedirectUrl(), $opts, $redirects);
 213                 }
 214
 215                 @curl_close($ch);
 216
 217                 $this->profiler->stopRecording();
 218
 219                 return $curlResponse;
 220         }
 221
 222         /**
 223          * {@inheritDoc}
 224          *
 225          * @param int $redirects The recursion counter for internal use - default 0
 226          *
 227          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 228          */
 229         public function post(string $url, $params, array $headers = [], int $timeout = 0, &$redirects = 0)
 230         {
 231                 $this->profiler->startRecording('network');
 232
 233                 if (Network::isLocalLink($url)) {
 234                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
 235                 }
 236
 237                 if (Network::isUrlBlocked($url)) {
 238                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 239                         $this->profiler->stopRecording();
 240                         return CurlResult::createErrorCurl($url);
 241                 }
 242
 243                 $ch = curl_init($url);
 244
 245                 if (($redirects > 8) || (!$ch)) {
 246                         $this->profiler->stopRecording();
 247                         return CurlResult::createErrorCurl($url);
 248                 }
 249
 250                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 251
 252                 curl_setopt($ch, CURLOPT_HEADER, true);
 253                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 254                 curl_setopt($ch, CURLOPT_POST, 1);
 255                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 256                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 257
 258                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 259                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 260                 }
 261
 262                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 263
 264                 if (intval($timeout)) {
 265                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 266                 } else {
 267                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 268                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 269                 }
 270
 271                 if (!empty($headers)) {
 272                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 273                 }
 274
 275                 $check_cert = $this->config->get('system', 'verifyssl');
 276                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 277
 278                 if ($check_cert) {
 279                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 280                 }
 281
 282                 $proxy = $this->config->get('system', 'proxy');
 283
 284                 if (!empty($proxy)) {
 285                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 286                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 287                         $proxyuser = $this->config->get('system', 'proxyuser');
 288                         if (!empty($proxyuser)) {
 289                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 290                         }
 291                 }
 292
 293                 // don't let curl abort the entire application
 294                 // if it throws any errors.
 295
 296                 $s = @curl_exec($ch);
 297
 298                 $curl_info = curl_getinfo($ch);
 299
 300                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 301
 302                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 303                         $redirects++;
 304                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 305                         curl_close($ch);
 306                         $this->profiler->stopRecording();
 307                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 308                 }
 309
 310                 curl_close($ch);
 311
 312                 $this->profiler->stopRecording();
 313
 314                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 315                 if ($curlResponse->getReturnCode() == 417) {
 316                         $redirects++;
 317
 318                         if (empty($headers)) {
 319                                 $headers = ['Expect:'];
 320                         } else {
 321                                 if (!in_array('Expect:', $headers)) {
 322                                         array_push($headers, 'Expect:');
 323                                 }
 324                         }
 325                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 326                         return $this->post($url, $params, $headers, $redirects, $timeout);
 327                 }
 328
 329                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 330
 331                 return $curlResponse;
 332         }
 333
 334         /**
 335          * {@inheritDoc}
 336          */
 337         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 338         {
 339                 if (Network::isLocalLink($url)) {
 340                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
 341                 }
 342
 343                 if (Network::isUrlBlocked($url)) {
 344                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 345                         return $url;
 346                 }
 347
 348                 if (Network::isRedirectBlocked($url)) {
 349                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 350                         return $url;
 351                 }
 352
 353                 $url = Network::stripTrackingQueryParams($url);
 354
 355                 if ($depth > 10) {
 356                         return $url;
 357                 }
 358
 359                 $url = trim($url, "'");
 360
 361                 $this->profiler->startRecording('network');
 362
 363                 $ch = curl_init();
 364                 curl_setopt($ch, CURLOPT_URL, $url);
 365                 curl_setopt($ch, CURLOPT_HEADER, 1);
 366                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 367                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 368                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 369                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 370                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 371
 372                 curl_exec($ch);
 373                 $curl_info = @curl_getinfo($ch);
 374                 $http_code = $curl_info['http_code'];
 375                 curl_close($ch);
 376
 377                 $this->profiler->stopRecording();
 378
 379                 if ($http_code == 0) {
 380                         return $url;
 381                 }
 382
 383                 if (in_array($http_code, ['301', '302'])) {
 384                         if (!empty($curl_info['redirect_url'])) {
 385                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 386                         } elseif (!empty($curl_info['location'])) {
 387                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 388                         }
 389                 }
 390
 391                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 392                 if (!$fetchbody) {
 393                         return $this->finalUrl($url, ++$depth, true);
 394                 }
 395
 396                 // if the file is too large then exit
 397                 if ($curl_info["download_content_length"] > 1000000) {
 398                         return $url;
 399                 }
 400
 401                 // if it isn't a HTML file then exit
 402                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 403                         return $url;
 404                 }
 405
 406                 $this->profiler->startRecording('network');
 407
 408                 $ch = curl_init();
 409                 curl_setopt($ch, CURLOPT_URL, $url);
 410                 curl_setopt($ch, CURLOPT_HEADER, 0);
 411                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 412                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 413                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 414                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 415                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 416
 417                 $body = curl_exec($ch);
 418                 curl_close($ch);
 419
 420                 $this->profiler->stopRecording();
 421
 422                 if (trim($body) == "") {
 423                         return $url;
 424                 }
 425
 426                 // Check for redirect in meta elements
 427                 $doc = new DOMDocument();
 428                 @$doc->loadHTML($body);
 429
 430                 $xpath = new DomXPath($doc);
 431
 432                 $list = $xpath->query("//meta[@content]");
 433                 foreach ($list as $node) {
 434                         $attr = [];
 435                         if ($node->attributes->length) {
 436                                 foreach ($node->attributes as $attribute) {
 437                                         $attr[$attribute->name] = $attribute->value;
 438                                 }
 439                         }
 440
 441                         if (@$attr["http-equiv"] == 'refresh') {
 442                                 $path = $attr["content"];
 443                                 $pathinfo = explode(";", $path);
 444                                 foreach ($pathinfo as $value) {
 445                                         if (substr(strtolower($value), 0, 4) == "url=") {
 446                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 447                                         }
 448                                 }
 449                         }
 450                 }
 451
 452                 return $url;
 453         }
 454
 455         /**
 456          * {@inheritDoc}
 457          *
 458          * @param int $redirects The recursion counter for internal use - default 0
 459          *
 460          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 461          */
 462         public function fetch(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '', &$redirects = 0)
 463         {
 464                 $ret = $this->fetchFull($url, $timeout, $accept_content, $cookiejar, $redirects);
 465
 466                 return $ret->getBody();
 467         }
 468
 469         /**
 470          * {@inheritDoc}
 471          *
 472          * @param int $redirects The recursion counter for internal use - default 0
 473          *
 474          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 475          */
 476         public function fetchFull(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '', &$redirects = 0)
 477         {
 478                 return $this->get(
 479                         $url,
 480                         [
 481                                 'timeout'        => $timeout,
 482                                 'accept_content' => $accept_content,
 483                                 'cookiejar'      => $cookiejar
 484                         ],
 485                         $redirects
 486                 );
 487         }
 488
 489         /**
 490          * {@inheritDoc}
 491          */
 492         public function getUserAgent()
 493         {
 494                 return
 495                         FRIENDICA_PLATFORM . " '" .
 496                         FRIENDICA_CODENAME . "' " .
 497                         FRIENDICA_VERSION . '-' .
 498                         DB_UPDATE_VERSION . '; ' .
 499                         $this->baseUrl;
 500         }
 501 }