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 GuzzleHttp\Client;
  32 use GuzzleHttp\Exception\RequestException;
  33 use GuzzleHttp\Exception\TransferException;
  34 use Psr\Http\Message\RequestInterface;
  35 use Psr\Http\Message\ResponseInterface;
  36 use Psr\Http\Message\UriInterface;
  37 use Psr\Log\LoggerInterface;
  38
  39 /**
  40  * Performs HTTP requests to a given URL
  41  */
  42 class HTTPRequest implements IHTTPRequest
  43 {
  44         /** @var LoggerInterface */
  45         private $logger;
  46         /** @var Profiler */
  47         private $profiler;
  48         /** @var IConfig */
  49         private $config;
  50         /** @var string */
  51         private $baseUrl;
  52
  53         public function __construct(LoggerInterface $logger, Profiler $profiler, IConfig $config, App\BaseURL $baseUrl)
  54         {
  55                 $this->logger   = $logger;
  56                 $this->profiler = $profiler;
  57                 $this->config   = $config;
  58                 $this->baseUrl  = $baseUrl->get();
  59         }
  60
  61         /** {@inheritDoc}
  62          *
  63          * @throws HTTPException\InternalServerErrorException
  64          */
  65         public function head(string $url, array $opts = [])
  66         {
  67                 $opts['nobody'] = true;
  68
  69                 return $this->get($url, $opts);
  70         }
  71
  72         /**
  73          * {@inheritDoc}
  74          */
  75         public function get(string $url, bool $binary = false, array $opts = [])
  76         {
  77                 $this->profiler->startRecording('network');
  78
  79                 if (Network::isLocalLink($url)) {
  80                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
  81                 }
  82
  83                 if (strlen($url) > 1000) {
  84                         $this->logger->debug('URL is longer than 1000 characters.', ['url' => $url, 'callstack' => System::callstack(20)]);
  85                         $this->profiler->stopRecording();
  86                         return CurlResult::createErrorCurl(substr($url, 0, 200));
  87                 }
  88
  89                 $parts2     = [];
  90                 $parts      = parse_url($url);
  91                 $path_parts = explode('/', $parts['path'] ?? '');
  92                 foreach ($path_parts as $part) {
  93                         if (strlen($part) <> mb_strlen($part)) {
  94                                 $parts2[] = rawurlencode($part);
  95                         } else {
  96                                 $parts2[] = $part;
  97                         }
  98                 }
  99                 $parts['path'] = implode('/', $parts2);
 100                 $url           = Network::unparseURL($parts);
 101
 102                 if (Network::isUrlBlocked($url)) {
 103                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 104                         $this->profiler->stopRecording();
 105                         return CurlResult::createErrorCurl($url);
 106                 }
 107
 108                 $curlOptions = [];
 109
 110                 $curlOptions[CURLOPT_HEADER] = true;
 111
 112                 if (!empty($opts['cookiejar'])) {
 113                         $curlOptions[CURLOPT_COOKIEJAR] = $opts["cookiejar"];
 114                         $curlOptions[CURLOPT_COOKIEFILE] = $opts["cookiejar"];
 115                 }
 116
 117                 // These settings aren't needed. We're following the location already.
 118                 //      $curlOptions[CURLOPT_FOLLOWLOCATION] =true;
 119                 //      $curlOptions[CURLOPT_MAXREDIRS] = 5;
 120
 121                 if (!empty($opts['accept_content'])) {
 122                         if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
 123                                 $curlOptions[CURLOPT_HTTPHEADER] = [];
 124                         }
 125                         array_push($curlOptions[CURLOPT_HTTPHEADER], 'Accept: ' . $opts['accept_content']);
 126                 }
 127
 128                 if (!empty($opts['header'])) {
 129                         if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
 130                                 $curlOptions[CURLOPT_HTTPHEADER] = [];
 131                         }
 132                         $curlOptions[CURLOPT_HTTPHEADER] = array_merge($opts['header'], $curlOptions[CURLOPT_HTTPHEADER]);
 133                 }
 134
 135                 $curlOptions[CURLOPT_RETURNTRANSFER] = true;
 136                 $curlOptions[CURLOPT_USERAGENT] = $this->getUserAgent();
 137
 138                 $range = intval($this->config->get('system', 'curl_range_bytes', 0));
 139
 140                 if ($range > 0) {
 141                         $curlOptions[CURLOPT_RANGE] = '0-' . $range;
 142                 }
 143
 144                 // Without this setting it seems as if some webservers send compressed content
 145                 // This seems to confuse curl so that it shows this uncompressed.
 146                 /// @todo  We could possibly set this value to "gzip" or something similar
 147                 $curlOptions[CURLOPT_ENCODING] = '';
 148
 149                 if (!empty($opts['headers'])) {
 150                         $this->logger->notice('Wrong option \'headers\' used.');
 151                         if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
 152                                 $curlOptions[CURLOPT_HTTPHEADER] = [];
 153                         }
 154                         $curlOptions[CURLOPT_HTTPHEADER] = array_merge($opts['headers'], $curlOptions[CURLOPT_HTTPHEADER]);
 155                 }
 156
 157                 if (!empty($opts['nobody'])) {
 158                         $curlOptions[CURLOPT_NOBODY] = $opts['nobody'];
 159                 }
 160
 161                 $curlOptions[CURLOPT_CONNECTTIMEOUT] = 10;
 162
 163                 if (!empty($opts['timeout'])) {
 164                         $curlOptions[CURLOPT_TIMEOUT] = $opts['timeout'];
 165                 } else {
 166                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 167                         $curlOptions[CURLOPT_TIMEOUT] = intval($curl_time);
 168                 }
 169
 170                 // by default we will allow self-signed certs
 171                 // but you can override this
 172
 173                 $check_cert = $this->config->get('system', 'verifyssl');
 174                 $curlOptions[CURLOPT_SSL_VERIFYPEER] = ($check_cert) ? true : false;
 175
 176                 if ($check_cert) {
 177                         $curlOptions[CURLOPT_SSL_VERIFYHOST] = 2;
 178                 }
 179
 180                 $proxy = $this->config->get('system', 'proxy');
 181
 182                 if (!empty($proxy)) {
 183                         $curlOptions[CURLOPT_HTTPPROXYTUNNEL] = 1;
 184                         $curlOptions[CURLOPT_PROXY] = $proxy;
 185                         $proxyuser = $this->config->get('system', 'proxyuser');
 186
 187                         if (!empty($proxyuser)) {
 188                                 $curlOptions[CURLOPT_PROXYUSERPWD] = $proxyuser;
 189                         }
 190                 }
 191
 192                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 193                         $curlOptions[CURLOPT_IPRESOLVE] = CURL_IPRESOLVE_V4;
 194                 }
 195
 196                 if ($binary) {
 197                         $curlOptions[CURLOPT_BINARYTRANSFER] = 1;
 198                 }
 199
 200                 $onRedirect = function(
 201                         RequestInterface $request,
 202                         ResponseInterface $response,
 203                         UriInterface $uri
 204                 ) {
 205                         $this->logger->notice('Curl redirect.', ['url' => $request->getUri(), 'to' => $uri]);
 206                 };
 207
 208                 $onHeaders = function (ResponseInterface $response) use ($opts) {
 209                         if (!empty($opts['content_length']) &&
 210                                 $response->getHeaderLine('Content-Length') > $opts['content_length']) {
 211                                 throw new TransferException('The file is too big!');
 212                         }
 213                 };
 214
 215                 $client = new Client([
 216                         'allow_redirect' => [
 217                                 'max' => 8,
 218                                 'on_redirect' => $onRedirect,
 219                                 'track_redirect' => true,
 220                                 'strict' => true,
 221                                 'referer' => true,
 222                         ],
 223                         'on_headers' => $onHeaders,
 224                         'sink' => tempnam(get_temppath(), 'guzzle'),
 225                         'curl' => $curlOptions
 226                 ]);
 227
 228                 try {
 229                         $response = $client->get($url);
 230                         return new GuzzleResponse($response, $url);
 231                 } catch (TransferException $exception) {
 232                         if ($exception instanceof RequestException &&
 233                                 $exception->hasResponse()) {
 234                                 return new GuzzleResponse($exception->getResponse(), $url, $exception->getCode(), $exception->getMessage());
 235                         } else {
 236                                 return new CurlResult($url, '', ['http_code' => $exception->getCode()], $exception->getCode(), $exception->getMessage());
 237                         }
 238                 } finally {
 239                         $this->profiler->stopRecording();
 240                 }
 241         }
 242
 243         /**
 244          * {@inheritDoc}
 245          *
 246          * @param int $redirects The recursion counter for internal use - default 0
 247          *
 248          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 249          */
 250         public function post(string $url, $params, array $headers = [], int $timeout = 0, &$redirects = 0)
 251         {
 252                 $this->profiler->startRecording('network');
 253
 254                 if (Network::isLocalLink($url)) {
 255                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
 256                 }
 257
 258                 if (Network::isUrlBlocked($url)) {
 259                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 260                         $this->profiler->stopRecording();
 261                         return CurlResult::createErrorCurl($url);
 262                 }
 263
 264                 $ch = curl_init($url);
 265
 266                 if (($redirects > 8) || (!$ch)) {
 267                         $this->profiler->stopRecording();
 268                         return CurlResult::createErrorCurl($url);
 269                 }
 270
 271                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 272
 273                 curl_setopt($ch, CURLOPT_HEADER, true);
 274                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 275                 curl_setopt($ch, CURLOPT_POST, 1);
 276                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 277                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 278
 279                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 280                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 281                 }
 282
 283                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 284
 285                 if (intval($timeout)) {
 286                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 287                 } else {
 288                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 289                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 290                 }
 291
 292                 if (!empty($headers)) {
 293                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 294                 }
 295
 296                 $check_cert = $this->config->get('system', 'verifyssl');
 297                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 298
 299                 if ($check_cert) {
 300                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 301                 }
 302
 303                 $proxy = $this->config->get('system', 'proxy');
 304
 305                 if (!empty($proxy)) {
 306                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 307                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 308                         $proxyuser = $this->config->get('system', 'proxyuser');
 309                         if (!empty($proxyuser)) {
 310                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 311                         }
 312                 }
 313
 314                 // don't let curl abort the entire application
 315                 // if it throws any errors.
 316
 317                 $s = @curl_exec($ch);
 318
 319                 $curl_info = curl_getinfo($ch);
 320
 321                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 322
 323                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 324                         $redirects++;
 325                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 326                         curl_close($ch);
 327                         $this->profiler->stopRecording();
 328                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 329                 }
 330
 331                 curl_close($ch);
 332
 333                 $this->profiler->stopRecording();
 334
 335                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 336                 if ($curlResponse->getReturnCode() == 417) {
 337                         $redirects++;
 338
 339                         if (empty($headers)) {
 340                                 $headers = ['Expect:'];
 341                         } else {
 342                                 if (!in_array('Expect:', $headers)) {
 343                                         array_push($headers, 'Expect:');
 344                                 }
 345                         }
 346                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 347                         return $this->post($url, $params, $headers, $redirects, $timeout);
 348                 }
 349
 350                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 351
 352                 return $curlResponse;
 353         }
 354
 355         /**
 356          * {@inheritDoc}
 357          */
 358         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 359         {
 360                 if (Network::isLocalLink($url)) {
 361                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
 362                 }
 363
 364                 if (Network::isUrlBlocked($url)) {
 365                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 366                         return $url;
 367                 }
 368
 369                 if (Network::isRedirectBlocked($url)) {
 370                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 371                         return $url;
 372                 }
 373
 374                 $url = Network::stripTrackingQueryParams($url);
 375
 376                 if ($depth > 10) {
 377                         return $url;
 378                 }
 379
 380                 $url = trim($url, "'");
 381
 382                 $this->profiler->startRecording('network');
 383
 384                 $ch = curl_init();
 385                 curl_setopt($ch, CURLOPT_URL, $url);
 386                 curl_setopt($ch, CURLOPT_HEADER, 1);
 387                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 388                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 389                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 390                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 391                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 392
 393                 curl_exec($ch);
 394                 $curl_info = @curl_getinfo($ch);
 395                 $http_code = $curl_info['http_code'];
 396                 curl_close($ch);
 397
 398                 $this->profiler->stopRecording();
 399
 400                 if ($http_code == 0) {
 401                         return $url;
 402                 }
 403
 404                 if (in_array($http_code, ['301', '302'])) {
 405                         if (!empty($curl_info['redirect_url'])) {
 406                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 407                         } elseif (!empty($curl_info['location'])) {
 408                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 409                         }
 410                 }
 411
 412                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 413                 if (!$fetchbody) {
 414                         return $this->finalUrl($url, ++$depth, true);
 415                 }
 416
 417                 // if the file is too large then exit
 418                 if ($curl_info["download_content_length"] > 1000000) {
 419                         return $url;
 420                 }
 421
 422                 // if it isn't a HTML file then exit
 423                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 424                         return $url;
 425                 }
 426
 427                 $this->profiler->startRecording('network');
 428
 429                 $ch = curl_init();
 430                 curl_setopt($ch, CURLOPT_URL, $url);
 431                 curl_setopt($ch, CURLOPT_HEADER, 0);
 432                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 433                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 434                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 435                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 436                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 437
 438                 $body = curl_exec($ch);
 439                 curl_close($ch);
 440
 441                 $this->profiler->stopRecording();
 442
 443                 if (trim($body) == "") {
 444                         return $url;
 445                 }
 446
 447                 // Check for redirect in meta elements
 448                 $doc = new DOMDocument();
 449                 @$doc->loadHTML($body);
 450
 451                 $xpath = new DomXPath($doc);
 452
 453                 $list = $xpath->query("//meta[@content]");
 454                 foreach ($list as $node) {
 455                         $attr = [];
 456                         if ($node->attributes->length) {
 457                                 foreach ($node->attributes as $attribute) {
 458                                         $attr[$attribute->name] = $attribute->value;
 459                                 }
 460                         }
 461
 462                         if (@$attr["http-equiv"] == 'refresh') {
 463                                 $path = $attr["content"];
 464                                 $pathinfo = explode(";", $path);
 465                                 foreach ($pathinfo as $value) {
 466                                         if (substr(strtolower($value), 0, 4) == "url=") {
 467                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 468                                         }
 469                                 }
 470                         }
 471                 }
 472
 473                 return $url;
 474         }
 475
 476         /**
 477          * {@inheritDoc}
 478          *
 479          * @param int $redirects The recursion counter for internal use - default 0
 480          *
 481          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 482          */
 483         public function fetch(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '', &$redirects = 0)
 484         {
 485                 $ret = $this->fetchFull($url, $timeout, $accept_content, $cookiejar, $redirects);
 486
 487                 return $ret->getBody();
 488         }
 489
 490         /**
 491          * {@inheritDoc}
 492          *
 493          * @param int $redirects The recursion counter for internal use - default 0
 494          *
 495          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 496          */
 497         public function fetchFull(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '', &$redirects = 0)
 498         {
 499                 return $this->get(
 500                         $url,
 501                         [
 502                                 'timeout'        => $timeout,
 503                                 'accept_content' => $accept_content,
 504                                 'cookiejar'      => $cookiejar
 505                         ]
 506                 );
 507         }
 508
 509         /**
 510          * {@inheritDoc}
 511          */
 512         public function getUserAgent()
 513         {
 514                 return
 515                         FRIENDICA_PLATFORM . " '" .
 516                         FRIENDICA_CODENAME . "' " .
 517                         FRIENDICA_VERSION . '-' .
 518                         DB_UPDATE_VERSION . '; ' .
 519                         $this->baseUrl;
 520         }
 521 }