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                 $logger = $this->logger;
 201
 202                 $onRedirect = function(
 203                         RequestInterface $request,
 204                         ResponseInterface $response,
 205                         UriInterface $uri
 206                 ) use ($logger) {
 207                         $logger->notice('Curl redirect.', ['url' => $request->getUri(), 'to' => $uri]);
 208                 };
 209
 210                 $onHeaders = function (ResponseInterface $response) use ($opts) {
 211                         if (!empty($opts['content_length']) &&
 212                                 $response->getHeaderLine('Content-Length') > $opts['content_length']) {
 213                                 throw new TransferException('The file is too big!');
 214                         }
 215                 };
 216
 217                 $client = new Client([
 218                         'allow_redirect' => [
 219                                 'max' => 8,
 220                                 'on_redirect' => $onRedirect,
 221                                 'track_redirect' => true,
 222                                 'strict' => true,
 223                                 'referer' => true,
 224                         ],
 225                         'on_headers' => $onHeaders,
 226                         'sink' => tempnam(get_temppath(), 'guzzle'),
 227                         'curl' => $curlOptions
 228                 ]);
 229
 230                 try {
 231                         $response = $client->get($url);
 232                         return new GuzzleResponse($response, $url);
 233                 } catch (TransferException $exception) {
 234                         if ($exception instanceof RequestException &&
 235                                 $exception->hasResponse()) {
 236                                 return new GuzzleResponse($exception->getResponse(), $url, $exception->getCode(), $exception->getMessage());
 237                         } else {
 238                                 return new CurlResult($url, '', ['http_code' => $exception->getCode()], $exception->getCode(), $exception->getMessage());
 239                         }
 240                 } finally {
 241                         $this->profiler->stopRecording();
 242                 }
 243         }
 244
 245         /**
 246          * {@inheritDoc}
 247          *
 248          * @param int $redirects The recursion counter for internal use - default 0
 249          *
 250          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 251          */
 252         public function post(string $url, $params, array $headers = [], int $timeout = 0, &$redirects = 0)
 253         {
 254                 $this->profiler->startRecording('network');
 255
 256                 if (Network::isLocalLink($url)) {
 257                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
 258                 }
 259
 260                 if (Network::isUrlBlocked($url)) {
 261                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 262                         $this->profiler->stopRecording();
 263                         return CurlResult::createErrorCurl($url);
 264                 }
 265
 266                 $ch = curl_init($url);
 267
 268                 if (($redirects > 8) || (!$ch)) {
 269                         $this->profiler->stopRecording();
 270                         return CurlResult::createErrorCurl($url);
 271                 }
 272
 273                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 274
 275                 curl_setopt($ch, CURLOPT_HEADER, true);
 276                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 277                 curl_setopt($ch, CURLOPT_POST, 1);
 278                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 279                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 280
 281                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 282                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 283                 }
 284
 285                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 286
 287                 if (intval($timeout)) {
 288                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 289                 } else {
 290                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 291                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 292                 }
 293
 294                 if (!empty($headers)) {
 295                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 296                 }
 297
 298                 $check_cert = $this->config->get('system', 'verifyssl');
 299                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 300
 301                 if ($check_cert) {
 302                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 303                 }
 304
 305                 $proxy = $this->config->get('system', 'proxy');
 306
 307                 if (!empty($proxy)) {
 308                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 309                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 310                         $proxyuser = $this->config->get('system', 'proxyuser');
 311                         if (!empty($proxyuser)) {
 312                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 313                         }
 314                 }
 315
 316                 // don't let curl abort the entire application
 317                 // if it throws any errors.
 318
 319                 $s = @curl_exec($ch);
 320
 321                 $curl_info = curl_getinfo($ch);
 322
 323                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 324
 325                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 326                         $redirects++;
 327                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 328                         curl_close($ch);
 329                         $this->profiler->stopRecording();
 330                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 331                 }
 332
 333                 curl_close($ch);
 334
 335                 $this->profiler->stopRecording();
 336
 337                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 338                 if ($curlResponse->getReturnCode() == 417) {
 339                         $redirects++;
 340
 341                         if (empty($headers)) {
 342                                 $headers = ['Expect:'];
 343                         } else {
 344                                 if (!in_array('Expect:', $headers)) {
 345                                         array_push($headers, 'Expect:');
 346                                 }
 347                         }
 348                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 349                         return $this->post($url, $params, $headers, $redirects, $timeout);
 350                 }
 351
 352                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 353
 354                 return $curlResponse;
 355         }
 356
 357         /**
 358          * {@inheritDoc}
 359          */
 360         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 361         {
 362                 if (Network::isLocalLink($url)) {
 363                         $this->logger->info('Local link', ['url' => $url, 'callstack' => System::callstack(20)]);
 364                 }
 365
 366                 if (Network::isUrlBlocked($url)) {
 367                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 368                         return $url;
 369                 }
 370
 371                 if (Network::isRedirectBlocked($url)) {
 372                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 373                         return $url;
 374                 }
 375
 376                 $url = Network::stripTrackingQueryParams($url);
 377
 378                 if ($depth > 10) {
 379                         return $url;
 380                 }
 381
 382                 $url = trim($url, "'");
 383
 384                 $this->profiler->startRecording('network');
 385
 386                 $ch = curl_init();
 387                 curl_setopt($ch, CURLOPT_URL, $url);
 388                 curl_setopt($ch, CURLOPT_HEADER, 1);
 389                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 390                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 391                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 392                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 393                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 394
 395                 curl_exec($ch);
 396                 $curl_info = @curl_getinfo($ch);
 397                 $http_code = $curl_info['http_code'];
 398                 curl_close($ch);
 399
 400                 $this->profiler->stopRecording();
 401
 402                 if ($http_code == 0) {
 403                         return $url;
 404                 }
 405
 406                 if (in_array($http_code, ['301', '302'])) {
 407                         if (!empty($curl_info['redirect_url'])) {
 408                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 409                         } elseif (!empty($curl_info['location'])) {
 410                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 411                         }
 412                 }
 413
 414                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 415                 if (!$fetchbody) {
 416                         return $this->finalUrl($url, ++$depth, true);
 417                 }
 418
 419                 // if the file is too large then exit
 420                 if ($curl_info["download_content_length"] > 1000000) {
 421                         return $url;
 422                 }
 423
 424                 // if it isn't a HTML file then exit
 425                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 426                         return $url;
 427                 }
 428
 429                 $this->profiler->startRecording('network');
 430
 431                 $ch = curl_init();
 432                 curl_setopt($ch, CURLOPT_URL, $url);
 433                 curl_setopt($ch, CURLOPT_HEADER, 0);
 434                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 435                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 436                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 437                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 438                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 439
 440                 $body = curl_exec($ch);
 441                 curl_close($ch);
 442
 443                 $this->profiler->stopRecording();
 444
 445                 if (trim($body) == "") {
 446                         return $url;
 447                 }
 448
 449                 // Check for redirect in meta elements
 450                 $doc = new DOMDocument();
 451                 @$doc->loadHTML($body);
 452
 453                 $xpath = new DomXPath($doc);
 454
 455                 $list = $xpath->query("//meta[@content]");
 456                 foreach ($list as $node) {
 457                         $attr = [];
 458                         if ($node->attributes->length) {
 459                                 foreach ($node->attributes as $attribute) {
 460                                         $attr[$attribute->name] = $attribute->value;
 461                                 }
 462                         }
 463
 464                         if (@$attr["http-equiv"] == 'refresh') {
 465                                 $path = $attr["content"];
 466                                 $pathinfo = explode(";", $path);
 467                                 foreach ($pathinfo as $value) {
 468                                         if (substr(strtolower($value), 0, 4) == "url=") {
 469                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 470                                         }
 471                                 }
 472                         }
 473                 }
 474
 475                 return $url;
 476         }
 477
 478         /**
 479          * {@inheritDoc}
 480          *
 481          * @param int $redirects The recursion counter for internal use - default 0
 482          *
 483          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 484          */
 485         public function fetch(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '', &$redirects = 0)
 486         {
 487                 $ret = $this->fetchFull($url, $timeout, $accept_content, $cookiejar, $redirects);
 488
 489                 return $ret->getBody();
 490         }
 491
 492         /**
 493          * {@inheritDoc}
 494          *
 495          * @param int $redirects The recursion counter for internal use - default 0
 496          *
 497          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 498          */
 499         public function fetchFull(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '', &$redirects = 0)
 500         {
 501                 return $this->get(
 502                         $url,
 503                         [
 504                                 'timeout'        => $timeout,
 505                                 'accept_content' => $accept_content,
 506                                 'cookiejar'      => $cookiejar
 507                         ]
 508                 );
 509         }
 510
 511         /**
 512          * {@inheritDoc}
 513          */
 514         public function getUserAgent()
 515         {
 516                 return
 517                         FRIENDICA_PLATFORM . " '" .
 518                         FRIENDICA_CODENAME . "' " .
 519                         FRIENDICA_VERSION . '-' .
 520                         DB_UPDATE_VERSION . '; ' .
 521                         $this->baseUrl;
 522         }
 523 }