src/Network/HTTPRequest.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   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 $opts = [], int &$redirects = 0)
  74         {
  75                 $stamp1 = microtime(true);
  76
  77                 if (strlen($url) > 1000) {
  78                         $this->logger->debug('URL is longer than 1000 characters.', ['url' => $url, 'callstack' => System::callstack(20)]);
  79                         return CurlResult::createErrorCurl(substr($url, 0, 200));
  80                 }
  81
  82                 $parts2     = [];
  83                 $parts      = parse_url($url);
  84                 $path_parts = explode('/', $parts['path'] ?? '');
  85                 foreach ($path_parts as $part) {
  86                         if (strlen($part) <> mb_strlen($part)) {
  87                                 $parts2[] = rawurlencode($part);
  88                         } else {
  89                                 $parts2[] = $part;
  90                         }
  91                 }
  92                 $parts['path'] = implode('/', $parts2);
  93                 $url           = Network::unparseURL($parts);
  94
  95                 if (Network::isUrlBlocked($url)) {
  96                         $this->logger->info('Domain is blocked.', ['url' => $url]);
  97                         return CurlResult::createErrorCurl($url);
  98                 }
  99
 100                 $ch = @curl_init($url);
 101
 102                 if (($redirects > 8) || (!$ch)) {
 103                         return CurlResult::createErrorCurl($url);
 104                 }
 105
 106                 @curl_setopt($ch, CURLOPT_HEADER, true);
 107
 108                 if (!empty($opts['cookiejar'])) {
 109                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
 110                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
 111                 }
 112
 113                 // These settings aren't needed. We're following the location already.
 114                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 115                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 116
 117                 if (!empty($opts['accept_content'])) {
 118                         curl_setopt(
 119                                 $ch,
 120                                 CURLOPT_HTTPHEADER,
 121                                 ['Accept: ' . $opts['accept_content']]
 122                         );
 123                 }
 124
 125                 if (!empty($opts['header'])) {
 126                         curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['header']);
 127                 }
 128
 129                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 130                 @curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 131
 132                 $range = intval($this->config->get('system', 'curl_range_bytes', 0));
 133
 134                 if ($range > 0) {
 135                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 136                 }
 137
 138                 // Without this setting it seems as if some webservers send compressed content
 139                 // This seems to confuse curl so that it shows this uncompressed.
 140                 /// @todo  We could possibly set this value to "gzip" or something similar
 141                 curl_setopt($ch, CURLOPT_ENCODING, '');
 142
 143                 if (!empty($opts['headers'])) {
 144                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 145                 }
 146
 147                 if (!empty($opts['nobody'])) {
 148                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 149                 }
 150
 151                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 152
 153                 if (!empty($opts['timeout'])) {
 154                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 155                 } else {
 156                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 157                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 158                 }
 159
 160                 // by default we will allow self-signed certs
 161                 // but you can override this
 162
 163                 $check_cert = $this->config->get('system', 'verifyssl');
 164                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 165
 166                 if ($check_cert) {
 167                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 168                 }
 169
 170                 $proxy = $this->config->get('system', 'proxy');
 171
 172                 if (!empty($proxy)) {
 173                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 174                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 175                         $proxyuser = $this->config->get('system', 'proxyuser');
 176
 177                         if (!empty($proxyuser)) {
 178                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 179                         }
 180                 }
 181
 182                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 183                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 184                 }
 185
 186                 $logger = $this->logger;
 187
 188                 $s         = @curl_exec($ch);
 189                 $curl_info = @curl_getinfo($ch);
 190
 191                 // Special treatment for HTTP Code 416
 192                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 193                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 194                         @curl_setopt($ch, CURLOPT_RANGE, '');
 195                         $s         = @curl_exec($ch);
 196                         $curl_info = @curl_getinfo($ch);
 197                 }
 198
 199                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 200
 201                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 202                         $redirects++;
 203                         $this->logger->notice('Curl redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 204                         @curl_close($ch);
 205                         return $this->get($curlResponse->getRedirectUrl(), $opts, $redirects);
 206                 }
 207
 208                 @curl_close($ch);
 209
 210                 $this->profiler->saveTimestamp($stamp1, 'network');
 211
 212                 return $curlResponse;
 213         }
 214
 215         /**
 216          * {@inheritDoc}
 217          *
 218          * @param int $redirects The recursion counter for internal use - default 0
 219          *
 220          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 221          */
 222         public function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
 223         {
 224                 $stamp1 = microtime(true);
 225
 226                 if (Network::isUrlBlocked($url)) {
 227                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 228                         return CurlResult::createErrorCurl($url);
 229                 }
 230
 231                 $ch = curl_init($url);
 232
 233                 if (($redirects > 8) || (!$ch)) {
 234                         return CurlResult::createErrorCurl($url);
 235                 }
 236
 237                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 238
 239                 curl_setopt($ch, CURLOPT_HEADER, true);
 240                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 241                 curl_setopt($ch, CURLOPT_POST, 1);
 242                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 243                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 244
 245                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 246                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 247                 }
 248
 249                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 250
 251                 if (intval($timeout)) {
 252                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 253                 } else {
 254                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 255                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 256                 }
 257
 258                 if (!empty($headers)) {
 259                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 260                 }
 261
 262                 $check_cert = $this->config->get('system', 'verifyssl');
 263                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 264
 265                 if ($check_cert) {
 266                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 267                 }
 268
 269                 $proxy = $this->config->get('system', 'proxy');
 270
 271                 if (!empty($proxy)) {
 272                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 273                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 274                         $proxyuser = $this->config->get('system', 'proxyuser');
 275                         if (!empty($proxyuser)) {
 276                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 277                         }
 278                 }
 279
 280                 // don't let curl abort the entire application
 281                 // if it throws any errors.
 282
 283                 $s = @curl_exec($ch);
 284
 285                 $curl_info = curl_getinfo($ch);
 286
 287                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 288
 289                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 290                         $redirects++;
 291                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 292                         curl_close($ch);
 293                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 294                 }
 295
 296                 curl_close($ch);
 297
 298                 $this->profiler->saveTimestamp($stamp1, 'network');
 299
 300                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 301                 if ($curlResponse->getReturnCode() == 417) {
 302                         $redirects++;
 303
 304                         if (empty($headers)) {
 305                                 $headers = ['Expect:'];
 306                         } else {
 307                                 if (!in_array('Expect:', $headers)) {
 308                                         array_push($headers, 'Expect:');
 309                                 }
 310                         }
 311                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 312                         return $this->post($url, $params, $headers, $redirects, $timeout);
 313                 }
 314
 315                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 316
 317                 return $curlResponse;
 318         }
 319
 320         /**
 321          * {@inheritDoc}
 322          */
 323         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 324         {
 325                 if (Network::isUrlBlocked($url)) {
 326                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 327                         return $url;
 328                 }
 329
 330                 if (Network::isRedirectBlocked($url)) {
 331                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 332                         return $url;
 333                 }
 334
 335                 $url = Network::stripTrackingQueryParams($url);
 336
 337                 if ($depth > 10) {
 338                         return $url;
 339                 }
 340
 341                 $url = trim($url, "'");
 342
 343                 $stamp1 = microtime(true);
 344
 345                 $ch = curl_init();
 346                 curl_setopt($ch, CURLOPT_URL, $url);
 347                 curl_setopt($ch, CURLOPT_HEADER, 1);
 348                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 349                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 350                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 351                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 352                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 353
 354                 curl_exec($ch);
 355                 $curl_info = @curl_getinfo($ch);
 356                 $http_code = $curl_info['http_code'];
 357                 curl_close($ch);
 358
 359                 $this->profiler->saveTimestamp($stamp1, "network");
 360
 361                 if ($http_code == 0) {
 362                         return $url;
 363                 }
 364
 365                 if (in_array($http_code, ['301', '302'])) {
 366                         if (!empty($curl_info['redirect_url'])) {
 367                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 368                         } elseif (!empty($curl_info['location'])) {
 369                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 370                         }
 371                 }
 372
 373                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 374                 if (!$fetchbody) {
 375                         return $this->finalUrl($url, ++$depth, true);
 376                 }
 377
 378                 // if the file is too large then exit
 379                 if ($curl_info["download_content_length"] > 1000000) {
 380                         return $url;
 381                 }
 382
 383                 // if it isn't a HTML file then exit
 384                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 385                         return $url;
 386                 }
 387
 388                 $stamp1 = microtime(true);
 389
 390                 $ch = curl_init();
 391                 curl_setopt($ch, CURLOPT_URL, $url);
 392                 curl_setopt($ch, CURLOPT_HEADER, 0);
 393                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 394                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 395                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 396                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 397                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 398
 399                 $body = curl_exec($ch);
 400                 curl_close($ch);
 401
 402                 $this->profiler->saveTimestamp($stamp1, "network");
 403
 404                 if (trim($body) == "") {
 405                         return $url;
 406                 }
 407
 408                 // Check for redirect in meta elements
 409                 $doc = new DOMDocument();
 410                 @$doc->loadHTML($body);
 411
 412                 $xpath = new DomXPath($doc);
 413
 414                 $list = $xpath->query("//meta[@content]");
 415                 foreach ($list as $node) {
 416                         $attr = [];
 417                         if ($node->attributes->length) {
 418                                 foreach ($node->attributes as $attribute) {
 419                                         $attr[$attribute->name] = $attribute->value;
 420                                 }
 421                         }
 422
 423                         if (@$attr["http-equiv"] == 'refresh') {
 424                                 $path = $attr["content"];
 425                                 $pathinfo = explode(";", $path);
 426                                 foreach ($pathinfo as $value) {
 427                                         if (substr(strtolower($value), 0, 4) == "url=") {
 428                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 429                                         }
 430                                 }
 431                         }
 432                 }
 433
 434                 return $url;
 435         }
 436
 437         /**
 438          * {@inheritDoc}
 439          *
 440          * @param int $redirects The recursion counter for internal use - default 0
 441          *
 442          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 443          */
 444         public function fetch(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 445         {
 446                 $ret = $this->fetchFull($url, $timeout, $accept_content, $cookiejar, $redirects);
 447
 448                 return $ret->getBody();
 449         }
 450
 451         /**
 452          * {@inheritDoc}
 453          *
 454          * @param int $redirects The recursion counter for internal use - default 0
 455          *
 456          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 457          */
 458         public function fetchFull(string $url, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 459         {
 460                 return $this->get(
 461                         $url,
 462                         [
 463                                 'timeout'        => $timeout,
 464                                 'accept_content' => $accept_content,
 465                                 'cookiejar'      => $cookiejar
 466                         ],
 467                         $redirects
 468                 );
 469         }
 470
 471         /**
 472          * {@inheritDoc}
 473          */
 474         public function getUserAgent()
 475         {
 476                 return
 477                         FRIENDICA_PLATFORM . " '" .
 478                         FRIENDICA_CODENAME . "' " .
 479                         FRIENDICA_VERSION . '-' .
 480                         DB_UPDATE_VERSION . '; ' .
 481                         $this->baseUrl;
 482         }
 483 }