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\Logger;
  29 use Friendica\Core\System;
  30 use Friendica\DI;
  31 use Friendica\Util\Network;
  32 use Friendica\Util\Profiler;
  33 use Psr\Log\LoggerInterface;
  34
  35 /**
  36  * Performs HTTP requests to a given URL
  37  */
  38 class HTTPRequest
  39 {
  40         /** @var LoggerInterface */
  41         private $logger;
  42         /** @var Profiler */
  43         private $profiler;
  44         /** @var IConfig */
  45         private $config;
  46         /** @var string */
  47         private $baseUrl;
  48
  49         public function __construct(LoggerInterface $logger, Profiler $profiler, IConfig $config, App\BaseURL $baseUrl)
  50         {
  51                 $this->logger   = $logger;
  52                 $this->profiler = $profiler;
  53                 $this->config   = $config;
  54                 $this->baseUrl  = $baseUrl->get();
  55         }
  56
  57         /**
  58          * fetches an URL.
  59          *
  60          * @param string $url        URL to fetch
  61          * @param bool   $binary     default false
  62          *                           TRUE if asked to return binary results (file download)
  63          * @param array  $opts       (optional parameters) assoziative array with:
  64          *                           'accept_content' => supply Accept: header with 'accept_content' as the value
  65          *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
  66          *                           'http_auth' => username:password
  67          *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
  68          *                           'nobody' => only return the header
  69          *                           'cookiejar' => path to cookie jar file
  70          *                           'header' => header array
  71          * @param int    $redirects  The recursion counter for internal use - default 0
  72          *
  73          * @return CurlResult
  74          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  75          */
  76         public function curl(string $url, bool $binary = false, array $opts = [], int &$redirects = 0)
  77         {
  78                 $stamp1 = microtime(true);
  79
  80                 if (strlen($url) > 1000) {
  81                         $this->logger->debug('URL is longer than 1000 characters.', ['url' => $url, 'callstack' => System::callstack(20)]);
  82                         return CurlResult::createErrorCurl(substr($url, 0, 200));
  83                 }
  84
  85                 $parts2     = [];
  86                 $parts      = parse_url($url);
  87                 $path_parts = explode('/', $parts['path'] ?? '');
  88                 foreach ($path_parts as $part) {
  89                         if (strlen($part) <> mb_strlen($part)) {
  90                                 $parts2[] = rawurlencode($part);
  91                         } else {
  92                                 $parts2[] = $part;
  93                         }
  94                 }
  95                 $parts['path'] = implode('/', $parts2);
  96                 $url           = Network::unparseURL($parts);
  97
  98                 if (Network::isUrlBlocked($url)) {
  99                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 100                         return CurlResult::createErrorCurl($url);
 101                 }
 102
 103                 $ch = @curl_init($url);
 104
 105                 if (($redirects > 8) || (!$ch)) {
 106                         return CurlResult::createErrorCurl($url);
 107                 }
 108
 109                 @curl_setopt($ch, CURLOPT_HEADER, true);
 110
 111                 if (!empty($opts['cookiejar'])) {
 112                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
 113                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
 114                 }
 115
 116                 // These settings aren't needed. We're following the location already.
 117                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 118                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 119
 120                 if (!empty($opts['accept_content'])) {
 121                         curl_setopt(
 122                                 $ch,
 123                                 CURLOPT_HTTPHEADER,
 124                                 ['Accept: ' . $opts['accept_content']]
 125                         );
 126                 }
 127
 128                 if (!empty($opts['header'])) {
 129                         curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['header']);
 130                 }
 131
 132                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 133                 @curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent);
 134
 135                 $range = intval($this->config->get('system', 'curl_range_bytes', 0));
 136
 137                 if ($range > 0) {
 138                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 139                 }
 140
 141                 // Without this setting it seems as if some webservers send compressed content
 142                 // This seems to confuse curl so that it shows this uncompressed.
 143                 /// @todo  We could possibly set this value to "gzip" or something similar
 144                 curl_setopt($ch, CURLOPT_ENCODING, '');
 145
 146                 if (!empty($opts['headers'])) {
 147                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 148                 }
 149
 150                 if (!empty($opts['nobody'])) {
 151                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 152                 }
 153
 154                 if (!empty($opts['timeout'])) {
 155                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 156                 } else {
 157                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 158                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 159                 }
 160
 161                 // by default we will allow self-signed certs
 162                 // but you can override this
 163
 164                 $check_cert = $this->config->get('system', 'verifyssl');
 165                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 166
 167                 if ($check_cert) {
 168                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 169                 }
 170
 171                 $proxy = $this->config->get('system', 'proxy');
 172
 173                 if (!empty($proxy)) {
 174                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 175                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 176                         $proxyuser = $this->config->get('system', 'proxyuser');
 177
 178                         if (!empty($proxyuser)) {
 179                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 180                         }
 181                 }
 182
 183                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 184                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 185                 }
 186
 187                 if ($binary) {
 188                         @curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
 189                 }
 190
 191                 // don't let curl abort the entire application
 192                 // if it throws any errors.
 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($ch));
 206
 207                 if ($curlResponse->isRedirectUrl()) {
 208                         $redirects++;
 209                         $this->logger->notice('Curl redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 210                         @curl_close($ch);
 211                         return self::curl($curlResponse->getRedirectUrl(), $binary, $opts, $redirects);
 212                 }
 213
 214                 @curl_close($ch);
 215
 216                 $this->profiler->saveTimestamp($stamp1, 'network', System::callstack());
 217
 218                 return $curlResponse;
 219         }
 220
 221         /**
 222          * Send POST request to $url
 223          *
 224          * @param string $url       URL to post
 225          * @param mixed  $params    array of POST variables
 226          * @param array  $headers   HTTP headers
 227          * @param int    $redirects Recursion counter for internal use - default = 0
 228          * @param int    $timeout   The timeout in seconds, default system config value or 60 seconds
 229          *
 230          * @return CurlResult The content
 231          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 232          */
 233         public function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
 234         {
 235                 $stamp1 = microtime(true);
 236
 237                 if (Network::isUrlBlocked($url)) {
 238                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 239                         return CurlResult::createErrorCurl($url);
 240                 }
 241
 242                 $ch = curl_init($url);
 243
 244                 if (($redirects > 8) || (!$ch)) {
 245                         return CurlResult::createErrorCurl($url);
 246                 }
 247
 248                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 249
 250                 curl_setopt($ch, CURLOPT_HEADER, true);
 251                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 252                 curl_setopt($ch, CURLOPT_POST, 1);
 253                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 254                 curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent);
 255
 256                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 257                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 258                 }
 259
 260                 if (intval($timeout)) {
 261                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 262                 } else {
 263                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 264                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 265                 }
 266
 267                 if (!empty($headers)) {
 268                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 269                 }
 270
 271                 $check_cert = $this->config->get('system', 'verifyssl');
 272                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 273
 274                 if ($check_cert) {
 275                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 276                 }
 277
 278                 $proxy = $this->config->get('system', 'proxy');
 279
 280                 if (!empty($proxy)) {
 281                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 282                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 283                         $proxyuser = $this->config->get('system', 'proxyuser');
 284                         if (!empty($proxyuser)) {
 285                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 286                         }
 287                 }
 288
 289                 // don't let curl abort the entire application
 290                 // if it throws any errors.
 291
 292                 $s = @curl_exec($ch);
 293
 294                 $curl_info = curl_getinfo($ch);
 295
 296                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 297
 298                 if ($curlResponse->isRedirectUrl()) {
 299                         $redirects++;
 300                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 301                         curl_close($ch);
 302                         return self::post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 303                 }
 304
 305                 curl_close($ch);
 306
 307                 $this->profiler->saveTimestamp($stamp1, 'network', System::callstack());
 308
 309                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 310                 if ($curlResponse->getReturnCode() == 417) {
 311                         $redirects++;
 312
 313                         if (empty($headers)) {
 314                                 $headers = ['Expect:'];
 315                         } else {
 316                                 if (!in_array('Expect:', $headers)) {
 317                                         array_push($headers, 'Expect:');
 318                                 }
 319                         }
 320                         Logger::info('Server responds with 417, applying workaround', ['url' => $url]);
 321                         return self::post($url, $params, $headers, $redirects, $timeout);
 322                 }
 323
 324                 Logger::log('post_url: end ' . $url, Logger::DATA);
 325
 326                 return $curlResponse;
 327         }
 328
 329         /**
 330          * Returns the original URL of the provided URL
 331          *
 332          * This function strips tracking query params and follows redirections, either
 333          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 334          *
 335          * @todo  Remove the $fetchbody parameter that generates an extraneous HEAD request
 336          *
 337          * @see   ParseUrl::getSiteinfo
 338          *
 339          * @param string $url       A user-submitted URL
 340          * @param int    $depth     The current redirection recursion level (internal)
 341          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 342          * @return string A canonical URL
 343          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 344          */
 345         public static function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 346         {
 347                 $url = Network::stripTrackingQueryParams($url);
 348
 349                 if ($depth > 10) {
 350                         return $url;
 351                 }
 352
 353                 $url = trim($url, "'");
 354
 355                 $stamp1 = microtime(true);
 356
 357                 $ch = curl_init();
 358                 curl_setopt($ch, CURLOPT_URL, $url);
 359                 curl_setopt($ch, CURLOPT_HEADER, 1);
 360                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 361                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 362                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 363                 curl_setopt($ch, CURLOPT_USERAGENT, DI::httpRequest()->getUserAgent());
 364
 365                 curl_exec($ch);
 366                 $curl_info = @curl_getinfo($ch);
 367                 $http_code = $curl_info['http_code'];
 368                 curl_close($ch);
 369
 370                 DI::profiler()->saveTimestamp($stamp1, "network", System::callstack());
 371
 372                 if ($http_code == 0) {
 373                         return $url;
 374                 }
 375
 376                 if (in_array($http_code, ['301', '302'])) {
 377                         if (!empty($curl_info['redirect_url'])) {
 378                                 return self::finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 379                         } elseif (!empty($curl_info['location'])) {
 380                                 return self::finalUrl($curl_info['location'], ++$depth, $fetchbody);
 381                         }
 382                 }
 383
 384                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 385                 if (!$fetchbody) {
 386                         return self::finalUrl($url, ++$depth, true);
 387                 }
 388
 389                 // if the file is too large then exit
 390                 if ($curl_info["download_content_length"] > 1000000) {
 391                         return $url;
 392                 }
 393
 394                 // if it isn't a HTML file then exit
 395                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 396                         return $url;
 397                 }
 398
 399                 $stamp1 = microtime(true);
 400
 401                 $ch = curl_init();
 402                 curl_setopt($ch, CURLOPT_URL, $url);
 403                 curl_setopt($ch, CURLOPT_HEADER, 0);
 404                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 405                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 406                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 407                 curl_setopt($ch, CURLOPT_USERAGENT, DI::httpRequest()->getUserAgent());
 408
 409                 $body = curl_exec($ch);
 410                 curl_close($ch);
 411
 412                 DI::profiler()->saveTimestamp($stamp1, "network", System::callstack());
 413
 414                 if (trim($body) == "") {
 415                         return $url;
 416                 }
 417
 418                 // Check for redirect in meta elements
 419                 $doc = new DOMDocument();
 420                 @$doc->loadHTML($body);
 421
 422                 $xpath = new DomXPath($doc);
 423
 424                 $list = $xpath->query("//meta[@content]");
 425                 foreach ($list as $node) {
 426                         $attr = [];
 427                         if ($node->attributes->length) {
 428                                 foreach ($node->attributes as $attribute) {
 429                                         $attr[$attribute->name] = $attribute->value;
 430                                 }
 431                         }
 432
 433                         if (@$attr["http-equiv"] == 'refresh') {
 434                                 $path = $attr["content"];
 435                                 $pathinfo = explode(";", $path);
 436                                 foreach ($pathinfo as $value) {
 437                                         if (substr(strtolower($value), 0, 4) == "url=") {
 438                                                 return self::finalUrl(substr($value, 4), ++$depth);
 439                                         }
 440                                 }
 441                         }
 442                 }
 443
 444                 return $url;
 445         }
 446
 447         /**
 448          * Curl wrapper
 449          *
 450          * If binary flag is true, return binary results.
 451          * Set the cookiejar argument to a string (e.g. "/tmp/friendica-cookies.txt")
 452          * to preserve cookies from one request to the next.
 453          *
 454          * @param string $url             URL to fetch
 455          * @param bool   $binary          default false
 456          *                                TRUE if asked to return binary results (file download)
 457          * @param int    $timeout         Timeout in seconds, default system config value or 60 seconds
 458          * @param string $accept_content  supply Accept: header with 'accept_content' as the value
 459          * @param string $cookiejar       Path to cookie jar file
 460          * @param int    $redirects       The recursion counter for internal use - default 0
 461          *
 462          * @return string The fetched content
 463          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 464          */
 465         public function fetchUrl(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 466         {
 467                 $ret = $this->fetchUrlFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
 468
 469                 return $ret->getBody();
 470         }
 471
 472         /**
 473          * Curl wrapper with array of return values.
 474          *
 475          * Inner workings and parameters are the same as @ref fetchUrl but returns an array with
 476          * all the information collected during the fetch.
 477          *
 478          * @param string $url             URL to fetch
 479          * @param bool   $binary          default false
 480          *                                TRUE if asked to return binary results (file download)
 481          * @param int    $timeout         Timeout in seconds, default system config value or 60 seconds
 482          * @param string $accept_content  supply Accept: header with 'accept_content' as the value
 483          * @param string $cookiejar       Path to cookie jar file
 484          * @param int    $redirects       The recursion counter for internal use - default 0
 485          *
 486          * @return CurlResult With all relevant information, 'body' contains the actual fetched content.
 487          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 488          */
 489         public function fetchUrlFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 490         {
 491                 return $this->curl(
 492                         $url,
 493                         $binary,
 494                         [
 495                                 'timeout'        => $timeout,
 496                                 'accept_content' => $accept_content,
 497                                 'cookiejar'      => $cookiejar
 498                         ],
 499                         $redirects
 500                 );
 501         }
 502
 503         /**
 504          * Returns the current UserAgent as a String
 505          *
 506          * @return string the UserAgent as a String
 507          */
 508         public function getUserAgent()
 509         {
 510                 return
 511                         FRIENDICA_PLATFORM . " '" .
 512                         FRIENDICA_CODENAME . "' " .
 513                         FRIENDICA_VERSION . '-' .
 514                         DB_UPDATE_VERSION . '; ' .
 515                         $this->baseUrl;
 516         }
 517 }