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
  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         /**
  56          * fetches an URL.
  57          *
  58          * @param string $url        URL to fetch
  59          * @param bool   $binary     default false
  60          *                           TRUE if asked to return binary results (file download)
  61          * @param array  $opts       (optional parameters) assoziative array with:
  62          *                           'accept_content' => supply Accept: header with 'accept_content' as the value
  63          *                           'timeout' => int Timeout in seconds, default system config value or 60 seconds
  64          *                           'http_auth' => username:password
  65          *                           'novalidate' => do not validate SSL certs, default is to validate using our CA list
  66          *                           'nobody' => only return the header
  67          *                           'cookiejar' => path to cookie jar file
  68          *                           'header' => header array
  69          * @param int    $redirects  The recursion counter for internal use - default 0
  70          *
  71          * @return CurlResult
  72          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  73          */
  74         public function get(string $url, bool $binary = false, array $opts = [], int &$redirects = 0)
  75         {
  76                 $stamp1 = microtime(true);
  77
  78                 if (strlen($url) > 1000) {
  79                         $this->logger->debug('URL is longer than 1000 characters.', ['url' => $url, 'callstack' => System::callstack(20)]);
  80                         return CurlResult::createErrorCurl(substr($url, 0, 200));
  81                 }
  82
  83                 $parts2     = [];
  84                 $parts      = parse_url($url);
  85                 $path_parts = explode('/', $parts['path'] ?? '');
  86                 foreach ($path_parts as $part) {
  87                         if (strlen($part) <> mb_strlen($part)) {
  88                                 $parts2[] = rawurlencode($part);
  89                         } else {
  90                                 $parts2[] = $part;
  91                         }
  92                 }
  93                 $parts['path'] = implode('/', $parts2);
  94                 $url           = Network::unparseURL($parts);
  95
  96                 if (Network::isUrlBlocked($url)) {
  97                         $this->logger->info('Domain is blocked.', ['url' => $url]);
  98                         return CurlResult::createErrorCurl($url);
  99                 }
 100
 101                 $ch = @curl_init($url);
 102
 103                 if (($redirects > 8) || (!$ch)) {
 104                         return CurlResult::createErrorCurl($url);
 105                 }
 106
 107                 @curl_setopt($ch, CURLOPT_HEADER, true);
 108
 109                 if (!empty($opts['cookiejar'])) {
 110                         curl_setopt($ch, CURLOPT_COOKIEJAR, $opts["cookiejar"]);
 111                         curl_setopt($ch, CURLOPT_COOKIEFILE, $opts["cookiejar"]);
 112                 }
 113
 114                 // These settings aren't needed. We're following the location already.
 115                 //      @curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
 116                 //      @curl_setopt($ch, CURLOPT_MAXREDIRS, 5);
 117
 118                 if (!empty($opts['accept_content'])) {
 119                         curl_setopt(
 120                                 $ch,
 121                                 CURLOPT_HTTPHEADER,
 122                                 ['Accept: ' . $opts['accept_content']]
 123                         );
 124                 }
 125
 126                 if (!empty($opts['header'])) {
 127                         curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['header']);
 128                 }
 129
 130                 @curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 131                 @curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 132
 133                 $range = intval($this->config->get('system', 'curl_range_bytes', 0));
 134
 135                 if ($range > 0) {
 136                         @curl_setopt($ch, CURLOPT_RANGE, '0-' . $range);
 137                 }
 138
 139                 // Without this setting it seems as if some webservers send compressed content
 140                 // This seems to confuse curl so that it shows this uncompressed.
 141                 /// @todo  We could possibly set this value to "gzip" or something similar
 142                 curl_setopt($ch, CURLOPT_ENCODING, '');
 143
 144                 if (!empty($opts['headers'])) {
 145                         @curl_setopt($ch, CURLOPT_HTTPHEADER, $opts['headers']);
 146                 }
 147
 148                 if (!empty($opts['nobody'])) {
 149                         @curl_setopt($ch, CURLOPT_NOBODY, $opts['nobody']);
 150                 }
 151
 152                 if (!empty($opts['timeout'])) {
 153                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 154                 } else {
 155                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 156                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 157                 }
 158
 159                 // by default we will allow self-signed certs
 160                 // but you can override this
 161
 162                 $check_cert = $this->config->get('system', 'verifyssl');
 163                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 164
 165                 if ($check_cert) {
 166                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 167                 }
 168
 169                 $proxy = $this->config->get('system', 'proxy');
 170
 171                 if (!empty($proxy)) {
 172                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 173                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 174                         $proxyuser = $this->config->get('system', 'proxyuser');
 175
 176                         if (!empty($proxyuser)) {
 177                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 178                         }
 179                 }
 180
 181                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 182                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 183                 }
 184
 185                 if ($binary) {
 186                         @curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
 187                 }
 188
 189                 // don't let curl abort the entire application
 190                 // if it throws any errors.
 191
 192                 $s         = @curl_exec($ch);
 193                 $curl_info = @curl_getinfo($ch);
 194
 195                 // Special treatment for HTTP Code 416
 196                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 197                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 198                         @curl_setopt($ch, CURLOPT_RANGE, '');
 199                         $s         = @curl_exec($ch);
 200                         $curl_info = @curl_getinfo($ch);
 201                 }
 202
 203                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 204
 205                 if ($curlResponse->isRedirectUrl()) {
 206                         $redirects++;
 207                         $this->logger->notice('Curl redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 208                         @curl_close($ch);
 209                         return $this->get($curlResponse->getRedirectUrl(), $binary, $opts, $redirects);
 210                 }
 211
 212                 @curl_close($ch);
 213
 214                 $this->profiler->saveTimestamp($stamp1, 'network', System::callstack());
 215
 216                 return $curlResponse;
 217         }
 218
 219         /**
 220          * Send POST request to $url
 221          *
 222          * @param string $url       URL to post
 223          * @param mixed  $params    array of POST variables
 224          * @param array  $headers   HTTP headers
 225          * @param int    $redirects Recursion counter for internal use - default = 0
 226          * @param int    $timeout   The timeout in seconds, default system config value or 60 seconds
 227          *
 228          * @return CurlResult The content
 229          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 230          */
 231         public function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
 232         {
 233                 $stamp1 = microtime(true);
 234
 235                 if (Network::isUrlBlocked($url)) {
 236                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 237                         return CurlResult::createErrorCurl($url);
 238                 }
 239
 240                 $ch = curl_init($url);
 241
 242                 if (($redirects > 8) || (!$ch)) {
 243                         return CurlResult::createErrorCurl($url);
 244                 }
 245
 246                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 247
 248                 curl_setopt($ch, CURLOPT_HEADER, true);
 249                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 250                 curl_setopt($ch, CURLOPT_POST, 1);
 251                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 252                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 253
 254                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 255                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 256                 }
 257
 258                 if (intval($timeout)) {
 259                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 260                 } else {
 261                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 262                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 263                 }
 264
 265                 if (!empty($headers)) {
 266                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 267                 }
 268
 269                 $check_cert = $this->config->get('system', 'verifyssl');
 270                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 271
 272                 if ($check_cert) {
 273                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 274                 }
 275
 276                 $proxy = $this->config->get('system', 'proxy');
 277
 278                 if (!empty($proxy)) {
 279                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 280                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 281                         $proxyuser = $this->config->get('system', 'proxyuser');
 282                         if (!empty($proxyuser)) {
 283                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 284                         }
 285                 }
 286
 287                 // don't let curl abort the entire application
 288                 // if it throws any errors.
 289
 290                 $s = @curl_exec($ch);
 291
 292                 $curl_info = curl_getinfo($ch);
 293
 294                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 295
 296                 if ($curlResponse->isRedirectUrl()) {
 297                         $redirects++;
 298                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 299                         curl_close($ch);
 300                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 301                 }
 302
 303                 curl_close($ch);
 304
 305                 $this->profiler->saveTimestamp($stamp1, 'network', System::callstack());
 306
 307                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 308                 if ($curlResponse->getReturnCode() == 417) {
 309                         $redirects++;
 310
 311                         if (empty($headers)) {
 312                                 $headers = ['Expect:'];
 313                         } else {
 314                                 if (!in_array('Expect:', $headers)) {
 315                                         array_push($headers, 'Expect:');
 316                                 }
 317                         }
 318                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 319                         return $this->post($url, $params, $headers, $redirects, $timeout);
 320                 }
 321
 322                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 323
 324                 return $curlResponse;
 325         }
 326
 327         /**
 328          * Returns the original URL of the provided URL
 329          *
 330          * This function strips tracking query params and follows redirections, either
 331          * through HTTP code or meta refresh tags. Stops after 10 redirections.
 332          *
 333          * @todo  Remove the $fetchbody parameter that generates an extraneous HEAD request
 334          *
 335          * @see   ParseUrl::getSiteinfo
 336          *
 337          * @param string $url       A user-submitted URL
 338          * @param int    $depth     The current redirection recursion level (internal)
 339          * @param bool   $fetchbody Wether to fetch the body or not after the HEAD requests
 340          * @return string A canonical URL
 341          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 342          */
 343         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 344         {
 345                 $url = Network::stripTrackingQueryParams($url);
 346
 347                 if ($depth > 10) {
 348                         return $url;
 349                 }
 350
 351                 $url = trim($url, "'");
 352
 353                 $stamp1 = microtime(true);
 354
 355                 $ch = curl_init();
 356                 curl_setopt($ch, CURLOPT_URL, $url);
 357                 curl_setopt($ch, CURLOPT_HEADER, 1);
 358                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 359                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 360                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 361                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 362
 363                 curl_exec($ch);
 364                 $curl_info = @curl_getinfo($ch);
 365                 $http_code = $curl_info['http_code'];
 366                 curl_close($ch);
 367
 368                 $this->profiler->saveTimestamp($stamp1, "network", System::callstack());
 369
 370                 if ($http_code == 0) {
 371                         return $url;
 372                 }
 373
 374                 if (in_array($http_code, ['301', '302'])) {
 375                         if (!empty($curl_info['redirect_url'])) {
 376                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 377                         } elseif (!empty($curl_info['location'])) {
 378                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 379                         }
 380                 }
 381
 382                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 383                 if (!$fetchbody) {
 384                         return $this->finalUrl($url, ++$depth, true);
 385                 }
 386
 387                 // if the file is too large then exit
 388                 if ($curl_info["download_content_length"] > 1000000) {
 389                         return $url;
 390                 }
 391
 392                 // if it isn't a HTML file then exit
 393                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 394                         return $url;
 395                 }
 396
 397                 $stamp1 = microtime(true);
 398
 399                 $ch = curl_init();
 400                 curl_setopt($ch, CURLOPT_URL, $url);
 401                 curl_setopt($ch, CURLOPT_HEADER, 0);
 402                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 403                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 404                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 405                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 406
 407                 $body = curl_exec($ch);
 408                 curl_close($ch);
 409
 410                 $this->profiler->saveTimestamp($stamp1, "network", System::callstack());
 411
 412                 if (trim($body) == "") {
 413                         return $url;
 414                 }
 415
 416                 // Check for redirect in meta elements
 417                 $doc = new DOMDocument();
 418                 @$doc->loadHTML($body);
 419
 420                 $xpath = new DomXPath($doc);
 421
 422                 $list = $xpath->query("//meta[@content]");
 423                 foreach ($list as $node) {
 424                         $attr = [];
 425                         if ($node->attributes->length) {
 426                                 foreach ($node->attributes as $attribute) {
 427                                         $attr[$attribute->name] = $attribute->value;
 428                                 }
 429                         }
 430
 431                         if (@$attr["http-equiv"] == 'refresh') {
 432                                 $path = $attr["content"];
 433                                 $pathinfo = explode(";", $path);
 434                                 foreach ($pathinfo as $value) {
 435                                         if (substr(strtolower($value), 0, 4) == "url=") {
 436                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 437                                         }
 438                                 }
 439                         }
 440                 }
 441
 442                 return $url;
 443         }
 444
 445         /**
 446          * Curl wrapper
 447          *
 448          * If binary flag is true, return binary results.
 449          * Set the cookiejar argument to a string (e.g. "/tmp/friendica-cookies.txt")
 450          * to preserve cookies from one request to the next.
 451          *
 452          * @param string $url             URL to fetch
 453          * @param bool   $binary          default false
 454          *                                TRUE if asked to return binary results (file download)
 455          * @param int    $timeout         Timeout in seconds, default system config value or 60 seconds
 456          * @param string $accept_content  supply Accept: header with 'accept_content' as the value
 457          * @param string $cookiejar       Path to cookie jar file
 458          * @param int    $redirects       The recursion counter for internal use - default 0
 459          *
 460          * @return string The fetched content
 461          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 462          */
 463         public function fetch(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 464         {
 465                 $ret = $this->fetchFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
 466
 467                 return $ret->getBody();
 468         }
 469
 470         /**
 471          * Curl wrapper with array of return values.
 472          *
 473          * Inner workings and parameters are the same as @ref fetchUrl but returns an array with
 474          * all the information collected during the fetch.
 475          *
 476          * @param string $url             URL to fetch
 477          * @param bool   $binary          default false
 478          *                                TRUE if asked to return binary results (file download)
 479          * @param int    $timeout         Timeout in seconds, default system config value or 60 seconds
 480          * @param string $accept_content  supply Accept: header with 'accept_content' as the value
 481          * @param string $cookiejar       Path to cookie jar file
 482          * @param int    $redirects       The recursion counter for internal use - default 0
 483          *
 484          * @return CurlResult With all relevant information, 'body' contains the actual fetched content.
 485          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 486          */
 487         public function fetchFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 488         {
 489                 return $this->get(
 490                         $url,
 491                         $binary,
 492                         [
 493                                 'timeout'        => $timeout,
 494                                 'accept_content' => $accept_content,
 495                                 'cookiejar'      => $cookiejar
 496                         ],
 497                         $redirects
 498                 );
 499         }
 500
 501         /**
 502          * Returns the current UserAgent as a String
 503          *
 504          * @return string the UserAgent as a String
 505          */
 506         public function getUserAgent()
 507         {
 508                 return
 509                         FRIENDICA_PLATFORM . " '" .
 510                         FRIENDICA_CODENAME . "' " .
 511                         FRIENDICA_VERSION . '-' .
 512                         DB_UPDATE_VERSION . '; ' .
 513                         $this->baseUrl;
 514         }
 515 }