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