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, $opts['nobody']);
 138                 }
 139
 140                 if (!empty($opts['timeout'])) {
 141                         @curl_setopt($ch, CURLOPT_TIMEOUT, $opts['timeout']);
 142                 } else {
 143                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 144                         @curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 145                 }
 146
 147                 // by default we will allow self-signed certs
 148                 // but you can override this
 149
 150                 $check_cert = $this->config->get('system', 'verifyssl');
 151                 @curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 152
 153                 if ($check_cert) {
 154                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 155                 }
 156
 157                 $proxy = $this->config->get('system', 'proxy');
 158
 159                 if (!empty($proxy)) {
 160                         @curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 161                         @curl_setopt($ch, CURLOPT_PROXY, $proxy);
 162                         $proxyuser = $this->config->get('system', 'proxyuser');
 163
 164                         if (!empty($proxyuser)) {
 165                                 @curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 166                         }
 167                 }
 168
 169                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 170                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 171                 }
 172
 173                 if ($binary) {
 174                         @curl_setopt($ch, CURLOPT_BINARYTRANSFER, 1);
 175                 }
 176
 177                 // don't let curl abort the entire application
 178                 // if it throws any errors.
 179
 180                 $s         = @curl_exec($ch);
 181                 $curl_info = @curl_getinfo($ch);
 182
 183                 // Special treatment for HTTP Code 416
 184                 // See https://developer.mozilla.org/en-US/docs/Web/HTTP/Status/416
 185                 if (($curl_info['http_code'] == 416) && ($range > 0)) {
 186                         @curl_setopt($ch, CURLOPT_RANGE, '');
 187                         $s         = @curl_exec($ch);
 188                         $curl_info = @curl_getinfo($ch);
 189                 }
 190
 191                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 192
 193                 if ($curlResponse->isRedirectUrl()) {
 194                         $redirects++;
 195                         $this->logger->notice('Curl redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 196                         @curl_close($ch);
 197                         return $this->get($curlResponse->getRedirectUrl(), $binary, $opts, $redirects);
 198                 }
 199
 200                 @curl_close($ch);
 201
 202                 $this->profiler->saveTimestamp($stamp1, 'network');
 203
 204                 return $curlResponse;
 205         }
 206
 207         /**
 208          * {@inheritDoc}
 209          *
 210          * @param int $redirects The recursion counter for internal use - default 0
 211          *
 212          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 213          */
 214         public function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
 215         {
 216                 $stamp1 = microtime(true);
 217
 218                 if (Network::isUrlBlocked($url)) {
 219                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 220                         return CurlResult::createErrorCurl($url);
 221                 }
 222
 223                 $ch = curl_init($url);
 224
 225                 if (($redirects > 8) || (!$ch)) {
 226                         return CurlResult::createErrorCurl($url);
 227                 }
 228
 229                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 230
 231                 curl_setopt($ch, CURLOPT_HEADER, true);
 232                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 233                 curl_setopt($ch, CURLOPT_POST, 1);
 234                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 235                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 236
 237                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 238                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 239                 }
 240
 241                 if (intval($timeout)) {
 242                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 243                 } else {
 244                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 245                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 246                 }
 247
 248                 if (!empty($headers)) {
 249                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 250                 }
 251
 252                 $check_cert = $this->config->get('system', 'verifyssl');
 253                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 254
 255                 if ($check_cert) {
 256                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 257                 }
 258
 259                 $proxy = $this->config->get('system', 'proxy');
 260
 261                 if (!empty($proxy)) {
 262                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 263                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 264                         $proxyuser = $this->config->get('system', 'proxyuser');
 265                         if (!empty($proxyuser)) {
 266                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 267                         }
 268                 }
 269
 270                 // don't let curl abort the entire application
 271                 // if it throws any errors.
 272
 273                 $s = @curl_exec($ch);
 274
 275                 $curl_info = curl_getinfo($ch);
 276
 277                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 278
 279                 if ($curlResponse->isRedirectUrl()) {
 280                         $redirects++;
 281                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 282                         curl_close($ch);
 283                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 284                 }
 285
 286                 curl_close($ch);
 287
 288                 $this->profiler->saveTimestamp($stamp1, 'network');
 289
 290                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 291                 if ($curlResponse->getReturnCode() == 417) {
 292                         $redirects++;
 293
 294                         if (empty($headers)) {
 295                                 $headers = ['Expect:'];
 296                         } else {
 297                                 if (!in_array('Expect:', $headers)) {
 298                                         array_push($headers, 'Expect:');
 299                                 }
 300                         }
 301                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 302                         return $this->post($url, $params, $headers, $redirects, $timeout);
 303                 }
 304
 305                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 306
 307                 return $curlResponse;
 308         }
 309
 310         /**
 311          * {@inheritDoc}
 312          */
 313         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 314         {
 315                 if (Network::isUrlBlocked($url)) {
 316                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 317                         return $url;
 318                 }
 319
 320                 $url = Network::stripTrackingQueryParams($url);
 321
 322                 if ($depth > 10) {
 323                         return $url;
 324                 }
 325
 326                 $url = trim($url, "'");
 327
 328                 $stamp1 = microtime(true);
 329
 330                 $ch = curl_init();
 331                 curl_setopt($ch, CURLOPT_URL, $url);
 332                 curl_setopt($ch, CURLOPT_HEADER, 1);
 333                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 334                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 335                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 336                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 337
 338                 curl_exec($ch);
 339                 $curl_info = @curl_getinfo($ch);
 340                 $http_code = $curl_info['http_code'];
 341                 curl_close($ch);
 342
 343                 $this->profiler->saveTimestamp($stamp1, "network");
 344
 345                 if ($http_code == 0) {
 346                         return $url;
 347                 }
 348
 349                 if (in_array($http_code, ['301', '302'])) {
 350                         if (!empty($curl_info['redirect_url'])) {
 351                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 352                         } elseif (!empty($curl_info['location'])) {
 353                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 354                         }
 355                 }
 356
 357                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 358                 if (!$fetchbody) {
 359                         return $this->finalUrl($url, ++$depth, true);
 360                 }
 361
 362                 // if the file is too large then exit
 363                 if ($curl_info["download_content_length"] > 1000000) {
 364                         return $url;
 365                 }
 366
 367                 // if it isn't a HTML file then exit
 368                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 369                         return $url;
 370                 }
 371
 372                 $stamp1 = microtime(true);
 373
 374                 $ch = curl_init();
 375                 curl_setopt($ch, CURLOPT_URL, $url);
 376                 curl_setopt($ch, CURLOPT_HEADER, 0);
 377                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 378                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 379                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 380                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 381
 382                 $body = curl_exec($ch);
 383                 curl_close($ch);
 384
 385                 $this->profiler->saveTimestamp($stamp1, "network");
 386
 387                 if (trim($body) == "") {
 388                         return $url;
 389                 }
 390
 391                 // Check for redirect in meta elements
 392                 $doc = new DOMDocument();
 393                 @$doc->loadHTML($body);
 394
 395                 $xpath = new DomXPath($doc);
 396
 397                 $list = $xpath->query("//meta[@content]");
 398                 foreach ($list as $node) {
 399                         $attr = [];
 400                         if ($node->attributes->length) {
 401                                 foreach ($node->attributes as $attribute) {
 402                                         $attr[$attribute->name] = $attribute->value;
 403                                 }
 404                         }
 405
 406                         if (@$attr["http-equiv"] == 'refresh') {
 407                                 $path = $attr["content"];
 408                                 $pathinfo = explode(";", $path);
 409                                 foreach ($pathinfo as $value) {
 410                                         if (substr(strtolower($value), 0, 4) == "url=") {
 411                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 412                                         }
 413                                 }
 414                         }
 415                 }
 416
 417                 return $url;
 418         }
 419
 420         /**
 421          * {@inheritDoc}
 422          *
 423          * @param int $redirects The recursion counter for internal use - default 0
 424          *
 425          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 426          */
 427         public function fetch(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 428         {
 429                 $ret = $this->fetchFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
 430
 431                 return $ret->getBody();
 432         }
 433
 434         /**
 435          * {@inheritDoc}
 436          *
 437          * @param int $redirects The recursion counter for internal use - default 0
 438          *
 439          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 440          */
 441         public function fetchFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 442         {
 443                 return $this->get(
 444                         $url,
 445                         $binary,
 446                         [
 447                                 'timeout'        => $timeout,
 448                                 'accept_content' => $accept_content,
 449                                 'cookiejar'      => $cookiejar
 450                         ],
 451                         $redirects
 452                 );
 453         }
 454
 455         /**
 456          * {@inheritDoc}
 457          */
 458         public function getUserAgent()
 459         {
 460                 return
 461                         FRIENDICA_PLATFORM . " '" .
 462                         FRIENDICA_CODENAME . "' " .
 463                         FRIENDICA_VERSION . '-' .
 464                         DB_UPDATE_VERSION . '; ' .
 465                         $this->baseUrl;
 466         }
 467 }