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 GuzzleHttp\Client;
  32 use GuzzleHttp\Exception\RequestException;
  33 use Psr\Http\Message\RequestInterface;
  34 use Psr\Http\Message\ResponseInterface;
  35 use Psr\Http\Message\UriInterface;
  36 use Psr\Log\LoggerInterface;
  37
  38 /**
  39  * Performs HTTP requests to a given URL
  40  */
  41 class HTTPRequest implements IHTTPRequest
  42 {
  43         /** @var LoggerInterface */
  44         private $logger;
  45         /** @var Profiler */
  46         private $profiler;
  47         /** @var IConfig */
  48         private $config;
  49         /** @var string */
  50         private $baseUrl;
  51
  52         public function __construct(LoggerInterface $logger, Profiler $profiler, IConfig $config, App\BaseURL $baseUrl)
  53         {
  54                 $this->logger   = $logger;
  55                 $this->profiler = $profiler;
  56                 $this->config   = $config;
  57                 $this->baseUrl  = $baseUrl->get();
  58         }
  59
  60         /**
  61          * {@inheritDoc}
  62          */
  63         public function get(string $url, bool $binary = false, array $opts = [])
  64         {
  65                 $stamp1 = microtime(true);
  66
  67                 if (strlen($url) > 1000) {
  68                         $this->logger->debug('URL is longer than 1000 characters.', ['url' => $url, 'callstack' => System::callstack(20)]);
  69                         return CurlResult::createErrorCurl(substr($url, 0, 200));
  70                 }
  71
  72                 $parts2     = [];
  73                 $parts      = parse_url($url);
  74                 $path_parts = explode('/', $parts['path'] ?? '');
  75                 foreach ($path_parts as $part) {
  76                         if (strlen($part) <> mb_strlen($part)) {
  77                                 $parts2[] = rawurlencode($part);
  78                         } else {
  79                                 $parts2[] = $part;
  80                         }
  81                 }
  82                 $parts['path'] = implode('/', $parts2);
  83                 $url           = Network::unparseURL($parts);
  84
  85                 if (Network::isUrlBlocked($url)) {
  86                         $this->logger->info('Domain is blocked.', ['url' => $url]);
  87                         return CurlResult::createErrorCurl($url);
  88                 }
  89
  90                 $curlOptions = [];
  91
  92                 $curlOptions[CURLOPT_HEADER] = true;
  93
  94                 if (!empty($opts['cookiejar'])) {
  95                         $curlOptions[CURLOPT_COOKIEJAR] = $opts["cookiejar"];
  96                         $curlOptions[CURLOPT_COOKIEFILE] = $opts["cookiejar"];
  97                 }
  98
  99                 // These settings aren't needed. We're following the location already.
 100                 //      $curlOptions[CURLOPT_FOLLOWLOCATION] =true;
 101                 //      $curlOptions[CURLOPT_MAXREDIRS] = 5;
 102
 103                 if (!empty($opts['accept_content'])) {
 104                         $curlOptions[CURLOPT_HTTPHEADER][] = ['Accept: ' . $opts['accept_content']];
 105                 }
 106
 107                 if (!empty($opts['header'])) {
 108                         $curlOptions[CURLOPT_HTTPHEADER][] = $opts['header'];
 109                 }
 110
 111                 $curlOptions[CURLOPT_RETURNTRANSFER] = true;
 112                 $curlOptions[CURLOPT_USERAGENT] = $this->getUserAgent();
 113
 114                 $range = intval($this->config->get('system', 'curl_range_bytes', 0));
 115
 116                 if ($range > 0) {
 117                         $curlOptions[CURLOPT_RANGE] = '0-' . $range;
 118                 }
 119
 120                 // Without this setting it seems as if some webservers send compressed content
 121                 // This seems to confuse curl so that it shows this uncompressed.
 122                 /// @todo  We could possibly set this value to "gzip" or something similar
 123                 $curlOptions[CURLOPT_ENCODING] = '';
 124
 125                 if (!empty($opts['headers'])) {
 126                         $curlOptions[CURLOPT_HTTPHEADER][] = $opts['headers'];
 127                 }
 128
 129                 if (!empty($opts['nobody'])) {
 130                         $curlOptions[CURLOPT_NOBODY] = $opts['nobody'];
 131                 }
 132
 133                 $curlOptions[CURLOPT_CONNECTTIMEOUT] = 10;
 134
 135                 if (!empty($opts['timeout'])) {
 136                         $curlOptions[CURLOPT_TIMEOUT] = $opts['timeout'];
 137                 } else {
 138                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 139                         $curlOptions[CURLOPT_TIMEOUT] = intval($curl_time);
 140                 }
 141
 142                 // by default we will allow self-signed certs
 143                 // but you can override this
 144
 145                 $check_cert = $this->config->get('system', 'verifyssl');
 146                 $curlOptions[CURLOPT_SSL_VERIFYPEER] = ($check_cert) ? true : false;
 147
 148                 if ($check_cert) {
 149                         $curlOptions[CURLOPT_SSL_VERIFYHOST] = 2;
 150                 }
 151
 152                 $proxy = $this->config->get('system', 'proxy');
 153
 154                 if (!empty($proxy)) {
 155                         $curlOptions[CURLOPT_HTTPPROXYTUNNEL] = 1;
 156                         $curlOptions[CURLOPT_PROXY] = $proxy;
 157                         $proxyuser = $this->config->get('system', 'proxyuser');
 158
 159                         if (!empty($proxyuser)) {
 160                                 $curlOptions[CURLOPT_PROXYUSERPWD] = $proxyuser;
 161                         }
 162                 }
 163
 164                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 165                         $curlOptions[CURLOPT_IPRESOLVE] = CURL_IPRESOLVE_V4;
 166                 }
 167
 168                 if ($binary) {
 169                         $curlOptions[CURLOPT_BINARYTRANSFER] = 1;
 170                 }
 171
 172                 $onRedirect = function(
 173                         RequestInterface $request,
 174                         ResponseInterface $response,
 175                         UriInterface $uri
 176                 ) {
 177                         $this->logger->notice('Curl redirect.', ['url' => $request->getUri(), 'to' => $uri]);
 178                 };
 179
 180                 $client = new Client([
 181                         'allow_redirect' => [
 182                                 'max' => 8,
 183                                 'on_redirect' => $onRedirect,
 184                                 'track_redirect' => true,
 185                                 'strict' => true,
 186                                 'referer' => true,
 187                         ],
 188                         'curl' => $curlOptions
 189                 ]);
 190
 191                 try {
 192                         $response = $client->get($url);
 193                         return new GuzzleResponse($response, $url);
 194                 } catch (RequestException $exception) {
 195                         if ($exception->hasResponse()) {
 196                                 return new GuzzleResponse($exception->getResponse(), $url, $exception->getCode(), $exception->getMessage());
 197                         } else {
 198                                 return new CurlResult($url, '', ['http_code' => $exception->getCode()], $exception->getCode(), $exception->getMessage());
 199                         }
 200                 } finally {
 201                         $this->profiler->saveTimestamp($stamp1, 'network');
 202                 }
 203         }
 204
 205         /**
 206          * {@inheritDoc}
 207          *
 208          * @param int $redirects The recursion counter for internal use - default 0
 209          *
 210          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 211          */
 212         public function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
 213         {
 214                 $stamp1 = microtime(true);
 215
 216                 if (Network::isUrlBlocked($url)) {
 217                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 218                         return CurlResult::createErrorCurl($url);
 219                 }
 220
 221                 $ch = curl_init($url);
 222
 223                 if (($redirects > 8) || (!$ch)) {
 224                         return CurlResult::createErrorCurl($url);
 225                 }
 226
 227                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 228
 229                 curl_setopt($ch, CURLOPT_HEADER, true);
 230                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 231                 curl_setopt($ch, CURLOPT_POST, 1);
 232                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 233                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 234
 235                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 236                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 237                 }
 238
 239                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 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($ch));
 278
 279                 if (!Network::isRedirectBlocked($url) && $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 $url;
 318                 }
 319
 320                 if (Network::isRedirectBlocked($url)) {
 321                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 322                         return $url;
 323                 }
 324
 325                 $url = Network::stripTrackingQueryParams($url);
 326
 327                 if ($depth > 10) {
 328                         return $url;
 329                 }
 330
 331                 $url = trim($url, "'");
 332
 333                 $stamp1 = microtime(true);
 334
 335                 $ch = curl_init();
 336                 curl_setopt($ch, CURLOPT_URL, $url);
 337                 curl_setopt($ch, CURLOPT_HEADER, 1);
 338                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 339                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 340                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 341                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 342                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 343
 344                 curl_exec($ch);
 345                 $curl_info = @curl_getinfo($ch);
 346                 $http_code = $curl_info['http_code'];
 347                 curl_close($ch);
 348
 349                 $this->profiler->saveTimestamp($stamp1, "network");
 350
 351                 if ($http_code == 0) {
 352                         return $url;
 353                 }
 354
 355                 if (in_array($http_code, ['301', '302'])) {
 356                         if (!empty($curl_info['redirect_url'])) {
 357                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 358                         } elseif (!empty($curl_info['location'])) {
 359                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 360                         }
 361                 }
 362
 363                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 364                 if (!$fetchbody) {
 365                         return $this->finalUrl($url, ++$depth, true);
 366                 }
 367
 368                 // if the file is too large then exit
 369                 if ($curl_info["download_content_length"] > 1000000) {
 370                         return $url;
 371                 }
 372
 373                 // if it isn't a HTML file then exit
 374                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 375                         return $url;
 376                 }
 377
 378                 $stamp1 = microtime(true);
 379
 380                 $ch = curl_init();
 381                 curl_setopt($ch, CURLOPT_URL, $url);
 382                 curl_setopt($ch, CURLOPT_HEADER, 0);
 383                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 384                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 385                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 386                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 387                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 388
 389                 $body = curl_exec($ch);
 390                 curl_close($ch);
 391
 392                 $this->profiler->saveTimestamp($stamp1, "network");
 393
 394                 if (trim($body) == "") {
 395                         return $url;
 396                 }
 397
 398                 // Check for redirect in meta elements
 399                 $doc = new DOMDocument();
 400                 @$doc->loadHTML($body);
 401
 402                 $xpath = new DomXPath($doc);
 403
 404                 $list = $xpath->query("//meta[@content]");
 405                 foreach ($list as $node) {
 406                         $attr = [];
 407                         if ($node->attributes->length) {
 408                                 foreach ($node->attributes as $attribute) {
 409                                         $attr[$attribute->name] = $attribute->value;
 410                                 }
 411                         }
 412
 413                         if (@$attr["http-equiv"] == 'refresh') {
 414                                 $path = $attr["content"];
 415                                 $pathinfo = explode(";", $path);
 416                                 foreach ($pathinfo as $value) {
 417                                         if (substr(strtolower($value), 0, 4) == "url=") {
 418                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 419                                         }
 420                                 }
 421                         }
 422                 }
 423
 424                 return $url;
 425         }
 426
 427         /**
 428          * {@inheritDoc}
 429          *
 430          * @param int $redirects The recursion counter for internal use - default 0
 431          *
 432          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 433          */
 434         public function fetch(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 435         {
 436                 $ret = $this->fetchFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
 437
 438                 return $ret->getBody();
 439         }
 440
 441         /**
 442          * {@inheritDoc}
 443          *
 444          * @param int $redirects The recursion counter for internal use - default 0
 445          *
 446          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 447          */
 448         public function fetchFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 449         {
 450                 return $this->get(
 451                         $url,
 452                         $binary,
 453                         [
 454                                 'timeout'        => $timeout,
 455                                 'accept_content' => $accept_content,
 456                                 'cookiejar'      => $cookiejar
 457                         ]
 458                 );
 459         }
 460
 461         /**
 462          * {@inheritDoc}
 463          */
 464         public function getUserAgent()
 465         {
 466                 return
 467                         FRIENDICA_PLATFORM . " '" .
 468                         FRIENDICA_CODENAME . "' " .
 469                         FRIENDICA_VERSION . '-' .
 470                         DB_UPDATE_VERSION . '; ' .
 471                         $this->baseUrl;
 472         }
 473 }