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                         if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
 105                                 $curlOptions[CURLOPT_HTTPHEADER] = [];
 106                         }
 107                         array_push($curlOptions[CURLOPT_HTTPHEADER], 'Accept: ' . $opts['accept_content']);
 108                 }
 109
 110                 if (!empty($opts['header'])) {
 111                         if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
 112                                 $curlOptions[CURLOPT_HTTPHEADER] = [];
 113                         }
 114                         $curlOptions[CURLOPT_HTTPHEADER] = array_merge($opts['header'], $curlOptions[CURLOPT_HTTPHEADER]);
 115                 }
 116
 117                 $curlOptions[CURLOPT_RETURNTRANSFER] = true;
 118                 $curlOptions[CURLOPT_USERAGENT] = $this->getUserAgent();
 119
 120                 $range = intval($this->config->get('system', 'curl_range_bytes', 0));
 121
 122                 if ($range > 0) {
 123                         $curlOptions[CURLOPT_RANGE] = '0-' . $range;
 124                 }
 125
 126                 // Without this setting it seems as if some webservers send compressed content
 127                 // This seems to confuse curl so that it shows this uncompressed.
 128                 /// @todo  We could possibly set this value to "gzip" or something similar
 129                 $curlOptions[CURLOPT_ENCODING] = '';
 130
 131                 if (!empty($opts['headers'])) {
 132                         if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
 133                                 $curlOptions[CURLOPT_HTTPHEADER] = [];
 134                         }
 135                         $curlOptions[CURLOPT_HTTPHEADER] = array_merge($opts['headers'], $curlOptions[CURLOPT_HTTPHEADER]);
 136                 }
 137
 138                 if (!empty($opts['nobody'])) {
 139                         $curlOptions[CURLOPT_NOBODY] = $opts['nobody'];
 140                 }
 141
 142                 $curlOptions[CURLOPT_CONNECTTIMEOUT] = 10;
 143
 144                 if (!empty($opts['timeout'])) {
 145                         $curlOptions[CURLOPT_TIMEOUT] = $opts['timeout'];
 146                 } else {
 147                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 148                         $curlOptions[CURLOPT_TIMEOUT] = intval($curl_time);
 149                 }
 150
 151                 // by default we will allow self-signed certs
 152                 // but you can override this
 153
 154                 $check_cert = $this->config->get('system', 'verifyssl');
 155                 $curlOptions[CURLOPT_SSL_VERIFYPEER] = ($check_cert) ? true : false;
 156
 157                 if ($check_cert) {
 158                         $curlOptions[CURLOPT_SSL_VERIFYHOST] = 2;
 159                 }
 160
 161                 $proxy = $this->config->get('system', 'proxy');
 162
 163                 if (!empty($proxy)) {
 164                         $curlOptions[CURLOPT_HTTPPROXYTUNNEL] = 1;
 165                         $curlOptions[CURLOPT_PROXY] = $proxy;
 166                         $proxyuser = $this->config->get('system', 'proxyuser');
 167
 168                         if (!empty($proxyuser)) {
 169                                 $curlOptions[CURLOPT_PROXYUSERPWD] = $proxyuser;
 170                         }
 171                 }
 172
 173                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 174                         $curlOptions[CURLOPT_IPRESOLVE] = CURL_IPRESOLVE_V4;
 175                 }
 176
 177                 if ($binary) {
 178                         $curlOptions[CURLOPT_BINARYTRANSFER] = 1;
 179                 }
 180
 181                 $onRedirect = function(
 182                         RequestInterface $request,
 183                         ResponseInterface $response,
 184                         UriInterface $uri
 185                 ) {
 186                         $this->logger->notice('Curl redirect.', ['url' => $request->getUri(), 'to' => $uri]);
 187                 };
 188
 189                 $client = new Client([
 190                         'allow_redirect' => [
 191                                 'max' => 8,
 192                                 'on_redirect' => $onRedirect,
 193                                 'track_redirect' => true,
 194                                 'strict' => true,
 195                                 'referer' => true,
 196                         ],
 197                         'on_headers' => $onHeaders,
 198                         'sink' => tempnam(get_temppath(), 'guzzle'),
 199                         'curl' => $curlOptions
 200                 ]);
 201
 202                 try {
 203                         $response = $client->get($url);
 204                         return new GuzzleResponse($response, $url);
 205                 } catch (RequestException $exception) {
 206                         if ($exception->hasResponse()) {
 207                                 return new GuzzleResponse($exception->getResponse(), $url, $exception->getCode(), $exception->getMessage());
 208                         } else {
 209                                 return new CurlResult($url, '', ['http_code' => $exception->getCode()], $exception->getCode(), $exception->getMessage());
 210                         }
 211                 } finally {
 212                         $this->profiler->saveTimestamp($stamp1, 'network');
 213                 }
 214         }
 215
 216         /**
 217          * {@inheritDoc}
 218          *
 219          * @param int $redirects The recursion counter for internal use - default 0
 220          *
 221          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 222          */
 223         public function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
 224         {
 225                 $stamp1 = microtime(true);
 226
 227                 if (Network::isUrlBlocked($url)) {
 228                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 229                         return CurlResult::createErrorCurl($url);
 230                 }
 231
 232                 $ch = curl_init($url);
 233
 234                 if (($redirects > 8) || (!$ch)) {
 235                         return CurlResult::createErrorCurl($url);
 236                 }
 237
 238                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 239
 240                 curl_setopt($ch, CURLOPT_HEADER, true);
 241                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 242                 curl_setopt($ch, CURLOPT_POST, 1);
 243                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 244                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 245
 246                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 247                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 248                 }
 249
 250                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 251
 252                 if (intval($timeout)) {
 253                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 254                 } else {
 255                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 256                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 257                 }
 258
 259                 if (!empty($headers)) {
 260                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 261                 }
 262
 263                 $check_cert = $this->config->get('system', 'verifyssl');
 264                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 265
 266                 if ($check_cert) {
 267                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 268                 }
 269
 270                 $proxy = $this->config->get('system', 'proxy');
 271
 272                 if (!empty($proxy)) {
 273                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 274                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 275                         $proxyuser = $this->config->get('system', 'proxyuser');
 276                         if (!empty($proxyuser)) {
 277                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 278                         }
 279                 }
 280
 281                 // don't let curl abort the entire application
 282                 // if it throws any errors.
 283
 284                 $s = @curl_exec($ch);
 285
 286                 $curl_info = curl_getinfo($ch);
 287
 288                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 289
 290                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 291                         $redirects++;
 292                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 293                         curl_close($ch);
 294                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 295                 }
 296
 297                 curl_close($ch);
 298
 299                 $this->profiler->saveTimestamp($stamp1, 'network');
 300
 301                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 302                 if ($curlResponse->getReturnCode() == 417) {
 303                         $redirects++;
 304
 305                         if (empty($headers)) {
 306                                 $headers = ['Expect:'];
 307                         } else {
 308                                 if (!in_array('Expect:', $headers)) {
 309                                         array_push($headers, 'Expect:');
 310                                 }
 311                         }
 312                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 313                         return $this->post($url, $params, $headers, $redirects, $timeout);
 314                 }
 315
 316                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 317
 318                 return $curlResponse;
 319         }
 320
 321         /**
 322          * {@inheritDoc}
 323          */
 324         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 325         {
 326                 if (Network::isUrlBlocked($url)) {
 327                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 328                         return $url;
 329                 }
 330
 331                 if (Network::isRedirectBlocked($url)) {
 332                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 333                         return $url;
 334                 }
 335
 336                 $url = Network::stripTrackingQueryParams($url);
 337
 338                 if ($depth > 10) {
 339                         return $url;
 340                 }
 341
 342                 $url = trim($url, "'");
 343
 344                 $stamp1 = microtime(true);
 345
 346                 $ch = curl_init();
 347                 curl_setopt($ch, CURLOPT_URL, $url);
 348                 curl_setopt($ch, CURLOPT_HEADER, 1);
 349                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 350                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 351                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 352                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 353                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 354
 355                 curl_exec($ch);
 356                 $curl_info = @curl_getinfo($ch);
 357                 $http_code = $curl_info['http_code'];
 358                 curl_close($ch);
 359
 360                 $this->profiler->saveTimestamp($stamp1, "network");
 361
 362                 if ($http_code == 0) {
 363                         return $url;
 364                 }
 365
 366                 if (in_array($http_code, ['301', '302'])) {
 367                         if (!empty($curl_info['redirect_url'])) {
 368                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 369                         } elseif (!empty($curl_info['location'])) {
 370                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 371                         }
 372                 }
 373
 374                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 375                 if (!$fetchbody) {
 376                         return $this->finalUrl($url, ++$depth, true);
 377                 }
 378
 379                 // if the file is too large then exit
 380                 if ($curl_info["download_content_length"] > 1000000) {
 381                         return $url;
 382                 }
 383
 384                 // if it isn't a HTML file then exit
 385                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 386                         return $url;
 387                 }
 388
 389                 $stamp1 = microtime(true);
 390
 391                 $ch = curl_init();
 392                 curl_setopt($ch, CURLOPT_URL, $url);
 393                 curl_setopt($ch, CURLOPT_HEADER, 0);
 394                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 395                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 396                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 397                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 398                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 399
 400                 $body = curl_exec($ch);
 401                 curl_close($ch);
 402
 403                 $this->profiler->saveTimestamp($stamp1, "network");
 404
 405                 if (trim($body) == "") {
 406                         return $url;
 407                 }
 408
 409                 // Check for redirect in meta elements
 410                 $doc = new DOMDocument();
 411                 @$doc->loadHTML($body);
 412
 413                 $xpath = new DomXPath($doc);
 414
 415                 $list = $xpath->query("//meta[@content]");
 416                 foreach ($list as $node) {
 417                         $attr = [];
 418                         if ($node->attributes->length) {
 419                                 foreach ($node->attributes as $attribute) {
 420                                         $attr[$attribute->name] = $attribute->value;
 421                                 }
 422                         }
 423
 424                         if (@$attr["http-equiv"] == 'refresh') {
 425                                 $path = $attr["content"];
 426                                 $pathinfo = explode(";", $path);
 427                                 foreach ($pathinfo as $value) {
 428                                         if (substr(strtolower($value), 0, 4) == "url=") {
 429                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 430                                         }
 431                                 }
 432                         }
 433                 }
 434
 435                 return $url;
 436         }
 437
 438         /**
 439          * {@inheritDoc}
 440          *
 441          * @param int $redirects The recursion counter for internal use - default 0
 442          *
 443          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 444          */
 445         public function fetch(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 446         {
 447                 $ret = $this->fetchFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
 448
 449                 return $ret->getBody();
 450         }
 451
 452         /**
 453          * {@inheritDoc}
 454          *
 455          * @param int $redirects The recursion counter for internal use - default 0
 456          *
 457          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 458          */
 459         public function fetchFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 460         {
 461                 return $this->get(
 462                         $url,
 463                         $binary,
 464                         [
 465                                 'timeout'        => $timeout,
 466                                 'accept_content' => $accept_content,
 467                                 'cookiejar'      => $cookiejar
 468                         ]
 469                 );
 470         }
 471
 472         /**
 473          * {@inheritDoc}
 474          */
 475         public function getUserAgent()
 476         {
 477                 return
 478                         FRIENDICA_PLATFORM . " '" .
 479                         FRIENDICA_CODENAME . "' " .
 480                         FRIENDICA_VERSION . '-' .
 481                         DB_UPDATE_VERSION . '; ' .
 482                         $this->baseUrl;
 483         }
 484 }