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 GuzzleHttp\Exception\TransferException;
  34 use Psr\Http\Message\RequestInterface;
  35 use Psr\Http\Message\ResponseInterface;
  36 use Psr\Http\Message\UriInterface;
  37 use Psr\Log\LoggerInterface;
  38
  39 /**
  40  * Performs HTTP requests to a given URL
  41  */
  42 class HTTPRequest implements IHTTPRequest
  43 {
  44         /** @var LoggerInterface */
  45         private $logger;
  46         /** @var Profiler */
  47         private $profiler;
  48         /** @var IConfig */
  49         private $config;
  50         /** @var string */
  51         private $baseUrl;
  52
  53         public function __construct(LoggerInterface $logger, Profiler $profiler, IConfig $config, App\BaseURL $baseUrl)
  54         {
  55                 $this->logger   = $logger;
  56                 $this->profiler = $profiler;
  57                 $this->config   = $config;
  58                 $this->baseUrl  = $baseUrl->get();
  59         }
  60
  61         /**
  62          * {@inheritDoc}
  63          */
  64         public function get(string $url, bool $binary = false, array $opts = [])
  65         {
  66                 $stamp1 = microtime(true);
  67
  68                 if (strlen($url) > 1000) {
  69                         $this->logger->debug('URL is longer than 1000 characters.', ['url' => $url, 'callstack' => System::callstack(20)]);
  70                         return CurlResult::createErrorCurl(substr($url, 0, 200));
  71                 }
  72
  73                 $parts2     = [];
  74                 $parts      = parse_url($url);
  75                 $path_parts = explode('/', $parts['path'] ?? '');
  76                 foreach ($path_parts as $part) {
  77                         if (strlen($part) <> mb_strlen($part)) {
  78                                 $parts2[] = rawurlencode($part);
  79                         } else {
  80                                 $parts2[] = $part;
  81                         }
  82                 }
  83                 $parts['path'] = implode('/', $parts2);
  84                 $url           = Network::unparseURL($parts);
  85
  86                 if (Network::isUrlBlocked($url)) {
  87                         $this->logger->info('Domain is blocked.', ['url' => $url]);
  88                         return CurlResult::createErrorCurl($url);
  89                 }
  90
  91                 $curlOptions = [];
  92
  93                 $curlOptions[CURLOPT_HEADER] = true;
  94
  95                 if (!empty($opts['cookiejar'])) {
  96                         $curlOptions[CURLOPT_COOKIEJAR] = $opts["cookiejar"];
  97                         $curlOptions[CURLOPT_COOKIEFILE] = $opts["cookiejar"];
  98                 }
  99
 100                 // These settings aren't needed. We're following the location already.
 101                 //      $curlOptions[CURLOPT_FOLLOWLOCATION] =true;
 102                 //      $curlOptions[CURLOPT_MAXREDIRS] = 5;
 103
 104                 if (!empty($opts['accept_content'])) {
 105                         if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
 106                                 $curlOptions[CURLOPT_HTTPHEADER] = [];
 107                         }
 108                         array_push($curlOptions[CURLOPT_HTTPHEADER], 'Accept: ' . $opts['accept_content']);
 109                 }
 110
 111                 if (!empty($opts['header'])) {
 112                         if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
 113                                 $curlOptions[CURLOPT_HTTPHEADER] = [];
 114                         }
 115                         $curlOptions[CURLOPT_HTTPHEADER] = array_merge($opts['header'], $curlOptions[CURLOPT_HTTPHEADER]);
 116                 }
 117
 118                 $curlOptions[CURLOPT_RETURNTRANSFER] = true;
 119                 $curlOptions[CURLOPT_USERAGENT] = $this->getUserAgent();
 120
 121                 $range = intval($this->config->get('system', 'curl_range_bytes', 0));
 122
 123                 if ($range > 0) {
 124                         $curlOptions[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                 $curlOptions[CURLOPT_ENCODING] = '';
 131
 132                 if (!empty($opts['headers'])) {
 133                         if (empty($curlOptions[CURLOPT_HTTPHEADER])) {
 134                                 $curlOptions[CURLOPT_HTTPHEADER] = [];
 135                         }
 136                         $curlOptions[CURLOPT_HTTPHEADER] = array_merge($opts['headers'], $curlOptions[CURLOPT_HTTPHEADER]);
 137                 }
 138
 139                 if (!empty($opts['nobody'])) {
 140                         $curlOptions[CURLOPT_NOBODY] = $opts['nobody'];
 141                 }
 142
 143                 $curlOptions[CURLOPT_CONNECTTIMEOUT] = 10;
 144
 145                 if (!empty($opts['timeout'])) {
 146                         $curlOptions[CURLOPT_TIMEOUT] = $opts['timeout'];
 147                 } else {
 148                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 149                         $curlOptions[CURLOPT_TIMEOUT] = intval($curl_time);
 150                 }
 151
 152                 // by default we will allow self-signed certs
 153                 // but you can override this
 154
 155                 $check_cert = $this->config->get('system', 'verifyssl');
 156                 $curlOptions[CURLOPT_SSL_VERIFYPEER] = ($check_cert) ? true : false;
 157
 158                 if ($check_cert) {
 159                         $curlOptions[CURLOPT_SSL_VERIFYHOST] = 2;
 160                 }
 161
 162                 $proxy = $this->config->get('system', 'proxy');
 163
 164                 if (!empty($proxy)) {
 165                         $curlOptions[CURLOPT_HTTPPROXYTUNNEL] = 1;
 166                         $curlOptions[CURLOPT_PROXY] = $proxy;
 167                         $proxyuser = $this->config->get('system', 'proxyuser');
 168
 169                         if (!empty($proxyuser)) {
 170                                 $curlOptions[CURLOPT_PROXYUSERPWD] = $proxyuser;
 171                         }
 172                 }
 173
 174                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 175                         $curlOptions[CURLOPT_IPRESOLVE] = CURL_IPRESOLVE_V4;
 176                 }
 177
 178                 if ($binary) {
 179                         $curlOptions[CURLOPT_BINARYTRANSFER] = 1;
 180                 }
 181
 182                 $onRedirect = function(
 183                         RequestInterface $request,
 184                         ResponseInterface $response,
 185                         UriInterface $uri
 186                 ) {
 187                         $this->logger->notice('Curl redirect.', ['url' => $request->getUri(), 'to' => $uri]);
 188                 };
 189
 190                 $onHeaders = function (ResponseInterface $response) use ($opts) {
 191                         if (!empty($opts['content_length']) &&
 192                                 $response->getHeaderLine('Content-Length') > $opts['content_length']) {
 193                                 throw new TransferException('The file is too big!');
 194                         }
 195                 };
 196
 197                 $client = new Client([
 198                         'allow_redirect' => [
 199                                 'max' => 8,
 200                                 'on_redirect' => $onRedirect,
 201                                 'track_redirect' => true,
 202                                 'strict' => true,
 203                                 'referer' => true,
 204                         ],
 205                         'on_headers' => $onHeaders,
 206                         'sink' => tempnam(get_temppath(), 'guzzle'),
 207                         'curl' => $curlOptions
 208                 ]);
 209
 210                 try {
 211                         $response = $client->get($url);
 212                         return new GuzzleResponse($response, $url);
 213                 } catch (TransferException $exception) {
 214                         if ($exception instanceof RequestException &&
 215                                 $exception->hasResponse()) {
 216                                 return new GuzzleResponse($exception->getResponse(), $url, $exception->getCode(), $exception->getMessage());
 217                         } else {
 218                                 return new CurlResult($url, '', ['http_code' => $exception->getCode()], $exception->getCode(), $exception->getMessage());
 219                         }
 220                 } finally {
 221                         $this->profiler->saveTimestamp($stamp1, 'network');
 222                 }
 223         }
 224
 225         /**
 226          * {@inheritDoc}
 227          *
 228          * @param int $redirects The recursion counter for internal use - default 0
 229          *
 230          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 231          */
 232         public function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
 233         {
 234                 $stamp1 = microtime(true);
 235
 236                 if (Network::isUrlBlocked($url)) {
 237                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 238                         return CurlResult::createErrorCurl($url);
 239                 }
 240
 241                 $ch = curl_init($url);
 242
 243                 if (($redirects > 8) || (!$ch)) {
 244                         return CurlResult::createErrorCurl($url);
 245                 }
 246
 247                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 248
 249                 curl_setopt($ch, CURLOPT_HEADER, true);
 250                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 251                 curl_setopt($ch, CURLOPT_POST, 1);
 252                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 253                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 254
 255                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 256                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 257                 }
 258
 259                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 260
 261                 if (intval($timeout)) {
 262                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 263                 } else {
 264                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 265                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 266                 }
 267
 268                 if (!empty($headers)) {
 269                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 270                 }
 271
 272                 $check_cert = $this->config->get('system', 'verifyssl');
 273                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 274
 275                 if ($check_cert) {
 276                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 277                 }
 278
 279                 $proxy = $this->config->get('system', 'proxy');
 280
 281                 if (!empty($proxy)) {
 282                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 283                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 284                         $proxyuser = $this->config->get('system', 'proxyuser');
 285                         if (!empty($proxyuser)) {
 286                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 287                         }
 288                 }
 289
 290                 // don't let curl abort the entire application
 291                 // if it throws any errors.
 292
 293                 $s = @curl_exec($ch);
 294
 295                 $curl_info = curl_getinfo($ch);
 296
 297                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 298
 299                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 300                         $redirects++;
 301                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 302                         curl_close($ch);
 303                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 304                 }
 305
 306                 curl_close($ch);
 307
 308                 $this->profiler->saveTimestamp($stamp1, 'network');
 309
 310                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 311                 if ($curlResponse->getReturnCode() == 417) {
 312                         $redirects++;
 313
 314                         if (empty($headers)) {
 315                                 $headers = ['Expect:'];
 316                         } else {
 317                                 if (!in_array('Expect:', $headers)) {
 318                                         array_push($headers, 'Expect:');
 319                                 }
 320                         }
 321                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 322                         return $this->post($url, $params, $headers, $redirects, $timeout);
 323                 }
 324
 325                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 326
 327                 return $curlResponse;
 328         }
 329
 330         /**
 331          * {@inheritDoc}
 332          */
 333         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 334         {
 335                 if (Network::isUrlBlocked($url)) {
 336                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 337                         return $url;
 338                 }
 339
 340                 if (Network::isRedirectBlocked($url)) {
 341                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 342                         return $url;
 343                 }
 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_CONNECTTIMEOUT, 10);
 360                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 361                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 362                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 363
 364                 curl_exec($ch);
 365                 $curl_info = @curl_getinfo($ch);
 366                 $http_code = $curl_info['http_code'];
 367                 curl_close($ch);
 368
 369                 $this->profiler->saveTimestamp($stamp1, "network");
 370
 371                 if ($http_code == 0) {
 372                         return $url;
 373                 }
 374
 375                 if (in_array($http_code, ['301', '302'])) {
 376                         if (!empty($curl_info['redirect_url'])) {
 377                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 378                         } elseif (!empty($curl_info['location'])) {
 379                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 380                         }
 381                 }
 382
 383                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 384                 if (!$fetchbody) {
 385                         return $this->finalUrl($url, ++$depth, true);
 386                 }
 387
 388                 // if the file is too large then exit
 389                 if ($curl_info["download_content_length"] > 1000000) {
 390                         return $url;
 391                 }
 392
 393                 // if it isn't a HTML file then exit
 394                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 395                         return $url;
 396                 }
 397
 398                 $stamp1 = microtime(true);
 399
 400                 $ch = curl_init();
 401                 curl_setopt($ch, CURLOPT_URL, $url);
 402                 curl_setopt($ch, CURLOPT_HEADER, 0);
 403                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 404                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 405                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 406                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 407                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 408
 409                 $body = curl_exec($ch);
 410                 curl_close($ch);
 411
 412                 $this->profiler->saveTimestamp($stamp1, "network");
 413
 414                 if (trim($body) == "") {
 415                         return $url;
 416                 }
 417
 418                 // Check for redirect in meta elements
 419                 $doc = new DOMDocument();
 420                 @$doc->loadHTML($body);
 421
 422                 $xpath = new DomXPath($doc);
 423
 424                 $list = $xpath->query("//meta[@content]");
 425                 foreach ($list as $node) {
 426                         $attr = [];
 427                         if ($node->attributes->length) {
 428                                 foreach ($node->attributes as $attribute) {
 429                                         $attr[$attribute->name] = $attribute->value;
 430                                 }
 431                         }
 432
 433                         if (@$attr["http-equiv"] == 'refresh') {
 434                                 $path = $attr["content"];
 435                                 $pathinfo = explode(";", $path);
 436                                 foreach ($pathinfo as $value) {
 437                                         if (substr(strtolower($value), 0, 4) == "url=") {
 438                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 439                                         }
 440                                 }
 441                         }
 442                 }
 443
 444                 return $url;
 445         }
 446
 447         /**
 448          * {@inheritDoc}
 449          *
 450          * @param int $redirects The recursion counter for internal use - default 0
 451          *
 452          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 453          */
 454         public function fetch(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 455         {
 456                 $ret = $this->fetchFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
 457
 458                 return $ret->getBody();
 459         }
 460
 461         /**
 462          * {@inheritDoc}
 463          *
 464          * @param int $redirects The recursion counter for internal use - default 0
 465          *
 466          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 467          */
 468         public function fetchFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 469         {
 470                 return $this->get(
 471                         $url,
 472                         $binary,
 473                         [
 474                                 'timeout'        => $timeout,
 475                                 'accept_content' => $accept_content,
 476                                 'cookiejar'      => $cookiejar
 477                         ]
 478                 );
 479         }
 480
 481         /**
 482          * {@inheritDoc}
 483          */
 484         public function getUserAgent()
 485         {
 486                 return
 487                         FRIENDICA_PLATFORM . " '" .
 488                         FRIENDICA_CODENAME . "' " .
 489                         FRIENDICA_VERSION . '-' .
 490                         DB_UPDATE_VERSION . '; ' .
 491                         $this->baseUrl;
 492         }
 493 }