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                 $logger = $this->logger;
 183
 184                 $onRedirect = function(
 185                         RequestInterface $request,
 186                         ResponseInterface $response,
 187                         UriInterface $uri
 188                 ) use ($logger) {
 189                         $logger->notice('Curl redirect.', ['url' => $request->getUri(), 'to' => $uri]);
 190                 };
 191
 192                 $onHeaders = function (ResponseInterface $response) use ($opts) {
 193                         if (!empty($opts['content_length']) &&
 194                                 $response->getHeaderLine('Content-Length') > $opts['content_length']) {
 195                                 throw new TransferException('The file is too big!');
 196                         }
 197                 };
 198
 199                 $client = new Client([
 200                         'allow_redirect' => [
 201                                 'max' => 8,
 202                                 'on_redirect' => $onRedirect,
 203                                 'track_redirect' => true,
 204                                 'strict' => true,
 205                                 'referer' => true,
 206                         ],
 207                         'on_headers' => $onHeaders,
 208                         'sink' => tempnam(get_temppath(), 'guzzle'),
 209                         'curl' => $curlOptions
 210                 ]);
 211
 212                 try {
 213                         $response = $client->get($url);
 214                         return new GuzzleResponse($response, $url);
 215                 } catch (TransferException $exception) {
 216                         if ($exception instanceof RequestException &&
 217                                 $exception->hasResponse()) {
 218                                 return new GuzzleResponse($exception->getResponse(), $url, $exception->getCode(), $exception->getMessage());
 219                         } else {
 220                                 return new CurlResult($url, '', ['http_code' => $exception->getCode()], $exception->getCode(), $exception->getMessage());
 221                         }
 222                 } finally {
 223                         $this->profiler->saveTimestamp($stamp1, 'network');
 224                 }
 225         }
 226
 227         /**
 228          * {@inheritDoc}
 229          *
 230          * @param int $redirects The recursion counter for internal use - default 0
 231          *
 232          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 233          */
 234         public function post(string $url, $params, array $headers = [], int $timeout = 0, int &$redirects = 0)
 235         {
 236                 $stamp1 = microtime(true);
 237
 238                 if (Network::isUrlBlocked($url)) {
 239                         $this->logger->info('Domain is blocked.' . ['url' => $url]);
 240                         return CurlResult::createErrorCurl($url);
 241                 }
 242
 243                 $ch = curl_init($url);
 244
 245                 if (($redirects > 8) || (!$ch)) {
 246                         return CurlResult::createErrorCurl($url);
 247                 }
 248
 249                 $this->logger->debug('Post_url: start.', ['url' => $url]);
 250
 251                 curl_setopt($ch, CURLOPT_HEADER, true);
 252                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 253                 curl_setopt($ch, CURLOPT_POST, 1);
 254                 curl_setopt($ch, CURLOPT_POSTFIELDS, $params);
 255                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 256
 257                 if ($this->config->get('system', 'ipv4_resolve', false)) {
 258                         curl_setopt($ch, CURLOPT_IPRESOLVE, CURL_IPRESOLVE_V4);
 259                 }
 260
 261                 @curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 262
 263                 if (intval($timeout)) {
 264                         curl_setopt($ch, CURLOPT_TIMEOUT, $timeout);
 265                 } else {
 266                         $curl_time = $this->config->get('system', 'curl_timeout', 60);
 267                         curl_setopt($ch, CURLOPT_TIMEOUT, intval($curl_time));
 268                 }
 269
 270                 if (!empty($headers)) {
 271                         curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
 272                 }
 273
 274                 $check_cert = $this->config->get('system', 'verifyssl');
 275                 curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, (($check_cert) ? true : false));
 276
 277                 if ($check_cert) {
 278                         @curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
 279                 }
 280
 281                 $proxy = $this->config->get('system', 'proxy');
 282
 283                 if (!empty($proxy)) {
 284                         curl_setopt($ch, CURLOPT_HTTPPROXYTUNNEL, 1);
 285                         curl_setopt($ch, CURLOPT_PROXY, $proxy);
 286                         $proxyuser = $this->config->get('system', 'proxyuser');
 287                         if (!empty($proxyuser)) {
 288                                 curl_setopt($ch, CURLOPT_PROXYUSERPWD, $proxyuser);
 289                         }
 290                 }
 291
 292                 // don't let curl abort the entire application
 293                 // if it throws any errors.
 294
 295                 $s = @curl_exec($ch);
 296
 297                 $curl_info = curl_getinfo($ch);
 298
 299                 $curlResponse = new CurlResult($url, $s, $curl_info, curl_errno($ch), curl_error($ch));
 300
 301                 if (!Network::isRedirectBlocked($url) && $curlResponse->isRedirectUrl()) {
 302                         $redirects++;
 303                         $this->logger->info('Post redirect.', ['url' => $url, 'to' => $curlResponse->getRedirectUrl()]);
 304                         curl_close($ch);
 305                         return $this->post($curlResponse->getRedirectUrl(), $params, $headers, $redirects, $timeout);
 306                 }
 307
 308                 curl_close($ch);
 309
 310                 $this->profiler->saveTimestamp($stamp1, 'network');
 311
 312                 // Very old versions of Lighttpd don't like the "Expect" header, so we remove it when needed
 313                 if ($curlResponse->getReturnCode() == 417) {
 314                         $redirects++;
 315
 316                         if (empty($headers)) {
 317                                 $headers = ['Expect:'];
 318                         } else {
 319                                 if (!in_array('Expect:', $headers)) {
 320                                         array_push($headers, 'Expect:');
 321                                 }
 322                         }
 323                         $this->logger->info('Server responds with 417, applying workaround', ['url' => $url]);
 324                         return $this->post($url, $params, $headers, $redirects, $timeout);
 325                 }
 326
 327                 $this->logger->debug('Post_url: End.', ['url' => $url]);
 328
 329                 return $curlResponse;
 330         }
 331
 332         /**
 333          * {@inheritDoc}
 334          */
 335         public function finalUrl(string $url, int $depth = 1, bool $fetchbody = false)
 336         {
 337                 if (Network::isUrlBlocked($url)) {
 338                         $this->logger->info('Domain is blocked.', ['url' => $url]);
 339                         return $url;
 340                 }
 341
 342                 if (Network::isRedirectBlocked($url)) {
 343                         $this->logger->info('Domain should not be redirected.', ['url' => $url]);
 344                         return $url;
 345                 }
 346
 347                 $url = Network::stripTrackingQueryParams($url);
 348
 349                 if ($depth > 10) {
 350                         return $url;
 351                 }
 352
 353                 $url = trim($url, "'");
 354
 355                 $stamp1 = microtime(true);
 356
 357                 $ch = curl_init();
 358                 curl_setopt($ch, CURLOPT_URL, $url);
 359                 curl_setopt($ch, CURLOPT_HEADER, 1);
 360                 curl_setopt($ch, CURLOPT_NOBODY, 1);
 361                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 362                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 363                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 364                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 365
 366                 curl_exec($ch);
 367                 $curl_info = @curl_getinfo($ch);
 368                 $http_code = $curl_info['http_code'];
 369                 curl_close($ch);
 370
 371                 $this->profiler->saveTimestamp($stamp1, "network");
 372
 373                 if ($http_code == 0) {
 374                         return $url;
 375                 }
 376
 377                 if (in_array($http_code, ['301', '302'])) {
 378                         if (!empty($curl_info['redirect_url'])) {
 379                                 return $this->finalUrl($curl_info['redirect_url'], ++$depth, $fetchbody);
 380                         } elseif (!empty($curl_info['location'])) {
 381                                 return $this->finalUrl($curl_info['location'], ++$depth, $fetchbody);
 382                         }
 383                 }
 384
 385                 // Check for redirects in the meta elements of the body if there are no redirects in the header.
 386                 if (!$fetchbody) {
 387                         return $this->finalUrl($url, ++$depth, true);
 388                 }
 389
 390                 // if the file is too large then exit
 391                 if ($curl_info["download_content_length"] > 1000000) {
 392                         return $url;
 393                 }
 394
 395                 // if it isn't a HTML file then exit
 396                 if (!empty($curl_info["content_type"]) && !strstr(strtolower($curl_info["content_type"]), "html")) {
 397                         return $url;
 398                 }
 399
 400                 $stamp1 = microtime(true);
 401
 402                 $ch = curl_init();
 403                 curl_setopt($ch, CURLOPT_URL, $url);
 404                 curl_setopt($ch, CURLOPT_HEADER, 0);
 405                 curl_setopt($ch, CURLOPT_NOBODY, 0);
 406                 curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
 407                 curl_setopt($ch, CURLOPT_TIMEOUT, 10);
 408                 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
 409                 curl_setopt($ch, CURLOPT_USERAGENT, $this->getUserAgent());
 410
 411                 $body = curl_exec($ch);
 412                 curl_close($ch);
 413
 414                 $this->profiler->saveTimestamp($stamp1, "network");
 415
 416                 if (trim($body) == "") {
 417                         return $url;
 418                 }
 419
 420                 // Check for redirect in meta elements
 421                 $doc = new DOMDocument();
 422                 @$doc->loadHTML($body);
 423
 424                 $xpath = new DomXPath($doc);
 425
 426                 $list = $xpath->query("//meta[@content]");
 427                 foreach ($list as $node) {
 428                         $attr = [];
 429                         if ($node->attributes->length) {
 430                                 foreach ($node->attributes as $attribute) {
 431                                         $attr[$attribute->name] = $attribute->value;
 432                                 }
 433                         }
 434
 435                         if (@$attr["http-equiv"] == 'refresh') {
 436                                 $path = $attr["content"];
 437                                 $pathinfo = explode(";", $path);
 438                                 foreach ($pathinfo as $value) {
 439                                         if (substr(strtolower($value), 0, 4) == "url=") {
 440                                                 return $this->finalUrl(substr($value, 4), ++$depth);
 441                                         }
 442                                 }
 443                         }
 444                 }
 445
 446                 return $url;
 447         }
 448
 449         /**
 450          * {@inheritDoc}
 451          *
 452          * @param int $redirects The recursion counter for internal use - default 0
 453          *
 454          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 455          */
 456         public function fetch(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 457         {
 458                 $ret = $this->fetchFull($url, $binary, $timeout, $accept_content, $cookiejar, $redirects);
 459
 460                 return $ret->getBody();
 461         }
 462
 463         /**
 464          * {@inheritDoc}
 465          *
 466          * @param int $redirects The recursion counter for internal use - default 0
 467          *
 468          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 469          */
 470         public function fetchFull(string $url, bool $binary = false, int $timeout = 0, string $accept_content = '', string $cookiejar = '', int &$redirects = 0)
 471         {
 472                 return $this->get(
 473                         $url,
 474                         $binary,
 475                         [
 476                                 'timeout'        => $timeout,
 477                                 'accept_content' => $accept_content,
 478                                 'cookiejar'      => $cookiejar
 479                         ]
 480                 );
 481         }
 482
 483         /**
 484          * {@inheritDoc}
 485          */
 486         public function getUserAgent()
 487         {
 488                 return
 489                         FRIENDICA_PLATFORM . " '" .
 490                         FRIENDICA_CODENAME . "' " .
 491                         FRIENDICA_VERSION . '-' .
 492                         DB_UPDATE_VERSION . '; ' .
 493                         $this->baseUrl;
 494         }
 495 }